JPH04322575A - Analog/digital conversion system - Google Patents

Abstract

PURPOSE:To digitize an analog signal with high accuracy by varying a reference voltage in response to a level of a signal in the case of digitizing the analog signal. CONSTITUTION:A dark current signal SD is read from a frame memory 15 synchronously with an output of a picture signal SO1 under the control of a controller 18. The signal SD is inputted to a D/A converter 16, in which the signal is converted into an analog signal by using a reference voltage RB used for A/D conversion. Since the dark current signal SD is digitized with high accuracy, the digital signal is reproduced into an analog signal more accurately. Then an output characteristic of the dark current signal SD depending on temperature is controlled by a control amplifier 17 and the result is inputted to a computing element 7. The computing element 7 subtracts the signal SD converted again into an analog signal from the picture signal SO1 and an obtained picture signal SO2 is inputted to an A/D converter 9. The converter 9 uses a reference voltage RA to digitize the picture signal SO2. R, G, B signals are separated from the picture signal SO2 by a sample-and-hold circuit 11 and they are transferred to other device from an output terminal 20 via a white balance circuit 12, a color correction matrix 13 and a gamma correction circuit 14.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an analog/digital conversion method used in image signal processing and the like.

0002

2. Description of the Related Art FIG. 3 is a schematic block diagram showing a conventional analog/digital conversion method (A/D conversion method) for black non-uniformity correction of image signals disclosed in Japanese Patent Application Laid-Open No. 141868/1986. . Black non-uniformity correction is to correct image unevenness due to unevenness in dark current flowing through each pixel of an image sensor. JP-A-62-141868 describes an example in which the result of image signal processing is output as an analog signal, but the invention is not limited to analog image signal output, and is applicable to digital image signals. It has the same effect when outputting . Therefore,
FIG. 3 shows the Japanese Patent Laid-Open No. 6
A schematic block diagram of the method disclosed in Japanese Patent No. 2-141868 is shown.

In the figure, 1 is an image sensor, 2 is a color filter, 3 is an iris mechanism, 4 is an optical system, 5 is a temperature sensor, 6 is an arithmetic unit, 8 is an AGC amplifier, 9 is an A/D converter, 10 is a 1→2 selector, 11 is a sample/hold circuit, 12 is a white balance circuit, 13 is a color correction matrix, 14 is a γ correction circuit, 15 is a frame memory, 16 is a D/A converter, 17 is a control amplifier, 1
8 is a controller, and 20 is an output terminal.

When a reading start button (not shown) is pressed, the iris mechanism 3 is activated under the control of the controller 18.
blocks the light emitted from the optical system 4, imaging is performed in a completely black state, and the image sensor 1 outputs a dark current signal SD. This signal SD is the sample/hold circuit 6
, an arithmetic unit 7 and an AGC amplifier 8 to an A/D converter 9 . At this time, under the control of the controller 18, both the arithmetic unit 7 and the AGC amplifier 8 output the input signals as they are. The A/D converter 9 uses a predetermined reference voltage RA supplied under the control of the controller 18 to digitize the dark current signal SD. The digitized dark current signal SD is sent to the frame memory 1 via a 1→2 selector 10 controlled by a controller 18.
5 is written. Writing to the frame memory 15 is performed under the control of the controller 18 in synchronization with the output of the image sensor 1.

[0005] Subsequently, under the control of the controller 18,
The iris mechanism 3 opens and the light emitted from the optical system 4 passes through to form an image on the light receiving surface of the image sensor 1. The image sensor 1 outputs an image signal SO1 obtained as a result of forming the image. The output image signal SO1 is input to the arithmetic unit 7 via the sample/hold circuit 6. At this time, the image signal SO1 is input to the arithmetic unit 7, and at the same time, the dark current signal SD read out from the frame memory 15 is supplied via the D/A converter 16 and the control amplifier 17. Reading from the frame memory 15 is performed under the control of the controller 18 in synchronization with the output of the image signal SO1 from the image sensor 1. The D/A converter 16 converts the dark current signal SD into an analog signal using a predetermined reference voltage RA supplied under the control of the controller 18 .

Further, gain control of the control amplifier 17 is performed by a controller 18 based on a detection signal SR from the temperature sensor 5. Since the dark current changes greatly depending on the temperature, the dark current signal SD also changes in proportion to the temperature. Therefore, it is necessary to correct the dark current signal SD to a value corresponding to the temperature at the time of outputting the image signal SO1. By controlling the control amplifier 17 using the temperature sensor 5, the dark current signal SD read out from the frame memory 15
The level of can be corrected according to the temperature.

The computing unit 7 receives the image signal SO1 as described above.
and the dark current signal SD are input, and by subtracting SD from SO1, the image signal S from which the dark current signal SD has been removed is obtained.
O2 is obtained and inputted to the AGC amplifier 8. The AGC amplifier 8 multiplies the image signal SO2 by a predetermined constant and inputs the result to the A/D converter 9. The A/D converter 9 is connected to the controller 18
The image signal SO2 is digitized using the same reference voltage RA used for A/D conversion and D/A conversion of the dark current signal SD, which is supplied by the control of 11. The sample/hold circuit 11 receives the input image signal S.
O2 is classified into R, G, and B color signals, and each color signal SO2 (R), SO2 (G), SO2 (B)
are supplied to the white balance circuit 12 in parallel. A white balance circuit 12 performs white balance, a color correction matrix 13 performs color correction, and a γ correction circuit 14 performs γ correction. Signals SO2 (R), SO2 (G), SO2 output from the γ correction circuit 14
(B) is the output terminal OP(R), OP(G), OP(
B) Transferred from 20 to another device.

[0008]

[Problem to be Solved by the Invention] Conventional A/D conversion methods used in image signal processing, etc., are performed as described above, so quantization errors occur during digitization, resulting in poor signal processing accuracy. There was a problem.

The cause of this problem will be explained in detail below using the conventional example of black non-uniformity correction of the image signal. For example, in the case of an 8-bit output, the signal is digitized by setting the magnitude of the reference voltage RA to 255, and comparing it with that value to represent the magnitude of the signal. Usually A/
Reference voltage R of D converter 9 and D/A converter 16
A value slightly larger than the output signal when reading the reference white color is used for A. The dark current signal SD taken in when capturing an all-black image is considerably smaller than the signal when reading the reference white color. Therefore, the signal difference between each pixel when the dark current signal SD is taken in is also smaller than the reference white signal. Therefore, it is difficult to represent the small signal difference between each pixel when the dark current signal SD is taken in by the reference voltage RA set using the reference white signal, and as a result, the quantum Errors will occur. Naturally, the quantization error also exists on the dark current signal SD converted into an analog using a D/A converter, so it is also quantized on the signal SO2 obtained by subtracting the dark current signal SD from the image signal SO1. Errors exist.

The quantization error associated with digitization is 1/255 or less in the case of 8-bit output. However, for example, when density conversion is performed in signal processing, that is, Log function conversion, as shown in Figure 2, when reading a dark image, that is, when the input is small, the input signal differs by at most 1/255. However, the difference in output is 10/255
It often grows to a certain extent. 1/25
An error that is not noticeable if it is about 5, but becomes quite noticeable if it is different by 10/255, and as a result, the effect of correcting black non-uniformity is weakened.

[0011] That is, when a small signal is digitized using a large reference voltage, especially when there is processing to dynamically convert the input signal such as Log conversion or γ correction, the quantization error associated with digitization causes black The accuracy of non-uniformity correction decreases.

The present invention has been made to solve the above-mentioned problems, and its purpose is to more accurately represent signals during digitization and improve the accuracy of signal processing.

[0013]

[Means for Solving the Problems] The A/D conversion method according to the present invention includes: an A/D conversion means for digitizing an analog signal; a reference voltage supply means for supplying a reference voltage to the A/D conversion means; A switching means is provided for changing the magnitude of the reference voltage according to the magnitude of the signal.

[0014]

In the present invention, when the signal input to the A/D conversion means is small, a small reference voltage is supplied, and when the signal is large, a large reference voltage is supplied.

[0015]

【Example】

Example 1. An embodiment of the present invention will be described below. FIG. 1 is a schematic block diagram when the present invention is used to correct black non-uniformity of an image signal.

1 to 18 and 20 are the same as those in the conventional example. 19 is a 2→1 selector that switches the reference voltage supplied to the A/D converter 9.

An embodiment of the present invention will be described below with reference to the A/D converter 9.
, the D/A converter 16, the controller 18, and the 2→1 selector 19 will be mainly explained.

When capturing an all-black image, the 2→1 selector 19 is switched by a control signal from the controller 18, and a reference voltage RB smaller than the reference voltage RA used during image capturing is supplied to the A/D converter 9. For example, a reference voltage RB of 1V is supplied to a reference voltage RA of 5V. The A/D converter 9 digitizes the input dark current signal SD using a reference voltage RB. Since a small dark current signal is digitized using a small reference voltage, it is possible to accurately represent the difference in dark current signal between each pixel, that is, the unevenness of dark current. The digitized dark current signal SD is written into the frame memory 15 as in the conventional example.

When capturing an image, the 2→1 selector 19 is switched under the control of the controller 18, and the reference voltage RA is supplied to the A/D converter 9. The image signal SO1 output from the image sensor 1 is input to the arithmetic unit 7. Further, under the control of the controller 18, the dark current signal SD is read out from the frame memory 15 in synchronization with the output of the image signal SO1. The dark current signal SD read out from the frame memory 15 is read out. The dark current signal SD read out from the frame memory 15 is
The signal is input to the converter 16 and converted into an analog signal using the reference voltage RB used during A/D conversion. Since the dark current signal SD is digitized with high precision, it can be more accurately reproduced as an analog signal. Next, dark current signal SD
is inputted to the arithmetic unit 7 after its output characteristic, which varies depending on the temperature, is controlled by the control amplifier 17 as in the conventional example. The arithmetic unit 7 subtracts the re-analogized dark current signal SD from the image signal SO1 and inputs the resulting image signal SO2 to the A/D converter 9. A/D
The converter 9 digitizes the image signal SO2 using the supplied reference voltage RA.The digitized image signal SO2 is processed by the sample/hold circuit 11 to convert R, G, B, as in the conventional example. separated, white balance circuit 12, color correction matrix 13, and γ
The signal is transferred from the output terminal 20 to another device via the correction circuit 14.

Since the subtraction is performed using the dark current signal SD that has been re-analogized more accurately, the resulting image signal SO2 is an image signal with less image unevenness due to unevenness of the dark current.

Example 2. In the above embodiment, the reference voltages RA and RB are always output, and the controller 1
Although the reference voltage supplied to the A/D converter is changed by switching the 2→1 selector 19 under the control of the controller 8, it is also possible to switch the output of the reference voltages RA and RB under the control of the controller 18. It is.

Example 3. In the above embodiment, a case is shown in which there are two types of reference voltages, but it is also possible to use three or more types of reference voltages as necessary.

Example 4. In the above embodiment, the case where the present invention is used to correct black non-uniformity of an image signal is described.
It can be used in other signal processing when you want to process signals of different sizes with high precision.

[0024]

Effects of the Invention As explained above, the present invention changes the reference voltage applied to the A/D conversion means according to the magnitude of the signal, so it is possible to represent the signal more accurately during digitization, and the signal Processing accuracy can be improved.

[Brief explanation of drawings]

FIG. 1 is a schematic block diagram showing a black non-uniformity correction method for image signals according to a first embodiment of the present invention.

FIG. 2 is a diagram showing the relationship between input and output of processing for dynamically converting input data, such as Log conversion and γ correction, in image processing.

FIG. 3 is a schematic block diagram showing a conventional image signal black non-uniformity correction method.

[Explanation of symbols]

1 Image sensor 7 Arithmetic unit 9 A/D converter 15 Frame memory 16 D/A converter 18 Controller 19 2→1 selector

Claims (1)

[Claims] 1. Analog/digital conversion means for digitizing an analog signal; reference voltage supply means for supplying a reference voltage to the analog/digital conversion means; Analog/
Digital conversion method.