JPH02217974A - Image recognizing device - Google Patents

Abstract

PURPOSE:To speed up the recognition processing of an image to be picked-up by processing the image signal of the image to be picked-up and calculating the peripheral length of the photographed object on an image. CONSTITUTION:The subject device has a telecamera (image pickup means) 1, a synchronizing separating circuit 2, a binarizing circuit 3, one-horizontal line delaying circuits 4, 5, and shift registers 6 to 8. The peripheral length components from respective directions, i.e. vertical, horizontal and oblique directions of the image to be picked-up are found out by the operation of each horizontal scanning of an image signal, the sum of respective horizontal lines is calculated over the effective vertical period and the peripheral length is calculated in the succeeding vertical blanking period. Consequently, the image pick-up signal of the image to be picked-up can be usefully and fully utilized and detected at real time and the recognition processing of the image to be picked-up can be accelerated.

Description

Detailed Description of the Invention (Field of the Invention) This invention processes an image signal of an object obtained from an imaging device such as a television or a video camera, and calculates and calculates the circumference of the object on the image. The present invention relates to an image recognition device.

(Summary of the Invention) The present invention includes an imaging means for imaging an object and outputting an image signal representing a grayscale image, and a binary image formed by a plurality of vertical and horizontal pixels by binarizing the output image signal. A binary processing means for generating the object and an arithmetic processing means for calculating the circumference of the bipedal object based on the binary data constituting each pixel are provided, and the object's image signal is fully utilized without waste. The perimeter is detected in real time.

(Prior art and its problems) Conventionally, an imaging device such as a television or a video camera is used to capture an object, and an image signal (
Most of the methods of processing a video signal to obtain the circumference of an object are as follows. That is,
Conventionally, the image signal of the object to be imaged is binarized and stored in an image memory, and then the image data stored in the image memory is processed by software to obtain the image signal of the object to be imaged. The peripheral length of the imaged object is determined by performing contour tracking and the like.

However, in this conventional method, the binarized image data is temporarily stored in the image memory, and then the data is transferred to the
Since software processing is performed by the PU, image memory for at least one frame is required. Since it is necessary to store data in an image memory and to read and calculate each pixel, the processing time becomes long, and there is a problem that real-time performance is impaired when performing real-time detection.

(Object of the Invention) This invention was made in view of the above-mentioned problems, and its purpose is to calculate the circumference of an imaged object at high speed with a simple configuration and without using an image memory. An object of the present invention is to provide an image recognition device that can perform processing, shorten processing time, and enable real-time detection.

(Structure and Effects of the Invention) In order to achieve the object stated in item 1, the present invention provides: an imaging means for imaging an object and outputting an image signal representing a grayscale image; a binary processing means for generating a binary image consisting of a plurality of vertical and horizontal pixels by performing a value processing; and a portion of the image signal based on the binary data constituting each pixel of the binary image generated by the binary processing means. It is characterized by comprising an arithmetic processing means comprising an extraction means for extracting a region and an integration means for integrating the circumference of the L object from each direction from each value of the partial region extracted by the extraction means. do.

According to the image recognition device of the present invention, which has a two-legged configuration, for each horizontal operation of the image signal, the arithmetic processing means calculates the vertical, horizontal, diagonal, left side, right side, etc. of the imaged object. The image signal of the object is obtained by calculating the perimeter component of the object from each direction, multiplying the sum of each horizontal line by the effective vertical period, and calculating the perimeter in the next vertical period. This means that real-time detection can be carried out by making full use of the system without waste, and when applied to robots, etc., the recognition processing of the imaged object can be sped up.

In addition, since it can be realized with a simple configuration that does not use software or image memory, it has many effects such as lowering costs overall, simplifying the circuit configuration, and increasing the speed of recognition processing, making the image recognition device more versatile. .

(Embodiment) Hereinafter, an embodiment of the image recognition device of the present invention will be described based on the drawings.

FIG. 1 is a block diagram showing the entire 1"t% of the image recognition apparatus of the present invention.

In the figure, 1 is a television camera which is an imaging means, and this television camera 1 is a stationary or moving object to be photographed (
(not shown) and outputs a video signal (image signal) representing a grayscale image of the object to be imaged.

2 and 3 in the figure are a synchronization separation circuit and a binarization circuit used as a binary processing stage, which binarize the video signal output from the television camera 1 and create a binary image consisting of a plurality of vertical and horizontal pixels. It is designed to generate.

4 to 8 in the figure are horizontal line delay circuits 4 and 5 and shift registers 6 and 7, respectively, which are manually used as extraction means. 8
It is. This extracting means is configured to extract a partial region of the video signal based on binary data constituting each pixel of a binary image generated by the binary processing means described below.

Reference numerals 9 to 19 in the figure are integration means for integrating the circumference of the object from each direction from various parts of the partial region extracted by the extraction means, and the perimeter detection circuit 9 counters 10 to 13, 3 respectively.
State buffers 14 to 17, decoder 18. CPU1
It consists of 9.

That is, the arithmetic processing means includes the above-mentioned extraction means and integration means.

Further, the configuration and operation of the image recognition device of the present invention will be explained in detail based on the drawings.

The television camera 1 is for capturing an image of a stationary or moving object, and outputs a video signal representing a grayscale image of the object. This video signal is given to a sync separation circuit 2, which separates a horizontal sync signal HD and a vertical sync signal VD from the video signal, generates a sampling clock signal ck, and converts the video signal into a binarization circuit. Give to 3. The Domei signal HD and vertical synchronization signal VD are as shown in Figure 5.In other words, the binarization circuit converts the video signal into black and white binarized for each field based on a predetermined threshold level. The current signal B3 of this binary signal is manually input to the shift register 6, and this binary signal is delayed by 1 horizontal line delay circuit 4 for an operation period of 17. The signal B2 inputted into the shift register 7, and the signal B3 obtained by delaying this binarized signal by 2 water operation periods by the 1 water 11 line delay circuits 4 and 5 are manually input into the shift register 8. , furthermore, the binarized signal currently being manually input from the television camera 1, the 3×3 8 neighborhood patterns a to i for the pixels approximately one water operator operation period ago are obtained.
is obtained.

This 8-neighborhood pattern is manually input to the perimeter detection circuit 9. Details of this circuit are shown in FIG.

In the perimeter detecting circuit 9 of FIG. 4, the output E1 is:
When the peripheral branch component in the horizontal direction is detected, the signal becomes H level. The output E2 becomes H level when a peripheral branch component in the vertical direction is detected. The output E3 becomes H level when the peripheral branch component in the diagonal direction of the right 1-lily is detected. Also output E4
becomes H level when a peripheral branch component in a diagonal direction to the left is detected.

These E1-E4 signals are then provided to enable counters 10, 11, 12 and 13, respectively, shown in FIG. These counters] 0, 11, 12 and 13 are all cleared by the inversion VD of the vertical synchronization signal VD,
As the count clock, an inverted version of the sampling clock ck is given. As a result, the outputs L1 to L4 of each counter become the total number of peripheral branch components in one field (one screen operation). That is, the output L1 is the total number of peripheral edge components in the horizontal direction, the output 2 is the total number of peripheral edge components in the vertical direction, the output 3 is the total number of peripheral edge components in the right-L direction, and the output 4 is the total number of peripheral edge components in the diagonal direction. is the total number of peripheral branch components in the diagonal direction to the right.

FIG. 2 is a diagram of a binary image for explaining how the number of peripheral branch components of the object to be imaged is integrated, and FIG. 3 is a time chart showing the process.

In FIG. 2, the dashed line indicates a black pixel where the object is located.

Now, assume that the peripheral branch components are detected only after the center of the 8-neighborhood pattern becomes X=n-1, Y=m.

At this time, since peripheral branch components in the horizontal direction and in the diagonal direction upward to the right are present in the 8-neighborhood pattern, the enable E1 of the counter 10 and the enable E3 of the counter 13 become H level. Then, at the falling edge of the sampling clock ck, the outputs of the counter 10 and the counter 12 both become 1.

Next, when the center of the 8-neighborhood pattern is X=n, Y=m, only the peripheral branch component in the diagonal direction upward to the right exists, so
The enable E4 of the counter 13 becomes 1. Then, the output of the counter 4 becomes 1 at the rising edge of the sampling clock ck. In this way, during one field (effective vertical operation period), each counter 10-13 obtains the total number of each peripheral branch component.

Then, when the line enters the blanking intercostal space, the CPU 19 shown in FIG. Read the values of L3 and L4. After doing so, the peripheral branch component in the diagonal h direction is multiplied by 77, that is, the following formula (%) is calculated to obtain the peripheral length of the object.

Note that the perimeter detection circuit 9 in FIG. 4 is just an example, and any circuit may be used as long as it can detect peripheral branch components.
It goes without saying that the numbers from 0 to 12 may vary.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the image recognition device of the present invention, FIG. 2 is a screen configuration diagram showing the imaging state of an object to be imaged,
FIG. 3 is a chart showing the imaging state of the object to be imaged, FIG. 4 is a circuit diagram showing an embodiment of the perimeter detection circuit, and FIG. 5 shows the relationship between the horizontal synchronization signal and vertical synchronization signal of the present invention. FIG. 1... Television camera (imaging means) 2... Synchronization separation circuit 3... Binarization circuit 4.5... 1 horizontal line delay circuit 6, 7.8... Shift register 9... Perimeter length detection circuits 10.11, 12.13...counters 14, 15.
1.6. 17...3 state buffer 18...
One decoder...CPU

Claims (1)

[Scope of Claims] 1. Imaging means for capturing an image of an object and outputting an image signal representing a grayscale image; and binary processing for binarizing the image signal output from the imaging means and comprising a plurality of vertical and horizontal pixels. a binary processing means for generating an image; an extraction means for extracting a partial region of the image signal based on binary data constituting each pixel of the binary image generated by the binary processing means; An image recognition device comprising: an arithmetic processing means including an integration means for integrating the circumference of the object from each direction from each value of the partial region.