JPH0215100B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (A) Technical Field of the Invention The present invention relates to an image processing device, particularly an image processing device that can extract the contour of an object at high speed, which is necessary when recognizing an object in a robot or the like. It is related to.

(B) Background of the technology A robot has a grasping unit, or hand, that grasps an object, an observation unit, or eyes, that observes the object to be grasped, and an image that extracts the outline of the object to be grasped from the image data observed by the observation unit. Robots are equipped with processing devices and the like as their main components, and needless to say, it is particularly important for robots to operate these components in real time.

(C) Prior Art and Problems FIG. 1 is a diagram showing an outline of a visual recognition device commonly used in the past.

In the figure, image processing for extracting the outline of an object begins by raster scanning the object to be recognized using an ITV camera 7 or the like to obtain an image signal. Next, this image signal is binarized or A/D converted by the preprocessing circuit 2, and the obtained image signal as shown in FIG. 2 (the hatched part corresponds to the object) is stored in the image memory 3. .

Next, the image signal stored in the image memory 3 is transferred pixel by pixel to the main memory 7 via the I/O port 4 and the bus line 5 under the control of the central processing unit (hereinafter referred to as CPU) 6. . Even if the data transfer rate at this time is 100KB/sec, the data transfer of 512 x 512 pixels requires approximately
It takes 2.6 seconds.

Furthermore, in order to recognize what the object image is, contour points (P ₁ , It is necessary to extract P ₂ ...) and create a two-dimensional coordinate table of contour points as shown in Figure 4. However, if this process is executed by software using a microcomputer, it will take several tens of seconds. It will require.

That is, the microcomputer sequentially scans the image signals as shown in FIG. 2 stored in the main memory using, for example, a known 3x3 matrix circuit, and calculates the contours (P ₁ , P ₂ . . . ). is extracted, and only the contour point image signal as shown in FIG. 3 is stored again in the main memory 7. Next, only the re-stored contour point image signals are scanned, for example, sequentially from the upper left corner, and if an image signal of a contour point (P ₁ /P ₂ ...) is found, the coordinates of this contour point are calculated, Since a contour point coordinate table as shown in FIG. 4 is created, there is a drawback that processing takes time.

Therefore, in the specification attached to the application of Japanese Patent Application No. 58-25719 (filed on February 18, 1988), the applicant of this application has proposed that the outline points of the object image can be calculated in real time from the image signal obtained by the ITV camera. We developed an image processing device that extracts the images to speed up the processing. However, although this image processing device can certainly extract the contour points of an object at high speed, the contour points necessary to recognize the object from the contour points extracted by this device are To create a point coordinate table, the contour point image signal stored in the image memory shown in FIG. A contour point coordinate table must be created by processing.

Therefore, it takes about several seconds for just one contour extraction process of an object image, and when considering the time required for the subsequent recognition process, one operation takes about 2 to 5 seconds. Therefore, since the coordinate extraction time by software is several tens of seconds, practical processing cannot be performed for 2 to 3 seconds per robot cycle operation. The main memory 7 stores not only a contour point coordinate table but also a program for executing processing of the visual recognition device. Further, 8 indicates a console, and 9 indicates an auxiliary recording device.

(D) Object of the invention In view of the above-mentioned drawbacks, the object of the invention is to
To provide an image processing device capable of raster scanning an object using a camera or the like to obtain image data, extracting the outline of the object from the image data at high speed, and creating a contour point coordinate table in real time. be.

(E) Structure of the Invention According to the present invention, the object is to scan an object using an imaging means to obtain image data, extract contour points of the object from the image data, and further extract the contour points of the contour points. A scanning signal that generates a timing signal for scanning an object of the imaging means in an image processing device that calculates contour point coordinates on the imaging surface and performs shape recognition processing based on the contour point coordinates in a processing device. generating means, and an analog/digital converter for converting the image data into analog/digital to obtain digital image data.
digital conversion means; contour point extraction means for extracting contour points of the object from the digital image data based on the scanning timing signal; and coordinate generation means for generating contour point coordinates of the contour points based on the scanning timing signal. and a bidirectional memory that utilizes a part of a memory address space of a processing device, is directly connected to a bus line, is configured to be directly accessible to the processing device, and sequentially stores the contour point coordinates. It is configured by providing an image processing device with characteristics.

(F) Embodiments of the Invention Hereinafter, embodiments of the image processing apparatus according to the present invention will be described in detail with reference to the drawings.

FIG. 5 is a diagram for explaining an embodiment of the image processing apparatus according to the present invention. In the figure, the same parts as in FIG. 1 are given the same numbers.

A scanning signal generator 11 outputs a timing signal for scanning an object of the ITV camera 1 to the ITV camera 1, a contour extraction circuit 10, and a contour point coordinate generation circuit 12, which will be described later.

Reference numeral 10 denotes a contour extraction circuit which detects the contour points of the object image (for example, P ₁ ,
P ₂ ...) is extracted based on the timing signal from the scanning signal generator 11.

As mentioned above, the method for extracting contour points of an object image in this contour extraction circuit 10 is described by the applicant in the specification attached to Japanese Patent Application No. 58-25719. We are proposing an image processing device that performs the processing.

Reference numeral 12 denotes a contour point coordinate generation circuit, which generates two-dimensional coordinates on the imaging plane of the ITV camera 1 of the contour points obtained by the contour extraction circuit 10 based on the timing signal from the scanning signal generator 11.

Reference numeral 13 denotes an address generation circuit which generates an address for storing the contour point coordinates of the contour points extracted by the contour extraction circuit 10 in the bidirectional memory 14.

14 is a bidirectional memory, which is configured using a part of the memory address space of the CPU, and has an internal address and data line switching circuit (not shown);
One side is connected to the contour point coordinate generation circuit 12 and is configured to store contour point coordinates that are sequentially generated.
The other is connected directly to bus line 5 and is connected to the CPU
Direct read/write from 6 is possible.

In other words, the bidirectional memory 14 has the same address system as the main memory 1, and has the same data transfer procedure as the main memory 1, allowing direct access without going through an I/O port, etc., and speeding up recognition processing after storing contour point coordinates. can be done.

With the configuration described above, while one screen of the ITV camera 1 is being scanned, the coordinates of contour points can be sequentially stored in the bidirectional memory based on the output results of the contour extraction circuit 10, and one field of A contour point coordinate table can be created almost immediately after scanning is completed (approximately 16.7ms). Note that after the contour point coordinate table is created in the bidirectional memory 14, the bidirectional memory 14 is directly connected to the bus line 5 by a bus switching circuit (not shown) included in the bidirectional memory 14, so no transfer means is necessary. For example, it is possible to directly read/write from the CPU 6 to the main memory 7 without requiring an I/O port or the like.

As explained above, in the image processing apparatus according to this embodiment, the contour point coordinate table can be created in approximately the same time as the time required for observing an object with an ITV camera. In addition, the contour point coordinate table is created simultaneously with the raster scanning of the object by the ITV camera and stored in the bidirectional memory, so it is necessary to transfer the data to the main memory again to create the contour point coordinate table. Therefore, the processing time can be further shortened.

In this embodiment, a case has been described in which the contour point coordinate table is stored in a bidirectional memory configured using a part of the memory address space of the CPU, but for example, if an I/O port other than the main memory is Even if the contour point coordinate table is stored in a separate memory connected to the bus line via the bus line, and the data is then transferred to the main memory, the processing speed is still sufficient compared to the conventional image processing device shown in Figure 1. The recognition device can be performed at high speed.

(G) Effects of the Invention As explained above, according to the image processing device according to the present invention, a contour point coordinate table is generated from image signals of contour points of an object from image signals obtained by scanning the object with an ITV camera. It can be created at high speed and can be fully applied to visual recognition devices such as robots.

[Brief explanation of drawings]

Fig. 1 is a diagram showing an outline of a conventional visual recognition device, and Fig. 2 shows an image signal obtained by scanning an object.
FIG. 3 is a diagram showing contour points of an object image, FIG. 4 is a diagram showing a contour point coordinate table, and FIG. 5 is a diagram for explaining the image processing apparatus according to the present invention. In the figure, 1 is an ITV camera, 2 is a preprocessing circuit, 5 is a bus line, 6 is a CPU, 7 is a main memory, 8 is a console, 9 is an auxiliary storage device, 10
11 is a scanning signal generator; 12 is a contour point coordinate generation circuit; 13 is an address generation circuit;
14 is a bidirectional memory.

Claims (1)

[Scope of Claims] 1. Obtaining image data by scanning an object with an imaging means, extracting contour points of the object from the image data, and further determining contour point coordinates of the contour points on the imaging surface, An image processing device that performs shape recognition processing based on the contour point coordinates in the processing device, comprising: a scanning signal generating device that generates a timing signal for scanning the object of the imaging device; analog/digital conversion means for converting to obtain digital image data; contour point extraction means for extracting contour points of the object from the digital image data based on the scanning timing signal; a coordinate generating means that generates based on a scanning timing signal; and a coordinate generating means that utilizes a part of the memory address space of the processing device and is directly connected to a bus line so that the processing device can directly access the processing device, and sequentially generates the contour point coordinates. An image processing device comprising: a bidirectional memory for storing information;