WO2021103341A1 - 一种像素采集电路及图像传感器 - Google Patents

一种像素采集电路及图像传感器 Download PDF

Info

Publication number
WO2021103341A1
WO2021103341A1 PCT/CN2020/078229 CN2020078229W WO2021103341A1 WO 2021103341 A1 WO2021103341 A1 WO 2021103341A1 CN 2020078229 W CN2020078229 W CN 2020078229W WO 2021103341 A1 WO2021103341 A1 WO 2021103341A1
Authority
WO
WIPO (PCT)
Prior art keywords
module
pixel
time signal
acquisition circuit
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2020/078229
Other languages
English (en)
French (fr)
Inventor
陈守顺
郭梦晗
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Omnivision Sensor Solution Shanghai Co Ltd
Original Assignee
Celepixel Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Celepixel Technology Co Ltd filed Critical Celepixel Technology Co Ltd
Priority to JP2022525427A priority Critical patent/JP7428307B2/ja
Priority to EP20892079.3A priority patent/EP4068760B1/en
Publication of WO2021103341A1 publication Critical patent/WO2021103341A1/zh
Priority to US17/731,305 priority patent/US12015861B2/en
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00—Details of television systems
    • H04N5/14—Picture signal circuitry for video frequency region
    • H04N5/144—Movement detection
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H04N23/61—Control of cameras or camera modules based on recognised objects
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
    • H04N25/47—Image sensors with pixel address output; Event-driven image sensors; Selection of pixels to be read out based on image data
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
    • H04N25/70—SSIS architectures; Circuits associated therewith
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
    • H04N25/70—SSIS architectures; Circuits associated therewith
    • H04N25/703—SSIS architectures incorporating pixels for producing signals other than image signals
    • H04N25/707—Pixels for event detection
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
    • H04N25/70—SSIS architectures; Circuits associated therewith
    • H04N25/709—Circuitry for control of the power supply
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
    • H04N25/70—SSIS architectures; Circuits associated therewith
    • H04N25/76—Addressed sensors, e.g. MOS or CMOS sensors
    • H04N25/77—Pixel circuitry, e.g. memories, A/D converters, pixel amplifiers, shared circuits or shared components
    • H04N25/772—Pixel circuitry, e.g. memories, A/D converters, pixel amplifiers, shared circuits or shared components comprising A/D, V/T, V/F, I/T or I/F converters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
    • H04N25/70—SSIS architectures; Circuits associated therewith
    • H04N25/76—Addressed sensors, e.g. MOS or CMOS sensors
    • H04N25/7795—Circuitry for generating timing or clock signals
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
    • H04N25/70—SSIS architectures; Circuits associated therewith
    • H04N25/76—Addressed sensors, e.g. MOS or CMOS sensors
    • H04N25/78—Readout circuits for addressed sensors, e.g. output amplifiers or A/D converters

Definitions

  • the present invention relates to the technical field of image acquisition, in particular to a pixel acquisition circuit and an image sensor.
  • machine vision technology As an important branch of artificial intelligence technology in the field of vision applications, machine vision technology has achieved rapid development in recent years. Relying on machine vision algorithms supported by high computing power, machine vision technology has been widely used in many fields such as pattern recognition, feature extraction, moving object detection and tracking.
  • image sensors are also an important part of the entire machine vision system. Since most image sensors output full-frame images at a fixed frame rate, mainstream machine vision algorithms also use full-frame image frame information as their input. However, these image frame information contains a lot of redundant information (mainly some useless background information), and at the same time, it also loses the dynamic information in the field of view that is concerned in the machine vision application. The above problems all increase the burden of storage and computing power on the back-end algorithms.
  • a dynamic vision image sensor hereinafter referred to as a "dynamic vision sensor”
  • a dynamic vision image sensor is more in line with the requirements of machine vision for image data sources.
  • the dynamic vision sensor only responds to and outputs dynamic information in the field of view, automatically shielding and filtering out useless background information.
  • the output of a dynamic vision sensor is an asynchronous event stream, which directly represents the dynamic information in the field of view in the temporal and spatial domain.
  • the dynamic vision sensor directly detects the dynamic information in the field of view, and reduces the redundancy of the output data.
  • the output format of the asynchronous event stream of the dynamic vision sensor is not compatible with the input data requirements of mainstream machine vision algorithms. Therefore, when the front-end image data is collected, if a dynamic vision sensor is used, the asynchronous event stream data output by it needs to be converted into synchronous frame data in some way. However, the data conversion operation still consumes a lot of external storage space and computing power, reducing the efficiency of the entire image processing system.
  • the present invention provides a pixel acquisition circuit and an image sensor to try to solve or at least alleviate at least one of the above problems.
  • a pixel acquisition circuit including: a motion detection module, adapted to detect changes in light intensity in the field of view, and when the change in light intensity exceeds a set threshold, generate a pixel trigger signal to indicate The pixel acquisition circuit enters the trigger state; the time signal generation module, coupled to the motion detection module, is adapted to generate a time signal when the pixel trigger signal is received, and the time signal is used to characterize the time information at the moment when the pixel acquisition circuit is triggered; wherein , The motion detection module and the time signal generation module are respectively coupled to the peripheral readout unit, so that the readout unit scans the pixel acquisition circuit to output the time information of the pixel acquisition circuit.
  • the time signal indicates an instantaneous voltage pulse when the pixel acquisition circuit enters the trigger state, and then decays with time.
  • the motion detection module includes: a photodetection sub-module, suitable for real-time monitoring of the light signal irradiated thereon, and output corresponding electrical signals; the trigger generation sub-module, the first An input end is coupled to the photodetection sub-module, its first output end is coupled to the readout interface sub-module, and its second output end is coupled to the time signal generation module.
  • the trigger generation sub-module is suitable for when the electrical signal meets a predetermined trigger When conditions are met, a trigger generation signal is generated to the readout interface submodule and the time signal generation module; and the readout interface submodule is coupled to the trigger generation submodule and is suitable for communicating with the readout unit.
  • the time signal generating module includes: a first switch (K1), the first terminal of which is coupled to the fixed voltage (VA), and the second terminal of which is coupled to the buffer (B1) The first end of the first capacitor (C1) and the first resistor (R1) in parallel, one end of which is grounded, and the other end is coupled to the second end of the first switch (K1) and the first buffer (B1) The first end of the first buffer (B1), the second end of which is coupled to the first end of the second switch (K2); and the second switch (K2), the second end of which is connected to the time signal column output line .
  • the time signal generation module includes: a third switch (K3), the first end of which is coupled to the fixed voltage (VA), and the second end of which is coupled to the second buffer ( The first end of B2); a second capacitor (C2) and a constant current source (I) connected in parallel, one end of which is grounded, and the other end is coupled to the second end of the third switch (K3) and the second buffer ( B2) the first end; the second buffer (B2), the second end of which is coupled to the first end of the fourth switch (K4); and the fourth switch (K4), the second end of which is connected to the time signal column Output line.
  • the first switch (K1) or the third switch (K3) is controlled by the pixel trigger signal output by the motion detection module.
  • the first switch ( K1) or the third switch (K3) is closed; and the second switch (K2) or the fourth switch (K4) is controlled by the time signal row selection signal output by the readout unit.
  • the second switch (K2) or the fourth switch (K4) is The switch (K2) or the fourth switch (K4) is closed.
  • the first resistor (R1) is an adjustable resistor; and the attenuation speed of the time signal is determined according to the time constant composed of the first capacitor (C1) and the first resistor (R1) .
  • the attenuation speed of the time signal is determined according to the second capacitor (C2) and the constant current source (I).
  • an image sensor including: a pixel acquisition circuit array, including a plurality of pixel acquisition circuits as described above; a global control unit, adapted to control the initial working state of the pixel acquisition circuit array; The output unit is adapted to scan the pixel collection circuit array row by row to output time information about the pixel collection circuit array.
  • the readout unit includes: a row selection module adapted to manage the pixel acquisition circuit array in the row direction; a column selection module adapted to manage the pixel acquisition circuit array in the column direction; reading
  • the output control module is respectively coupled with the row selection module and the column selection module, and is suitable for controlling the output of the row selection module and the column selection module.
  • the row selection module includes: a motion detection row selection sub-module, which is coupled to the motion detection module in the pixel acquisition circuit through a motion detection row request line and a motion detection row selection line, suitable To respond to the row request signal from the pixel acquisition circuit array; the time signal row selection sub-module, which is coupled to the time signal generation module in the pixel acquisition circuit through the time signal row selection line, is suitable for row-by-row pixel acquisition circuit
  • the time signal line selection line is set to be valid.
  • the row selection module includes: a motion detection row selection sub-module, which is coupled to the motion detection module in the pixel acquisition circuit through a motion detection row request line and a motion detection row selection line, suitable To respond to the row request signal from the pixel acquisition circuit array; the time signal row selection sub-module, which is coupled to the time signal generation module in the pixel acquisition circuit through the time signal row selection line, is suitable for row-by-row pixel acquisition circuit
  • the time signal line selection line is set to be valid.
  • the global control unit is further adapted to reset the pixel acquisition circuit array when the image sensor is powered on, and is also adapted to release when the pixel acquisition circuit array maintains a stable initial state Reset, so that the pixel acquisition circuit array starts to work.
  • a time signal generation module 220 is added to the pixel acquisition circuit 200.
  • the time signal generation module 220 generates a local time signal.
  • the time signal is generally a voltage signal. When an event is detected, it appears as an instantaneous voltage pulse, and then, the time signal gradually decays to characterize the passage of time.
  • the peripheral readout unit 130 adds a time signal row selection sub-module 1324 and a time signal column selection sub-module 1344, which periodically scan all the time signals output by the time signal generation module 220 in the pixel acquisition circuit 200 to form a time surface frame And output.
  • the image sensor 100 not only retains the advantages of the traditional dynamic vision sensor for asynchronous detection of dynamic information, but also can directly output the time surface frame representing the time information of the pixel acquisition circuit for the back-end machine vision algorithm.
  • the time surface frame can be used by the back-end machine vision algorithm as a training sample and to implement tasks such as object recognition and tracking. In this way, the image sensor 100 can effectively save the storage and computing overhead required for data conversion at the back end of the image processing system, thereby improving the efficiency of the image processing system.
  • FIG. 1 shows a schematic diagram of an image sensor 100 according to some embodiments of the present invention
  • FIG. 2 shows a schematic diagram of a pixel acquisition circuit 200 according to some embodiments of the present invention
  • FIG. 3 shows a schematic diagram of a time signal generating module 220 according to an embodiment of the present invention
  • FIGS. 4A-4D show waveform diagrams of time signals according to an embodiment of the present invention.
  • FIG. 5 shows a schematic diagram of a time signal generating module 220 according to another embodiment of the present invention.
  • 6A-6D show waveform diagrams of time signals according to another embodiment of the present invention.
  • FIG. 7 shows a schematic diagram of a scene of a field of view according to an embodiment of the present invention
  • FIGS. 8A to 8G respectively show schematic diagrams of output time surface frames in the scene shown in FIG. 7.
  • FIG. 1 shows a schematic diagram of an image sensor 100 according to some embodiments of the present invention.
  • the image sensor 100 can be applied to scenes that detect high-speed moving objects to output event data streams related to motion events.
  • the image sensor 100 is coupled to an external image processor (not shown), and outputs image frames representing time information to the external image processor for further calculation and application.
  • the image processor is generally used to implement certain machine vision algorithms, such as object detection and target tracking. The embodiment of the present invention does not limit this.
  • the image sensor 100 includes a pixel collection circuit array 110, a global control unit 120 and a readout unit 130.
  • the pixel acquisition circuit array 110 as the core part of the image sensor 100, is composed of a plurality of pixel acquisition circuits 200 evenly distributed in the row and column direction ( Figure 1 shows a 3 ⁇ 3 pixel acquisition circuit array, which is not limited to this).
  • the global control unit 120 is coupled to the pixel collection circuit array 110, and the global control unit 120 can control the initial working state of the pixel collection circuit array 110.
  • the global control unit 120 is responsible for resetting the entire pixel collection circuit array 110 when the image sensor 100 is powered on to ensure that each pixel collection circuit 112 maintains a stable initial state. Subsequently, the global control unit 120 releases the reset, and the pixel collection circuit array 110 starts to work normally.
  • the readout unit 130 is coupled to the pixel collection circuit array 110. According to the embodiment of the present invention, the readout unit 130 performs two functions: one is to manage events generated by the pixel collection circuit array 200, and the other is to scan the time signal generated by the pixel collection circuit array 110 to generate information about the pixel collection circuit array. 110 image frames of time information. The position of the pixel in the image frame is the row and column address of the pixel acquisition circuit, and the size of the pixel is the size of the time signal of the pixel acquisition circuit. In an embodiment, the image frame is called a time surface frame (TimeSurface), which reflects the temporal and spatial distribution of dynamic information in the field of view in the past short period of time. At the same time, these two functions are independent of each other.
  • TimeSurface time surface frame
  • the readout unit 130 includes three sub-units in total, namely a row selection module 132, a column selection module 134, and a readout control module 136.
  • the row selection module 132 manages the pixel collection circuit array 110 in the row direction
  • the column selection module 134 manages the pixel collection circuit array 110 in the column direction
  • the readout control module 136 coordinates the row selection module 132 and the column selection module 134.
  • the pixel acquisition circuit 200 not only includes a motion detection module 210, but also includes a time signal generation module 220.
  • FIG. 2 shows a schematic diagram of a pixel acquisition circuit 200 according to some embodiments of the present invention.
  • the motion detection module 210 is used to detect dynamic information in the field of view
  • the time signal generation module 220 is used to generate a local time signal to represent the time information of the pixel acquisition circuit 200 at the moment when it is triggered.
  • the motion detection module 210 responds to changes in the light intensity in the field of view in real time, and causes the pixel collection circuit 200 to enter a trigger state after the external light intensity changes exceed a set threshold to indicate that the pixel collection circuit 200 detects an event.
  • the time signal generation module 220 in the pixel acquisition circuit 200 is responsible for generating a local time signal.
  • the time signal is generally a voltage signal, which appears as an instantaneous voltage pulse signal when the pixel acquisition circuit 200 enters the trigger state.
  • the amplitude of the time signal gradually decays over time in some way. Therefore, by acquiring the amplitude of the time signal at a certain moment, it is possible to know whether the corresponding pixel acquisition circuit 200 detects an event and when the event is detected.
  • the row selection module 132 is further divided into a motion detection row selection sub-module 1322 and a time signal row selection sub-module 1324; column selection The module 134 is further divided into a motion detection column selection sub-module 1342 and a time signal column selection sub-module 1344. As shown in Figure 1.
  • the readout unit 130 For each pixel acquisition circuit 200 in the pixel acquisition circuit array 110, the readout unit 130 is coupled to it through a corresponding row and column bus to realize event management and time signal scanning. At the same time, the management of the event by the reading unit 130 and the scanning of the time signal are independent of each other. The two processing procedures will be described in detail below in conjunction with FIG. 2.
  • the motion detection row selection sub-module 1322 is coupled to the motion detection module 210 in the pixel acquisition circuit 200 through a motion detection row request line and a motion detection row selection line; the motion detection column selection sub-module 1342 uses a motion detection column request line and a motion detection column The selection line is coupled to the motion detection unit 210 in the pixel acquisition circuit 200.
  • the motion detection row selection sub-module 1322 manages all the motion detection row request lines of the pixel acquisition circuit array 110, and responds to the received row request signal.
  • the motion detection row selection sub-module 1322 responds to a valid row request signal by setting its corresponding motion detection row selection line to be valid.
  • the motion detection column selection sub-module 1342 responds to these column request signals.
  • the motion detection column selection sub-module 1342 manages all column request signals in the column direction, and releases the trigger state of the corresponding pixel acquisition circuit by setting the corresponding motion detection column selection line as valid, so that it can respond to external light intensity again Variety.
  • the readout control module 136 When all the pixel acquisition circuits in the trigger state of the row are released from the trigger state, the readout control module 136 notifies the motion detection row selection sub-module 1322 to perform a line change operation, and the motion detection row selection sub-module 1322 cancels the motion detection row selection signal of the current row , And select the next row, and so on, until all the pixel collection circuits in the pixel collection circuit array 110 have performed the above scanning process.
  • the time signal row selection submodule 1324 is coupled to the time signal generation module 220 in the pixel acquisition circuit 200 through a time signal row selection line; the time signal column selection submodule 1344 is coupled to the pixel acquisition circuit 200 through a time signal column output line The time signal generating module 220.
  • the time signal row selection sub-module 1324 activates the time signal row selection line generated by the pixel acquisition circuit 200 row by row (for example, first sets the time signal row selection line of the first row to be valid) ), the pixel acquisition circuit 200 outputs the time signal to the time signal column output line when the time signal row selection line coupled to it is valid.
  • the time signal column selection sub-module 1344 scans and reads out one by one through the time signal column output line The time signal output by the pixel acquisition circuit 200.
  • the readout control module 136 notifies the time signal row selection sub-module 1324 to perform a line change operation, and the time signal row selection sub-module 1324 cancels the time signal row selection of the current row Signal and select the next line. Proceed in this way, until the time signal of the pixel acquisition circuit 200 of the last row is read, a time surface frame is formed. Then, the readout control module 136 cancels the time signal line selection signal of the last line, re-enables the time signal line selection signal of the first line, and starts the read operation of the next time surface frame.
  • the output of the image sensor 100 is a synchronized time surface frame, which contains the spatio-temporal information of the motion in the field of view of the image sensor 100, which can be directly used by the back-end machine vision algorithm as
  • the training samples are used to achieve tasks such as object recognition and target tracking.
  • the motion detection row selection sub-module 1322 and the motion detection column selection sub-module 1342 can be implemented by selecting a scanning circuit or a decider.
  • the time signal row selection sub-module 1324 and the time signal column selection sub-module 1344 can pass the full scan Circuit implementation. Since the above circuits belong to technologies well known in the art, they will not be repeated here.
  • the structures of the motion detection module 210 and the time signal generation module 220 in the pixel acquisition circuit 200 are respectively described below.
  • the motion detection module 210 includes: a photodetection sub-module 212, a trigger generation sub-module 214, and a readout interface sub-module 216 coupled to each other.
  • the photodetection sub-module 212 monitors the light signal irradiated on it in real time, and outputs corresponding electrical signals.
  • the first input end of the trigger generation sub-module 214 is coupled to the photodetection sub-module 212, the first output end is coupled to the readout interface sub-module 216, and the second output end is coupled to the time signal generation module 220.
  • the trigger generation sub-module 214 generates a trigger generation signal to the readout interface sub-module 216 and the time signal generation module 220 when the electrical signal meets the predetermined trigger condition.
  • the reading interface sub-module 216 is coupled to the trigger generation sub-module 214 to communicate with the reading unit 130. Since the motion detection module 210 already exists in the existing dynamic vision sensor, it will not be repeated here.
  • the time signal represents an instantaneous voltage pulse when the pixel acquisition circuit 200 enters the trigger state, and then decays with time.
  • the time signal may decay exponentially or linearly, but is not limited thereto.
  • FIG. 3 and FIG. 5 respectively show schematic diagrams of two time signal generation modules 220 according to an embodiment of the present invention.
  • Figure 3 shows the attenuation in exponential form
  • Figure 5 shows the attenuation in linear form.
  • the exponential decay type time signal generating module 220 includes: a first switch (K1), a first capacitor (C1), a first resistor (R1), a first buffer (B1), and a second switch ( K2).
  • the first terminal of the first switch (K1) is coupled to the fixed voltage (VA), and the second terminal thereof is coupled to the first terminal of the first buffer (B1).
  • the first capacitor (C1) and the first resistor (R1) are connected in parallel, one end of which is grounded, and the other end is coupled to the second end of the first switch (K1) and the first end of the first buffer (B1).
  • the second end of the first buffer (B1) is coupled to the first end of the second switch (K2).
  • the second terminal of the second switch (K2) is connected to the time signal column output line.
  • the first switch (K1) is controlled by the pixel trigger signal output by the motion detection module 210. When the pixel trigger signal is valid, the first switch (K1) is closed.
  • the second switch (K2) is controlled by the time signal row selection signal output by the readout unit 130. When the time signal row selection signal is valid, the second switch (K2) is closed.
  • the first resistor (R1) is an adjustable resistor. The decay speed of the time signal is determined according to the time constant composed of the first capacitor (C1) and the first resistor (R1).
  • 4A-4D show waveform diagrams of the time signal generated by the time signal generating module 220 shown in FIG. 3.
  • FIG. 4A shows the time signal when the pixel acquisition circuit 200 is triggered for a single time
  • FIG. 4B shows the time signal when the pixel acquisition circuit 200 is triggered for multiple times (here shown as two consecutive triggers).
  • the horizontal axis represents time t
  • the vertical axis represents pixel trigger signal VT.
  • 4C and 4D show waveform diagrams of the time signals corresponding to FIGS. 4A and 4B, respectively.
  • the horizontal axis represents time t
  • the vertical axis represents the voltage value V1 of the output time signal.
  • the pixel trigger signal VT is low, the first switch K1 is turned off, and V1 is discharged to the zero position through the first resistor R1; At t1, the pixel acquisition circuit 200 is triggered, and the pixel trigger signal VT is a short pulse signal.
  • the first switch K1 is closed, and the first capacitor C1 is forcibly charged to a fixed voltage VA, that is, V1 is pulled instantaneously at t1 Increase to VA; then the pixel trigger signal is invalid, the first switch K1 is turned off again, the charge on the first capacitor C1 is slowly discharged through the first resistor R1, so V1 gradually decays in an exponential form; when t is greater than t3, V1 decays It is zero potential.
  • the pixel acquisition circuit 200 is triggered at time t1 and time t2, respectively, and the time between the two triggers is relatively short.
  • the pixel trigger signal VT gives a narrow pulse, which makes the first switch K1 close and forcibly pulls the time signal V1 to a fixed voltage VA, then the first switch K1 opens, and V1 begins to decay exponentially.
  • the pixel acquisition circuit 200 is triggered again at time t2, the pixel trigger signal VT again gives a narrow pulse and pulls the time signal V1 to VA again, then the first switch K1 is turned off, and V1 decays exponentially.
  • the time signal row selection signal is valid
  • the second switch K2 is closed, and V1 is sent to the time signal column output line through the first buffer B1, and then is acquired and output by the column selection module 134.
  • the instantaneous amplitude of the time signal at a certain moment can represent the time information between the moment when the pixel acquisition circuit is triggered and the current moment, that is, the higher the instantaneous amplitude, the distance from the moment when the pixel acquisition circuit is triggered. The closer, the lower the instantaneous amplitude, and the farther from the moment when the pixel acquisition circuit is triggered.
  • the attenuation speed of the time signal is determined by the time constant composed of the first capacitor C1 and the first resistor R1.
  • the first resistor R1 can be an adjustable resistor, according to different back-end applications Different resistance values need to be changed to adjust the speed at which the time signal decays.
  • Fig. 5 shows a linear attenuation type time signal generating module 220.
  • the difference from Fig. 3 is that the first resistor R1 is replaced by a constant current source I.
  • the third switch K3 is turned off, the second capacitor C2 is stored The charge of is discharged through the constant current source I, thereby forming the waveform of the time signal (V2) that decays linearly with time.
  • the time signal generating module 220 includes: a third switch (K3), a second capacitor (C2), a constant current source (I), a second buffer (B2), and a fourth switch (K4).
  • the first terminal of the third switch (K3) is coupled to the fixed voltage (VA), and the second terminal is coupled to the first terminal of the second buffer (B2).
  • the second capacitor (C2) is connected in parallel with the constant current source (I), one end of which is grounded, and the other end is coupled to the second end of the third switch (K3) and the first end of the second buffer (B2).
  • the second end of the second buffer (B2) is coupled to the first end of the fourth switch (K4), and the second end of the fourth switch (K4) is connected to the time signal column output line.
  • the third switch (K3) is controlled by the pixel trigger signal output by the motion detection module 210. When the pixel trigger signal is valid, the third switch (K3) is closed.
  • the fourth switch (K4) is controlled by the time signal row selection signal output by the readout unit 130. When the time signal row selection signal is valid, the fourth switch (K4) is closed. In an embodiment, the decay speed of the time signal is determined according to the second capacitor (C2) and the constant current source (I).
  • FIG. 6A-6D show waveform diagrams of the time signal generated by the time signal generating module 220 shown in FIG. 5.
  • FIG. 6A shows the time signal when the pixel acquisition circuit 200 is triggered once
  • FIG. 6B shows the time signal when the pixel acquisition circuit 200 is triggered multiple times (here, two consecutive triggers).
  • the horizontal axis represents time t
  • the vertical axis represents pixel trigger signal VT.
  • 6C and 6D show waveform diagrams of the time signals corresponding to FIGS. 6A and 6B, respectively.
  • the horizontal axis represents time t
  • the vertical axis represents the voltage value V2 of the output time signal.
  • the pixel trigger signal VT is low, the third switch K3 is turned off, and V2 is discharged to the zero position through the constant current source I; At t1, the pixel acquisition circuit 200 is triggered, and the pixel trigger signal VT is a short pulse signal.
  • the third switch K3 is closed, and the second capacitor C2 is forcibly charged to a fixed voltage VA, that is, V2 is pulled instantaneously at t1 Rise to VA; then the pixel trigger signal is invalid, the third switch K3 is turned off again, the charge on the second capacitor C2 is discharged through the constant current source I, so V2 decays in a linear form; when t is greater than t3, V2 decays to zero Potential.
  • the pixel acquisition circuit 200 is triggered at time t1 and time t2, respectively, and the time between the two triggers is relatively short.
  • the pixel trigger signal VT gives a narrow pulse, which makes the third switch K3 close and forcibly pulls the time signal V2 to a fixed voltage VA, then the third switch K3 opens, and V2 starts to decay in a linear fashion.
  • the pixel acquisition circuit 200 is triggered again at time t2, the pixel trigger signal VT again gives a narrow pulse and pulls the time signal V2 to VA again, then the third switch K1 is turned off, and V2 is attenuated in a linear fashion.
  • the time signal row selection signal is valid
  • the fourth switch K2 is closed
  • V2 is sent to the time signal column output line through the second buffer B2, and then is acquired and output by the column selection module 134.
  • the waveform diagram of the single-triggered time signal shown in FIG. 6C can better illustrate the time information represented by the instantaneous amplitude of the time signal.
  • the linear discharge time (t1 to t3) is 20ms, that is, the time signal will be completely discharged from VA to zero potential after 20ms.
  • the instantaneous voltage amplitude is normalized according to the amplitude of VA and referred to as the normalized amplitude, which ranges from 0 to 1. As shown in FIG.
  • the pixel acquisition circuit 200 when the pixel acquisition circuit 200 is read out, if the normalized amplitude acquired on the time signal output line is 1, it means that the pixel acquisition circuit 200 is triggered at the readout time, or By the time of readout, the pixel acquisition circuit 200 has been triggered for 0ms; if the normalized amplitude obtained is 0.75 (that is, VB shown in FIG. 6C), it means that at the time of readout, the pixel acquisition circuit 200 It has been triggered for 5ms; if the normalized amplitude obtained is 0.5 (that is, the VC shown in FIG.
  • the pixel acquisition circuit 200 has been triggered for 10ms at the time of reading; if the normalized amplitude obtained is If the amplitude is 0, it means that the pixel acquisition circuit 200 has been triggered for more than 20 ms, or the pixel acquisition circuit 200 has not been triggered at all.
  • the pixel acquisition circuit 200 that has been triggered for a long time can be regarded as a pixel acquisition circuit that has not been triggered. This is because the output time surface frame only pays attention to the movement that occurred at the current moment and the previous nearby time, and does not pay attention to the movement that occurred long ago. It should be understood that the assumption and normalization of the linear discharge time here are only for ease of understanding, and the embodiments of the present invention are not limited thereto.
  • the speed at which the time signal decays can be determined by the size of the second capacitor C2 and the constant current source I.
  • time signal generating modules 220 of exponential decay type and linear decay type shown in FIG. 3 and FIG. 5 are only examples.
  • the time signal generating modules 220 of other types of attenuation can also be used in different ways. To achieve, I won’t repeat them here.
  • Fig. 7 shows a schematic diagram of a scene of a field of view according to an embodiment of the present invention.
  • 8A to 8G respectively show schematic diagrams of output time surface frames in the scene shown in FIG. 7.
  • the pixel collection circuit array 110 of the image sensor 100 includes 6 rows and 6 columns of pixel collection circuits 200.
  • the field of view area corresponding to the pixel acquisition circuit array 110 is a white background, and a black moving object A (to distinguish the moving object A, grid lines are used to fill the black moving object A in FIG. 7) from Enter the lower left of the field of view area, and move along the horizontal direction to the far right side of the field of view area, and its motion track covers all the pixel acquisition circuits in the 4th to 6th rows.
  • Figure 8A uses a diagonal line "/" to mark the pixel acquisition circuit triggered by the object A during its movement. And correspondingly mark its corresponding trigger time. For example, the pixel acquisition circuit in row 5 and column 4 is triggered at 20 ms. The unfilled pixel acquisition circuit 200 indicates that it has not been triggered during this process.
  • the output of a complete time surface frame requires the time signal row selection sub-module 1324 and the time signal column selection sub-module 1344 to scan the entire pixel acquisition circuit array 110 row by row under the control of the readout unit 130 to obtain The time signal of each pixel acquisition circuit 200 generates the time signal output by the module 220.
  • the pixel acquisition circuit 200 here uses the linear attenuation time signal generation module 220 shown in FIG. 5, and at the same time, the discharge time is set to 20ms and the output amplitude is normalized to illustrate the output of the pixel acquisition circuit.
  • the process of temporal information ie, temporal surface frames).
  • the image sensor 100 When the trigger time of a pixel acquisition circuit 200 is closer to the readout time of the surface frame at the current time, the corresponding normalized amplitude is closer to 1; otherwise, the corresponding normalized amplitude is closer to 0.
  • the image sensor 100 outputs a total of 6 time surface frames from the 5th ms to the 30th ms, as shown in FIGS. 8B-8G, respectively.
  • the time surface frames are respectively denoted as time surface frame 1, time surface frame 2, time surface frame 3,..., time surface frame 6, and their corresponding output times are: 5ms, 10ms, 15ms, 20ms, 25ms, 30ms.
  • the readout time is the 5th ms.
  • the normalized amplitudes of these pixel acquisition circuits are 1, and the normalized amplitude values of the remaining pixel acquisition circuits are all 0.
  • the readout time is the 10th ms.
  • the normalized amplitude is 1; and the normalized amplitudes of the remaining pixel acquisition circuits are all 0.
  • the readout time is 15 ms.
  • the readout time is 20 ms.
  • the readout time is 25 ms.
  • the pixel acquisition circuits in the 4th to 6th rows of the first column are triggered at 5ms, so in the time surface frame 5, the normalized amplitudes of these pixel acquisition circuits are 0.
  • the pixel acquisition circuits of rows 4 to 6 in the second column are triggered at the 10th ms, so in the time surface frame 5, these pixel acquisition circuits are normalized
  • the readout time is 30 ms.
  • the pixel acquisition circuits in the 4th to 6th rows of the first column are triggered at 5 ms, so in the time surface frame 6, the normalized amplitudes of these pixel acquisition circuits are 0.
  • a time signal generation module 220 is added to the pixel acquisition circuit 200.
  • the time signal generation module 220 generates a local time signal.
  • the time signal is generally a voltage signal. When an event is detected, it appears as an instantaneous voltage pulse, and then, the time signal gradually decays to characterize the passage of time.
  • the peripheral readout unit 130 adds a time signal row selection sub-module 1324 and a time signal column selection sub-module 1344, which periodically scan all the time signals output by the time signal generation module 220 in the pixel acquisition circuit 200 to form a time surface frame And output.
  • the image sensor 100 not only retains the advantages of the traditional dynamic vision sensor for asynchronous detection of dynamic information, but also can directly output the time surface frame representing the time information of the pixel acquisition circuit, and send it to the back-end image processor (for example, execute machine vision). Algorithmic image processor).
  • the time surface frame can be used by the back-end machine vision algorithm as a training sample and to implement tasks such as object recognition and tracking. In this way, the image sensor 100 can effectively save the storage and computing overhead required for data conversion at the back end of the image processing system, thereby improving the efficiency of the image processing system.
  • modules or units or components of the device in the example disclosed herein can be arranged in the device as described in this embodiment, or alternatively can be positioned differently from the device in this example Of one or more devices.
  • the modules in the foregoing examples can be combined into one module or, in addition, can be divided into multiple sub-modules.
  • modules or units or components in the embodiments can be combined into one module or unit or component, and in addition, they can be divided into multiple sub-modules or sub-units or sub-components. Except that at least some of such features and/or processes or units are mutually exclusive, any combination can be used to compare all the features disclosed in this specification (including the accompanying claims, abstract and drawings) and any method or methods disclosed in this manner or All the processes or units of the equipment are combined. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Transforming Light Signals Into Electric Signals (AREA)

Abstract

本发明公开了一种像素采集电路及图像传感器。其中，像素采集电路包括：运动检测模块，适于检测视场中的光强变化，并在光强变化超出设定阈值时，生成像素触发信号，来指示像素采集电路进入触发状态；时间信号生成模块，与运动检测模块耦接，适于在接收到像素触发信号时生成时间信号，时间信号用于表征像素采集电路被触发时刻的时间信息；其中运动检测模块和时间信号生成模块分别耦接到外围的读出单元，以便读出单元对像素采集电路进行扫描来输出像素采集电路的时间信息。

Description

一种像素采集电路及图像传感器 技术领域
本发明涉及图像采集技术领域，尤其涉及一种像素采集电路及图像传感器。
背景技术
作为人工智能技术在视觉应用领域的一个重要分支，机器视觉技术近年来获得了飞速的发展。依托于高算力支撑下的机器视觉算法，机器视觉技术在模式识别、特征提取、运动物体检测和跟踪等诸多领域均得到了广泛的应用。
作为前端图像数据获取的源头，图像传感器也是整个机器视觉系统的重要组成部分。由于绝大部分图像传感器按照固定帧率输出全幅图像，因而主流的机器视觉算法也都采用全幅的图像帧信息作为其输入。然而，这些图像帧信息中包含了大量的冗余信息(其中主要是一些无用的背景信息)，同时，还丢失了机器视觉应用中所关心的视场中的动态信息。上述问题均加大了后端算法在存储和算力方面的负担。相比之下，另外一种图像传感器，即动态视觉图像传感器(以下简称“动态视觉传感器”)，则从原理上更加符合机器视觉对图像数据源的要求。动态视觉传感器仅响应并输出视场中的动态信息，自动屏蔽并过滤掉无用的背景信息。与固定帧率输出的图像传感器不同，动态视觉传感器的输出是异步的事件流，该事件流在时空域上直接表征视场中动态信息。
如上所述，动态视觉传感器直接检测视场中的动态信息、并降低了输出数据的冗余程度。但是，动态视觉传感器的异步事件流的输出格式却与主流的机器视觉算法对输入数据的要求不兼容。因此，在前端采集图像数据时，若使用动态视觉传感器，则需要将其输出的异步事件流数据按照某种方式转化为同步的帧数据。然而，数据转换操作依然会耗费大量的外部存储空间以 及算力，降低整个图像处理系统的效率。
鉴于此，需要一种新的图像传感器，来解决上述问题。
发明内容
本发明提供了一种像素采集电路及图像传感器，以力图解决或至少缓解上面存在的至少一个问题。
根据本发明的一个方面，提供了一种像素采集电路，包括：运动检测模块，适于检测视场中的光强变化，并在光强变化超出设定阈值时，生成像素触发信号，来指示像素采集电路进入触发状态；时间信号生成模块，与运动检测模块耦接，适于在接收到像素触发信号时生成时间信号，时间信号用于表征所述像素采集电路被触发时刻的时间信息；其中，运动检测模块和时间信号生成模块分别耦接到外围的读出单元，以便读出单元对像素采集电路进行扫描来输出像素采集电路的时间信息。
可选地，在根据本发明的像素采集电路中，时间信号在像素采集电路进入触发状态时指示一个瞬时电压脉冲、且之后随时间而衰减。
可选地，在根据本发明的像素采集电路中，运动检测模块包括：光电探测子模块，适于实时监测照射在其上的光信号，并输出相应的电信号；触发生成子模块，其第一输入端耦接到光电探测子模块，其第一输出端耦接到读出接口子模块，其第二输出端耦接到时间信号生成模块，触发生成子模块适于在电信号满足预定触发条件时，生成触发生成信号给读出接口子模块和时间信号生成模块；以及读出接口子模块，耦接到触发生成子模块，适于与读出单元进行通信。
可选地，在根据本发明的像素采集电路中，时间信号生成模块包括：第一开关(K1)，其第一端耦接固定电压(VA)，其第二端耦接缓冲器(B1)的第一端；并联的第一电容(C1)和第一电阻(R1)，其一端接地，另一端耦接于所述第一开关(K1)的第二端以及第一缓冲器(B1)的第一端；第一缓冲器(B1)，其第二端耦接至第二开关(K2)的第一端；以及第二开关(K2)，其第二端接至时间信号列输出线。
可选地，在根据本发明的像素采集电路中，时间信号生成模块包括：第三开关(K3)，其第一端耦接固定电压(VA)，其第二端耦接第二缓冲器(B2)的第一端；并联的第二电容(C2)和恒流源(I)，其一端接地，另一端耦接于所述第三开关(K3)的第二端以及第二缓冲器(B2)的第一端；第二缓冲器(B2)，其第二端耦接至第四开关(K4)的第一端；以及第四开关(K4)，其第二端接至时间信号列输出线。
可选地，在根据本发明的像素采集电路中，第一开关(K1)或第三开关(K3)由运动检测模块输出的像素触发信号来控制，当像素触发信号有效时，第一开关(K1)或第三开关(K3)闭合；以及第二开关(K2)或第四开关(K4)由读出单元输出的时间信号行选择信号来控制，当时间信号行选择信号有效时，第二开关(K2)或第四开关(K4)闭合。
可选地，在根据本发明的像素采集电路中，第一电阻(R1)为可调电阻；以及根据第一电容(C1)和第一电阻(R1)组成的时间常数确定时间信号的衰减速度。
可选地，在根据本发明的像素采集电路中，根据第二电容(C2)和恒流源(I)确定时间信号的衰减速度。
根据本发明的又一个方面，提供了一种图像传感器，包括：像素采集电路阵列，包括多个如上所述的像素采集电路；全局控制单元，适于控制像素采集电路阵列的初始工作状态；读出单元，适于逐行地对像素采集电路阵列进行扫描，来输出关于像素采集电路阵列的时间信息。
可选地，在根据本发明的图像传感器中，读出单元包括：行选择模块，适于在行方向上管理像素采集电路阵列；列选择模块，适于在列方向上管理像素采集电路阵列；读出控制模块，分别与行选择模块和列选择模块相耦接，适于控制行选择模块与列选择模块的输出。
可选地，在根据本发明的图像传感器中，行选择模块包括：运动检测行选择子模块，通过运动检测行请求线和运动检测行选择线耦接到像素采集电路中的运动检测模块，适于对来自像素采集电路阵列的行请求信号进行响应；时间信号行选择子模块，通过时间信号行选择线耦接到像素采集电路中的时间信号生成模块，适于逐行地将像素采集电路的时间信号行选择线置为有效。
可选地，在根据本发明的图像传感器中，行选择模块包括：运动检测行选择子模块，通过运动检测行请求线和运动检测行选择线耦接到像素采集电路中的运动检测模块，适于对来自像素采集电路阵列的行请求信号进行响应；时间信号行选择子模块，通过时间信号行选择线耦接到像素采集电路中的时间信号生成模块，适于逐行地将像素采集电路的时间信号行选择线置为有效。
可选地，在根据本发明的图像传感器中，全局控制单元还适于在图像传感器上电时，复位像素采集电路阵列，以及，还适于在像素采集电路阵列保持稳定的初始状态时，解除复位，使得像素采集电路阵列开始工作。
根据本公开的图像传感器100，在像素采集电路200中添加了时间信号生成模块220，该时间信号生成模块220生成一个本地的时间信号，该时间信号一般是一个电压信号，其在像素采集电路200检测到事件时表现为一个瞬时的电压脉冲，而后，该时间信号逐渐衰减以表征时间的流逝。外围的读出单元130添加了时间信号行选择子模块1324和时间信号列选择子模块1344，其定时地扫描所有像素采集电路200中时间信号生成模块220所输出的时间信号，从而形成时间表面帧并输出。
综上，该图像传感器100既保留了传统动态视觉传感器的异步检测动态信息的优点，又可以直接输出表征像素采集电路时间信息的时间表面帧，给到后端的机器视觉算法。该时间表面帧可以被后端的机器视觉算法用来作为训练的样本以及实现物体识别、追踪等任务。这样，该图像传感器100能够有效节省图像处理系统后端进行数据转换所需的存储和算力的开销，从而提高图像处理系统的效率。
附图说明
为了实现上述以及相关目的，本文结合下面的描述和附图来描述某些说明性方面，这些方面指示了可以实践本文所公开的原理的各种方式，并且所有方面及其等效方面旨在落入所要求保护的主题的范围内。通过结合附图阅读下面的详细描述，本公开的上述以及其它目的、特征和优势将变得更加明显。遍及本公开，相同的附图标记通常指代相同的部件或元素。
图1示出了根据本发明一些实施例的图像传感器100的示意图；
图2示出了根据本发明一些实施例的像素采集电路200的示意图；
图3示出了根据本发明一个实施例的时间信号生成模块220的示意图；
图4A-图4D示出了根据本发明一个实施例的时间信号的波形图；
图5示出了根据本发明另一个实施例的时间信号生成模块220的示意图；
图6A-图6D示出了根据本发明另一个实施例的时间信号的波形图；以及
图7示出了根据本发明一个实施例的视场的场景示意图，图8A-图8G分别示出了在图7所示的场景中，输出的时间表面帧的示意图。
具体实施方式
下面将参照附图更详细地描述本公开的示例性实施例。虽然附图中显示了本公开的示例性实施例，然而应当理解，可以以各种形式实现本公开而不应被这里阐述的实施例所限制。相反，提供这些实施例是为了能够更透彻地理解本公开，并且能够将本公开的范围完整的传达给本领域的技术人员。
图1示出了根据本发明一些实施例的图像传感器100的示意图。该图像传感器100能够应用于检测高速运动物体的场景，以输出与运动事件相关的事件数据流。根据一种实现方式，该图像传感器100与外部的图像处理器(未示出)相耦接，输出表征时间信息的图像帧给外部的图像处理器，以进行下一步的计算和应用。该图像处理器一般用于实现某些机器视觉算法，例如物体检测、目标跟踪等。本发明的实施例对此不作限制。
如图1所示，图像传感器100包括像素采集电路阵列110、全局控制单元120和读出单元130。其中，像素采集电路阵列110作为图像传感器100的核心部分，由在行列方向上均匀分布的多个像素采集电路200组成(如图1示出了一个3×3大小的像素采集电路阵列，不限于此)。
全局控制单元120与像素采集电路阵列110相耦接，该全局控制单元120能够控制像素采集电路阵列110的初始工作状态。一般来说，全局控制单元120负责在图像传感器100上电时，复位整个像素采集电路阵列110，以确保每个像素采集电路112均保持稳定的初始状态。随后，全局控制单元120解除复位，像素采集电路阵列110开始正常工作。
读出单元130与像素采集电路阵列110相耦接。根据本发明的实施方式，读出单元130完成两个功能：一是管理像素采集电路阵列200所产生的事件，二是扫描像素采集电路阵列110所生成的时间信号，以生成关于像素采集电路阵列110的时间信息的图像帧。该图像帧中像素的位置即像素采集电路的行列地址，像素的大小即像素采集电路的时间信号的大小。在一种实施例中，该图像帧被称作为时间表面帧(TimeSurface)，其反映了过去一小段时间内视场中的动态信息在时间和空间上的分布。同时，这两个功能是相互独立的。
为实现上述功能，读出单元130一共包括三个子单元，分别是行选择模块132、列选择模块134以及读出控制模块136。行选择模块132在行方向上管理像素采集电路阵列110，列选择模块134在列方向上管理像素采集电路阵列110，读出控制模块136协调行选择模块132与列选择模块134。
与一般的动态视觉传感器不同的是，在根据本发明的图像传感器100中，像素采集电路200不仅包含运动检测模块210，还包含时间信号生成模块220。如图2中示出了根据本发明一些实施例的像素采集电路200的示意图。其中，运动检测模块210用于实现对视场中动态信息的检测，时间信号生成模块220用于生成一个本地的时间信号，来表征像素采集电路200在被触发时刻的时间信息。具体地，运动检测模块210实时响应视场中光强变化，并在外界光强变化超出设定的阈值后使像素采集电路200进入到触发状态，以表征该像素采集电路200检测到一个事件。像素采集电路200中的时间信号生成模块220负责生成一个本地的时间信号，该时间信号一般是一个电压信号，其在像素采集电路200进入触发状态时表现为一个瞬时的电压脉冲信号，而后，该时间信号的幅值以某种方式随着时间逐渐衰减。由此，通过获取某一瞬间该时间信号的幅值即可得知对应像素采集电路200是否检测到事件以及何时检测到了事件。在一些实施例中，对应于像素采集电路200中的运动检测模块210和时间信号生成模块220，行选择模块132又分为运动检测行选择子模块1322和时间信号行选择子模块1324；列选择模块134又分为运动检测列选择子模块1342和时间信号列选择子模块1344。如图1所示。
对于像素采集电路阵列110中的每个像素采集电路200，读出单元130都通过相应的行列总线与其耦接，来实现对事件的管理和对时间信号的扫描。 同时，读出单元130对事件的管理以及时间信号的扫描是相互独立的。以下结合图2，对这两个处理过程进行详细的说明。
运动检测行选择子模块1322通过运动检测行请求线和运动检测行选择线耦接到像素采集电路200中的运动检测模块210；运动检测列选择子模块1342通过运动检测列请求线和运动检测列选择线耦接到像素采集电路200中的运动检测单元210。在一种实施例中，当某个像素采集电路200进入触发状态后，其内部的运动检测模块210会将相应的运动检测行请求线置为有效。运动检测行选择子模块1322管理像素采集电路阵列110的所有运动检测行请求线，对所接收到的行请求信号进行响应。可选地，运动检测行选择子模块1322通过将其对应的运动检测行选择线置为有效，来响应一个有效的行请求信号。当某行像素采集电路的行请求信号得到响应时，该行像素采集电路的列请求信号有效，此时，运动检测列选择子模块1342对这些列请求信号进行响应。运动检测列选择子模块1342在列方向上管理所有的列请求信号，并通过将相应的运动检测列选择线置为有效来解除对应像素采集电路的触发状态，使其可以重新响应外界的光强变化。当该行所有处于触发状态的像素采集电路均解除触发状态后，读出控制模块136通知运动检测行选择子模块1322进行换行操作，运动检测行选择子模块1322撤销当前行的运动检测行选择信号，并选中下一行，依次类推，直到对像素采集电路阵列110中所有像素采集电路均执行上述扫描过程。
同样，时间信号行选择子模块1324通过时间信号行选择线耦接到像素采集电路200中的时间信号生成模块220；时间信号列选择子模块1344通过时间信号列输出线耦接到像素采集电路200的时间信号生成模块220。在一种实施例中，时间信号行选择子模块1324逐行地将像素采集电路200所产生的时间信号行选择线置为有效(例如，先将第一行的时间信号行选择线置为有效)，像素采集电路200在与之耦接的时间信号行选择线有效时，输出时间信号至时间信号列输出线，此时，时间信号列选择子模块1344通过时间信号列输出线逐一扫描读出该像素采集电路200所输出的时间信号。当该行所有像素采集电路200的时间信号均被读取完毕后，读出控制模块136通知时间信号行选择子模块1324进行换行操作，时间信号行选择子模块1324撤销当前行的 时间信号行选择信号，并选中下一行。照此进行下去，直到最后一行像素采集电路200的时间信号被读取完毕时，便形成了一个时间表面帧。然后，读出控制模块136撤销最后一行的时间信号行选择信号，重新将第一行的时间信号行选择信号置为有效，开始下一时间表面帧的读出操作。
按照如上所述的读出机制，图像传感器100的输出是同步的时间表面帧，该时间表面帧包含了图像传感器100视场中运动的时空信息，它可以直接被后端的机器视觉算法用来作为训练的样本以实现物体识别、目标追踪等任务。在具体实现上，运动检测行选择子模块1322和运动检测列选择子模块1342可以通过选择扫描电路或者判决器来实现，时间信号行选择子模块1324和时间信号列选择子模块1344可以通过全扫描电路实现，由于以上电路均属于本领域公知的技术，所以在此不再赘述。
以下对像素采集电路200中的运动检测模块210和时间信号生成模块220的结构分别进行阐述。
与传统的动态视觉传感器的像素采集电路类似，运动检测模块210包括：相互耦接的光电探测子模块212、触发生成子模块214和读出接口子模块216。其中，光电探测子模块212实时监测照射在其上的光信号，并输出相应的电信号。触发生成子模块214的第一输入端耦接到光电探测子模块212，其第一输出端耦接到读出接口子模块216，且第二输出端耦接到时间信号生成模块220。触发生成子模块214在电信号满足预定触发条件时，生成触发生成信号给读出接口子模块216和时间信号生成模块220。读出接口子模块216耦接到触发生成子模块214，以便与读出单元130进行通信。基于运动检测模块210在现有的动态视觉传感器中已经存在，故此处不再展开赘述。
如前文所述，时间信号在像素采集电路200进入触发状态时表征一个瞬时电压脉冲、且之后随时间而衰减。根据本发明的实施方式，时间信号可以以指数形式衰减、也可以以线性形式衰减，不限于此。
根据时间信号的衰减形式的不同，图3和图5分别给出了根据本发明的实施例的两种时间信号生成模块220的示意图。其中，图3示出的是指数形式的衰减，图5示出的是线性形式的衰减。
如图3所示，指数衰减型的时间信号生成模块220包括：第一开关(K1)、 第一电容(C1)、第一电阻(R1)、第一缓冲器(B1)和第二开关(K2)。其中，第一开关(K1)的第一端耦接固定电压(VA)，其第二端耦接第一缓冲器(B1)的第一端。第一电容(C1)和第一电阻(R1)相并联，其一端接地，另一端耦接于第一开关(K1)的第二端以及第一缓冲器(B1)的第一端。第一缓冲器(B1)的第二端耦接至第二开关(K2)的第一端。第二开关(K2)的第二端接至时间信号列输出线。此外，第一开关(K1)由运动检测模块210输出的像素触发信号来控制，当像素触发信号有效时，第一开关(K1)闭合。第二开关(K2)由读出单元130输出的时间信号行选择信号来控制，当时间信号行选择信号有效时，第二开关(K2)闭合。在一种实施例中，第一电阻(R1)为可调电阻。根据第一电容(C1)和第一电阻(R1)组成的时间常数，来确定时间信号的衰减速度。
图4A-图4D示出了如图3所示出的时间信号生成模块220所产生的时间信号的波形图。
其中图4A示出的是像素采集电路200单次触发时的时间信号，图4B示出的是像素采集电路200多次触发时的时间信号(此处示出的为连续两次的触发)。在时间信号图中，横轴表示时间t，纵轴表示像素触发信号VT。图4C和图4D分别示出了对应于图4A和图4B的时间信号的波形图。在波形图中，横轴表示时间t，纵轴表示输出的时间信号的电压值V1。
结合图4A和图4C所示，当像素采集电路未被触发时(t<t1)，像素触发信号VT为低电平，第一开关K1断开，V1通过第一电阻R1放电到零点位；t1时刻，像素采集电路200被触发，像素触发信号VT为一个很短的脉冲信号，此时第一开关K1闭合，第一电容C1被强制充电至固定电压VA，即V1在t1时刻被瞬间拉升至VA；随后像素触发信号无效，第一开关K1再次断开，第一电容C1上的电荷通过第一电阻R1慢慢放电，因而V1以指数形式逐渐衰减；当t大于t3时，V1衰减为零电位。
结合图4B和图4D所示，像素采集电路200分别在t1时刻和t2时刻被触发，并且两次触发相距时间较短。在t1时刻，像素触发信号VT给出一个窄脉冲，它使得第一开关K1闭合、并将时间信号V1强制拉至固定电压VA，随后第一开关K1断开，V1开始以指数形式衰减。紧接着，像素采集电路200 在t2时刻被再次触发，像素触发信号VT再次给出一个窄脉冲并将时间信号V1再次拉至VA，随后第一开关K1断开，V1以指数形式衰减。在该像素采集电路被读出时，时间信号行选择信号有效，第二开关K2闭合，V1经第一缓冲器B1被送至时间信号列输出线，继而被列选择模块134获取并输出。
通过上述时间信号的波形图可以看出，时间信号在某时刻的瞬时幅值可以表征像素采集电路被触发时刻距离当前时刻的时间信息，即，瞬时幅值越高，距离像素采集电路被触发时刻越近，瞬时幅值越低，距离像素采集电路被触发时刻越远。如前文所述，时间信号衰减的速度由第一电容C1和第一电阻R1组成的时间常数决定，为了方便调节该衰减速度，第一电阻R1可以是一个可调电阻，根据不同的后端应用需求改变不同的阻值，来调节时间信号衰减的速度。
图5示出了线性衰减型的时间信号生成模块220，其与图3的区别在于，用恒流源I替代了第一电阻R1，当第三开关K3断开后，第二电容C2上存储的电荷通过恒流源I放电，从而形成随时间线性衰减的时间信号(V2)的波形。如图5所示，该时间信号生成模块220包括：第三开关(K3)、第二电容(C2)、恒流源(I)、第二缓冲器(B2)和第四开关(K4)。其中，第三开关(K3)，其第一端耦接固定电压(VA)，其第二端耦接第二缓冲器(B2)的第一端。第二电容(C2)和恒流源(I)并联，其一端接地，另一端耦接于第三开关(K3)的第二端以及第二缓冲器(B2)的第一端。第二缓冲器(B2)的第二端耦接至第四开关(K4)的第一端，第四开关(K4)的第二端接至时间信号列输出线。在该时间信号生成模块220中，第三开关(K3)由运动检测模块210输出的像素触发信号来控制，当像素触发信号有效时，第三开关(K3)闭合。第四开关(K4)由读出单元130输出的时间信号行选择信号来控制，当时间信号行选择信号有效时，第四开关(K4)闭合。在一种实施例中，根据第二电容(C2)和恒流源(I)确定时间信号的衰减速度。
图6A-图6D示出了如图5所示出的时间信号生成模块220所产生的时间信号的波形图。其中图6A示出的是像素采集电路200被单次触发时的时间信号，图6B示出的是像素采集电路200被多次触发时的时间信号(在此为连续两次的触发)。在时间信号图中，横轴表示时间t，纵轴表示像素触发信号 VT。图6C和图6D分别示出了对应于图6A和图6B的时间信号的波形图。在波形图中，横轴表示时间t，纵轴表示输出的时间信号的电压值V2。
结合图6A和图6C所示，当像素采集电路未被触发时(t<t1)，像素触发信号VT为低电平，第三开关K3断开，V2通过恒流源I放电到零点位；t1时刻，像素采集电路200被触发，像素触发信号VT为一个很短的脉冲信号，此时第三开关K3闭合，第二电容C2被强制充电至固定电压VA，即V2在t1时刻被瞬间拉升至VA；随后像素触发信号无效，第三开关K3再次断开，第二电容C2上的电荷通过恒流源I放电，因而V2以线性形式进行衰减；当t大于t3时，V2衰减为零电位。
同理，结合图6B和图6D所示，像素采集电路200分别在t1时刻和t2时刻被触发，并且两次触发相距时间较短。在t1时刻，像素触发信号VT给出一个窄脉冲，它使得第三开关K3闭合、并将时间信号V2强制拉至固定电压VA，随后第三开关K3断开，V2开始以线性形式衰减。紧接着，像素采集电路200在t2时刻被再次触发，像素触发信号VT再次给出一个窄脉冲并将时间信号V2再次拉至VA，随后第三开关K1断开，V2以线性形式衰减。在该像素采集电路200被读出时，时间信号行选择信号有效，第四开关K2闭合，V2经第二缓冲器B2被送至时间信号列输出线，继而被列选择模块134获取并输出。
通过图6C所示的单次触发的时间信号的波形图，可以更好地说明该时间信号瞬时幅值所表征的时间信息。为了叙述的方便，假设线性放电时间(t1至t3)为20ms，即在20ms后时间信号将由VA完全放电为零电位。另外，为便于理解，将瞬时电压幅值按照VA的幅值进行归一化并称为归一化幅值，其范围为从0至1。如图6C所示，在像素采集电路200被读出时，如果时间信号输出线上获取的归一化幅值为1，则说明该像素采集电路200恰好在该读出时刻被触发，或者说明到该读出时刻，该像素采集电路200已经触发了0ms；如果获取的归一化幅值为0.75(即图6C中所示的VB)，则说明到该读出时刻，该像素采集电路200已经触发了5ms；如果获取的归一化幅值为0.5(即图6C中所示的VC)，则说明到该读出时刻，该像素采集电路200已经触发了10ms；如果获取的归一化幅值为0，则说明该像素采集电路200已经被触 发了超过20ms，或者该像素采集电路200根本就没有触发。触发很久的像素采集电路200可以被认为是没有触发的像素采集电路，这是因为输出的时间表面帧仅关注当前时刻以及之前临近时间内所发生的运动，而不关注很早之前发生的运动。应当了解，此处对线性放电时间的假设和归一化处理仅为了便于理解，本发明的实施例并不受限于此。
如前文所述，时间信号衰减的速度可由第二电容C2和恒流源I的大小决定。第二电容C2越大、恒流源I越小，时间信号衰减的速度越慢；第二电容C2越小、恒流源I越大，时间信号衰减的速度越快。
此外，需要指出的是，图3和图5所示出的仅仅是代表指数衰减型和线性衰减型的时间信号生成模块220的示例，其他衰减形式的时间信号生成模块220亦可以通过不同的方式来实现，在此不再赘述。
下面通过一个简单的场景来进一步说明，基于本公开的图像传感器100所输出的时间表面帧的信息。图7示出了根据本发明一个实施例的视场的场景示意图。图8A-图8G分别示出了在图7所示的场景中，输出的时间表面帧的示意图。
为了叙述的方便，假定图像传感器100的像素采集电路阵列110包含6行6列的像素采集电路200。如图7所示，像素采集电路阵列110所对应的视场区域为白色背景，一个黑色的运动物体A(为区分运动物体A，在图7中采用网格线填充该黑色运动物体A)从视场区域的左下方进入，并沿着水平方向一直运动至视场区域的最右侧，其运动轨迹覆盖了第4至第6行的所有像素采集电路。
假定运动物体A在0时刻位于视场区域的左侧边界，且其运动速度为5ms/像素采集电路，图8A用斜线“/”标出了物体A在运动过程中触发的像素采集电路，并对应标出其相应的触发时间。例如，第5行第4列的像素采集电路在第20ms时被触发。无填充的像素采集电路200表示在该过程中未被触发。如前所述，一个完整的时间表面帧的输出需要时间信号行选择子模块1324和时间信号列选择子模块1344，在读出单元130的控制下逐行扫描整个像素采集电路阵列110，以获取每个像素采集电路200的时间信号生成模块220所输出的时间信号。
此处的像素采集电路200采用如图5所示的线性衰减型时间信号生成模块220，同时，设放电时间为20ms、并对输出幅值进行归一化处理，来说明输出关于像素采集电路的时间信息(即，时间表面帧)的过程。
当一个像素采集电路200的触发时间越临近当前时间表面帧的读出时刻时，其对应的归一化幅值越接近于1，反之，其对应的归一化幅值越接近于0。假定读出单元130每隔5ms获取一幅时间表面帧，那么从第5ms至第30ms，图像传感器100共输出6幅时间表面帧，分别如图8B-8G所示。其中，时间表面帧分别记作时间表面帧1、时间表面帧2、时间表面帧3、…、时间表面帧6，其对应的输出时间分别为：5ms、10ms、15ms、20ms、25ms、30ms。
对应于图8B(即，时间表面帧1)，其读出时刻为第5ms。在该读出时刻及之前，只有第1列的第4至6行的像素采集电路在第5ms被触发，因此在时间表面帧1中，这几个像素采集电路对应的归一化幅值为1，而其余像素采集电路的归一化幅值全部为0。
对应于图8C(即，时间表面帧2)，其读出时刻为第10ms。在该读出时刻及之前，第1列的第4至6行的像素采集电路在第5ms被触发，因此在时间表面帧2中，这几个像素采集电路对应的归一化幅值为0.75(0.75＝1-(10ms-5ms)/20ms)；第2列的第4至6行像素采集电路在第10ms被触发，因此在时间表面帧2中，这几个像素采集电路对应的归一化幅值为1；而其余像素采集电路的归一化幅值全部为0。
对应于图8D(即，时间表面帧3)，其读出时刻为15ms。在该读出时刻及之前，第1列的第4至6行的像素采集电路在第5ms被触发，因此在时间表面帧3中，这几个像素采集电路对应的归一化幅值为0.5(0.5＝1-(15ms-5ms)/20ms)；第2列的第4至6行像素采集电路在第10ms被触发，因此在时间表面帧3中，这几个像素采集电路对应的归一化幅值为0.75(0.75＝1-(15ms-10ms)/20ms)；第3列的第4至6行像素采集电路在第15ms被触发，因此在时间表面帧3中，这几个像素采集电路对应的归一化幅值为1；其余像素采集电路的归一化幅值均为0。
对应于图8E(即，时间表面帧4)，其读出时刻为20ms。在该读出时刻及之前，第1列的第4至6行的像素采集电路在第5ms被触发，因此在时间 表面帧4中，这几个像素采集电路对应的归一化幅值为0.25(0.25＝1-(20ms-5ms)/20ms)；第2列的第4至6行像素采集电路在第10ms被触发，因此在时间表面帧4中，这几个像素采集电路对应的归一化幅值为0.5(0.5＝1-(20ms-10ms)/20ms)；第3列的第4至6行像素采集电路在第15ms被触发，因此在时间表面帧4中，这几个像素采集电路对应的归一化幅值为0.75(0.75＝1-(20ms-15ms)/20ms)；第4列的第4至6行像素采集电路在第20ms被触发，因此在时间表面帧4中，这几个像素采集电路对应的归一化幅值为1；其余像素采集电路的归一化幅值均为0。
对应于图8F(即，时间表面帧5)，其读出时刻为25ms。在该读出时刻及之前，第1列的第4至6行的像素采集电路在第5ms被触发，因此在时间表面帧5中，这几个像素采集电路对应的归一化幅值为0(0＝1-(25ms-5ms)/20ms)；第2列的第4至6行像素采集电路在第10ms被触发，因此在时间表面帧5中，这几个像素采集电路对应的归一化幅值为0.25(0.25＝1-(25ms-10ms)/20ms)；第3列的第4至6行像素采集电路在第15ms被触发，因此在时间表面帧5中，这几个像素采集电路对应的归一化幅值为0.5(0.5＝1-(25ms-15ms)/20ms)；第4列的第4至6行像素采集电路在第20ms被触发，因此在时间表面帧5中，这几个像素采集电路对应的归一化幅值为0.75(0.75＝1-(25ms-20ms)/20ms)；第5列的第4至6行像素采集电路在第25ms被触发，因此在时间表面帧5中，这几个像素采集电路对应的归一化幅值为1；其余像素采集电路的归一化幅值均为0。
对应于图8G(即，时间表面帧6)，其读出时刻为30ms。在该读出时刻及之前，第1列的第4至6行的像素采集电路在第5ms被触发，因此在时间表面帧6中，这几个像素采集电路对应的归一化幅值为0(当前读出时刻距离这些像素采集电路上次被触发时刻的时间大于放电时间20ms)；第2列的第4至6行像素采集电路在第10ms被触发，因此在时间表面帧6中，这几个像素采集电路对应的归一化幅值为0(0＝1-(30ms-10ms)/20ms)；第3列的第4至6行像素采集电路在第15ms被触发，因此在时间表面帧6中，这几个像素采集电路对应的归一化幅值为0.25(0.25＝1-(30ms-15ms)/20ms)；第4列的第4至6行像素采集电路在第20ms被触发，因此在时间表面帧6中，这 几个像素采集电路对应的归一化幅值为0.5(0.5＝1-(30ms-20ms)/20ms)；第5列的第4至6行像素采集电路在第25ms被触发，因此在时间表面帧6中，这几个像素采集电路对应的归一化幅值为0.75(0.75＝1-(30ms-25ms)/20ms)；第6列的第4至6行像素采集电路在第30ms被触发，因此在时间表面帧6中，这几个像素采集电路对应的归一化幅值为1。
根据本公开的图像传感器100，在像素采集电路200中添加了时间信号生成模块220，该时间信号生成模块220生成一个本地的时间信号，该时间信号一般是一个电压信号，其在像素采集电路200检测到事件时表现为一个瞬时的电压脉冲，而后，该时间信号逐渐衰减以表征时间的流逝。外围的读出单元130添加了时间信号行选择子模块1324和时间信号列选择子模块1344，其定时地扫描所有像素采集电路200中时间信号生成模块220所输出的时间信号，从而形成时间表面帧并输出。
综上，该图像传感器100既保留了传统动态视觉传感器的异步检测动态信息的优点，又可以直接输出表征像素采集电路时间信息的时间表面帧，给到后端的图像处理器(例如，执行机器视觉算法的图像处理器)。该时间表面帧可以被后端的机器视觉算法用来作为训练的样本以及实现物体识别、追踪等任务。这样，该图像传感器100能够有效节省图像处理系统后端进行数据转换所需的存储和算力的开销，从而提高图像处理系统的效率。
在此处所提供的说明书中，说明了大量具体细节。然而，能够理解，本发明的实施例可以在没有这些具体细节的情况下被实践。在一些实例中，并未详细示出公知的方法、结构和技术，以便不模糊对本说明书的理解。
类似地，应当理解，为了精简本公开并帮助理解各个发明方面中的一个或多个，在上面对本发明的示例性实施例的描述中，本发明的各个特征有时被一起分组到单个实施例、图、或者对其的描述中。然而，并不应将该公开的方法解释成反映如下意图：即所要求保护的本发明要求比在每个权利要求中所明确记载的特征更多特征。更确切地说，如下面的权利要求书所反映的那样，发明方面在于少于前面公开的单个实施例的所有特征。因此，遵循具体实施方式的权利要求书由此明确地并入该具体实施方式，其中每个权利要求本身都作为本发明的单独实施例。
本领域那些技术人员应当理解在本文所公开的示例中的设备的模块或单元或组件可以布置在如该实施例中所描述的设备中，或者可替换地可以定位在与该示例中的设备不同的一个或多个设备中。前述示例中的模块可以组合为一个模块或者此外可以分成多个子模块。
本领域那些技术人员可以理解，可以对实施例中的设备中的模块进行自适应性地改变并且把它们设置在与该实施例不同的一个或多个设备中。可以把实施例中的模块或单元或组件组合成一个模块或单元或组件，以及此外可以把它们分成多个子模块或子单元或子组件。除了这样的特征和/或过程或者单元中的至少一些是相互排斥之外，可以采用任何组合对本说明书(包括伴随的权利要求、摘要和附图)中公开的所有特征以及如此公开的任何方法或者设备的所有过程或单元进行组合。除非另外明确陈述，本说明书(包括伴随的权利要求、摘要和附图)中公开的每个特征可以由提供相同、等同或相似目的的替代特征来代替。
此外，本领域的技术人员能够理解，尽管在此所述的一些实施例包括其它实施例中所包括的某些特征而不是其它特征，但是不同实施例的特征的组合意味着处于本发明的范围之内并且形成不同的实施例。例如，在下面的权利要求书中，所要求保护的实施例的任意之一都可以以任意的组合方式来使用。
此外，所述实施例中的一些在此被描述成可以由计算机系统的处理器或者由执行所述功能的其它装置实施的方法或方法元素的组合。因此，具有用于实施所述方法或方法元素的必要指令的处理器形成用于实施该方法或方法元素的装置。此外，装置实施例的在此所述的元素是如下装置的例子：该装置用于实施由为了实施该发明的目的的元素所执行的功能。
如在此所使用的那样，除非另行规定，使用序数词“第一”、“第二”、“第三”等等来描述普通对象仅仅表示涉及类似对象的不同实例，并且并不意图暗示这样被描述的对象必须具有时间上、空间上、排序方面或者以任意其它方式的给定顺序。
尽管根据有限数量的实施例描述了本发明，但是受益于上面的描述，本技术领域内的技术人员明白，在由此描述的本发明的范围内，可以设想其它 实施例。此外，应当注意，本说明书中使用的语言主要是为了可读性和教导的目的而选择的，而不是为了解释或者限定本发明的主题而选择的。因此，在不偏离所附权利要求书的范围和精神的情况下，对于本技术领域的普通技术人员来说许多修改和变更都是显而易见的。对于本发明的范围，对本发明所做的公开是说明性的，而非限制性的，本发明的范围由所附权利要求书限定。

Claims (13)

  1. 一种像素采集电路，包括：
    运动检测模块，适于检测视场中的光强变化，并在光强变化超出设定阈值时，生成像素触发信号，来指示所述像素采集电路进入触发状态；
    时间信号生成模块，与所述运动检测模块耦接，适于在接收到所述像素触发信号时生成时间信号，所述时间信号用于表征所述像素采集电路被触发时刻的时间信息；
    其中，所述运动检测模块和所述时间信号生成模块分别耦接到外围的读出单元，以便所述读出单元对像素采集电路进行扫描来输出像素采集电路的时间信息。
  2. 如权利要求1所述的像素采集电路，其中，
    所述时间信号在所述像素采集电路进入触发状态时指示一个瞬时电压脉冲、且之后随时间而衰减。
  3. 如权利要求1或2所述的像素采集电路，其中，所述运动检测模块包括：
    光电探测子模块，适于实时监测照射在其上的光信号，并输出相应的电信号；
    触发生成子模块，其第一输入端耦接到所述光电探测子模块，其第一输出端耦接到读出接口子模块，其第二输出端耦接到所述时间信号生成模块，所述触发生成子模块适于在所述电信号满足预定触发条件时，生成触发生成信号给读出接口子模块和所述时间信号生成模块；以及
    读出接口子模块，耦接到所述触发生成子模块，适于与读出单元进行通信。
  4. 如权利要求1-3中任一项所述的像素采集电路，其中，所述时间信号生成模块包括：
    第一开关(K1)，其第一端耦接固定电压(VA)，其第二端耦接缓冲器(B1)的第一端；
    并联的第一电容(C1)和第一电阻(R1)，其一端接地，另一端耦接于所述第一开关(K1)的第二端以及第一缓冲器(B1)的第一端；
    第一缓冲器(B1)，其第二端耦接至第二开关(K2)的第一端；以及
    第二开关(K2)，其第二端接至时间信号列输出线。
  5. 如权利要求1-3中任一项所述的像素采集电路，其中，所述时间信号生成模块包括：
    第三开关(K3)，其第一端耦接固定电压(VA)，其第二端耦接第二缓冲器(B2)的第一端；
    并联的第二电容(C2)和恒流源(I)，其一端接地，另一端耦接于所述第三开关(K3)的第二端以及第二缓冲器(B2)的第一端；
    第二缓冲器(B2)，其第二端耦接至第四开关(K4)的第一端；以及
    第四开关(K4)，其第二端接至时间信号列输出线。
  6. 如权利要求4或5所述的像素采集电路，其中，
    所述第一开关(K1)或第三开关(K3)由所述运动检测模块输出的像素触发信号来控制，当所述像素触发信号有效时，第一开关(K1)或第三开关(K3)闭合；以及
    所述第二开关(K2)或第四开关(K4)由所述读出单元输出的时间信号行选择信号来控制，当所述时间信号行选择信号有效时，第二开关(K2)或第四开关(K4)闭合。
  7. 如权利要求4所述的像素采集电路，其中，
    所述第一电阻(R1)为可调电阻；以及
    根据第一电容(C1)和第一电阻(R1)组成的时间常数确定时间信号的衰减速度。
  8. 如权利要求5所述的像素采集电路，其中，
    根据第二电容(C2)和恒流源(I)确定时间信号的衰减速度。
  9. 一种图像传感器，包括：
    像素采集电路阵列，包括多个如权利要求1-8中任一项所述的像素采集电 路；
    全局控制单元，适于控制所述像素采集电路阵列的初始工作状态；
    读出单元，适于逐行地对所述像素采集电路阵列进行扫描，来输出关于所述像素采集电路阵列的时间信息。
  10. 如权利要求9所述的图像传感器，其中，所述读出单元包括：
    行选择模块，适于在行方向上管理所述像素采集电路阵列；
    列选择模块，适于在列方向上管理所述像素采集电路阵列；
    读出控制模块，分别与所述行选择模块和所述列选择模块相耦接，适于控制所述行选择模块与所述列选择模块的输出。
  11. 如权利要求9或10所述的图像传感器，其中，所述行选择模块包括：
    运动检测行选择子模块，通过运动检测行请求线和运动检测行选择线耦接到所述像素采集电路中的运动检测模块，适于对来自所述像素采集电路阵列的行请求信号进行响应；
    时间信号行选择子模块，通过时间信号行选择线耦接到所述像素采集电路中的时间信号生成模块，适于逐行地将所述像素采集电路的时间信号行选择线置为有效。
  12. 如权利要求9-11中任一项所述的图像传感器，其中，所述列选择模块包括：
    运动检测列选择子模块，通过运动检测列请求线和运动检测列选择线耦接到所述像素采集电路中的运动检测模块，适于对来自所述像素采集电路阵列的列请求信号进行响应，其中所述列请求信号在所述像素采集电路得到行响应时有效；
    时间信号列选择子模块，通过时间信号列输出线耦接到所述像素采集电路中的时间信号生成模块，适于通过所述时间信号列输出线读取所述像素采集电路的时间信号，其中，所述像素采集电路在与之耦接的时间信号行选择线有效时，输出时间信号。
  13. 如权利要求9-12中任一项所述的图像传感器，其中，
    所述全局控制单元还适于在所述图像传感器上电时，复位所述像素采集电路阵列，以及，还适于在像素采集电路阵列保持稳定的初始状态时，解除复位，使得所述像素采集电路阵列开始工作。
PCT/CN2020/078229 2019-11-29 2020-03-06 一种像素采集电路及图像传感器 Ceased WO2021103341A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP2022525427A JP7428307B2 (ja) 2019-11-29 2020-03-06 ピクセル取得回路及び画像センサ
EP20892079.3A EP4068760B1 (en) 2019-11-29 2020-03-06 Image sensor
US17/731,305 US12015861B2 (en) 2019-11-29 2022-04-28 Pixel collection circuit and image sensor

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911203365.3A CN111107287B (zh) 2019-11-29 2019-11-29 一种像素采集电路及图像传感器
CN201911203365.3 2019-11-29

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/731,305 Continuation US12015861B2 (en) 2019-11-29 2022-04-28 Pixel collection circuit and image sensor

Publications (1)

Publication Number Publication Date
WO2021103341A1 true WO2021103341A1 (zh) 2021-06-03

Family

ID=70421369

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/078229 Ceased WO2021103341A1 (zh) 2019-11-29 2020-03-06 一种像素采集电路及图像传感器

Country Status (5)

Country Link
US (1) US12015861B2 (zh)
EP (1) EP4068760B1 (zh)
JP (1) JP7428307B2 (zh)
CN (1) CN111107287B (zh)
WO (1) WO2021103341A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11694304B2 (en) * 2019-11-26 2023-07-04 Samsung Electronics Co., Ltd. Jointly learning visual motion and confidence from local patches in event cameras
WO2022047651A1 (zh) * 2020-09-02 2022-03-10 华为技术有限公司 视觉传感器芯片及其操作方法和电子设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8687060B1 (en) * 2009-12-28 2014-04-01 Cognex Corporation System and method for providing distance-based pulses relative to motion of a surface scanned by a vision system
CN108449557A (zh) * 2018-03-23 2018-08-24 上海芯仑光电科技有限公司 像素采集电路、光流传感器和光流及图像信息采集系统
CN108632546A (zh) * 2017-03-17 2018-10-09 上海芯仑光电科技有限公司 像素采集电路、光流传感器及图像采集系统
US20180295298A1 (en) * 2017-04-06 2018-10-11 Samsung Electronics Co., Ltd. Intensity image acquisition from dynamic vision sensors
CN109842768A (zh) * 2019-01-29 2019-06-04 上海芯仑光电科技有限公司 一种像素采集电路及图像传感器

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
AT504582B1 (de) 2006-11-23 2008-12-15 Arc Austrian Res Centers Gmbh Verfahren zur generierung eines bildes in elektronischer form, bildelement für einen bildsensor zur generierung eines bildes sowie bildsensor
KR102022970B1 (ko) * 2013-04-30 2019-11-04 삼성전자주식회사 시각 센서에 기반하여 공간 정보를 감지하는 장치 및 방법
KR102136055B1 (ko) 2014-01-08 2020-07-21 삼성전자 주식회사 오픈-루프 증폭기를 포함하는 비전 센서 칩, 이의 동작 방법, 및 이를 포함하는 데이터 처리 시스템
CN106162000B (zh) * 2016-07-08 2019-03-15 上海芯仑光电科技有限公司 像素采集电路、图像传感器及图像采集系统
JP7074954B2 (ja) * 2016-12-30 2022-05-25 ソニー アドバンスト ビジュアル センシング アーゲー 動的ビジョンセンサアーキテクチャ
CN108827461B (zh) * 2018-04-25 2019-05-03 上海芯仑光电科技有限公司 像素采集电路及光流传感器
WO2019211317A1 (en) 2018-04-30 2019-11-07 Prophesee Systems and methods for asynchronous, time-based image sensing
CN109377516B (zh) * 2018-11-01 2020-09-29 清华大学 一种使用动态视觉数据的光流计算方法
CN109842767B (zh) * 2019-01-09 2020-07-14 上海芯仑光电科技有限公司 一种防闪光电路组件及图像传感器

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8687060B1 (en) * 2009-12-28 2014-04-01 Cognex Corporation System and method for providing distance-based pulses relative to motion of a surface scanned by a vision system
CN108632546A (zh) * 2017-03-17 2018-10-09 上海芯仑光电科技有限公司 像素采集电路、光流传感器及图像采集系统
US20180295298A1 (en) * 2017-04-06 2018-10-11 Samsung Electronics Co., Ltd. Intensity image acquisition from dynamic vision sensors
CN108449557A (zh) * 2018-03-23 2018-08-24 上海芯仑光电科技有限公司 像素采集电路、光流传感器和光流及图像信息采集系统
CN109842768A (zh) * 2019-01-29 2019-06-04 上海芯仑光电科技有限公司 一种像素采集电路及图像传感器

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4068760A4 *

Also Published As

Publication number Publication date
EP4068760A4 (en) 2023-01-04
JP7428307B2 (ja) 2024-02-06
CN111107287A (zh) 2020-05-05
JP2023501237A (ja) 2023-01-18
US20220256109A1 (en) 2022-08-11
EP4068760B1 (en) 2025-07-16
EP4068760A1 (en) 2022-10-05
CN111107287B (zh) 2021-06-04
US12015861B2 (en) 2024-06-18

Similar Documents

Publication Publication Date Title
JP7383616B2 (ja) イベントベースセンサから信号を出力するための方法、およびそのような方法を使用するイベントベースセンサ
US12022211B1 (en) Pixel collection circuit, optical flow sensor, and optical flow and image information collection system
CN108632546B (zh) 像素采集电路、光流传感器及图像采集系统
CN111709967B (zh) 一种目标检测方法、目标跟踪方法、装置及可读存储介质
CN107238727B (zh) 基于动态视觉传感器芯片的光电式转速传感器及探测方法
JP2020505801A (ja) 事象ベースの視覚センサのためのデータ速度制御
US11815394B2 (en) Photoelectric conversion device, method of controlling photoelectric conversion device, and information processing apparatus
JP7099783B2 (ja) 画素収集回路及びオプティカルフローセンサー
US12425750B2 (en) Pixel collection circuit and image sensor
CN108200362B (zh) 基于空间对比度的仿生视网膜摄像电路及子电路
EP3267678B1 (en) Pixel acquisition circuit, image sensor and image acquisition system
CN115413320A (zh) 像素传感器系统
WO2021103341A1 (zh) 一种像素采集电路及图像传感器
WO2022003013A1 (en) Eye tracking device, eye tracking method, and computer-readable medium
US20240377539A1 (en) Control of the brightness of a display
EP3376754A1 (en) Pixel acquisition circuit, optical flow sensor, and image acquisition system
Brajovic et al. Temporal photoreception for adaptive dynamic range image sensing and encoding
US12477239B2 (en) Event-based sensor system and method for operating an event-based sensor system
CN113747090A (zh) 一种像素采集电路及图像传感器
Chan et al. An address-event vision sensor for multiple transient object detection
CN107866058A (zh) 一种基于多层映射解析定位的乒乓球运动系统
CN109788219B (zh) 一种用于人眼视线追踪的高速cmos图像传感器读出方法
IL318724B1 (en) Event camera with optical element for modifying point spread function to improve signal to noise ratio
CN118695091A (zh) 一种事件传感器的双模式控制方法及相关装置
CN121967648A (zh) 基于事件的传感器的投影指向的标定方法与系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20892079

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022525427

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020892079

Country of ref document: EP

Effective date: 20220629

WWG Wipo information: grant in national office

Ref document number: 2020892079

Country of ref document: EP