JPH0962845A - Labeling method for picture data - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the labeling method of picture data by which a labeling processing can inexpensively be executed at high speed with a simple method. SOLUTION: A process (S2) for A/D-converting a picture signal inputted as an analog into a digital signal in the same way as a conventional system, a run coding process (S3) executing the binarization processing to the digital signal and expressing respective lines with coordinates of the start point and the end point of continuous picture element strings having specified gradation values, a block coding process (S4) evaluating the continuity of respective run code which are run-coded and generating blocks and a labeling process (S5) evaluating the continuity of the blocks are provided. Inputted camera video data is recognized (S6) as one continuous lump (area). Here, the run code is obtained by expressing the start point and the end point of a horizontal designation color by the picture element coordinates.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a labeling method of image data in a labeling process performed when image data such as a camera image is taken into a computer.

[0002]

2. Description of the Related Art Normally, when a camera video signal is input to a computer such as a personal computer, as shown in FIG.
The D / D converted (S102) digital signal is written as a digital signal in a digital memory corresponding to the physical position on the screen (S103), and the labeling process (S104) is performed by the software built in the computer. , The input camera image data was recognized as one continuous block.

[0003]

However, according to the conventional method described above, not only a large-capacity digital memory is required but also a large amount of digital memory information (see FIG. 6).
In the example shown in (1), a lot of time is required for the labeling process to handle 256 kilobytes. Therefore, in order to realize image processing in real time (real time), the cost is unnecessarily high and the feasibility is poor. The present invention has been made in order to eliminate the drawbacks in the prior art, and provides a labeling method of image data which enables a labeling process at a low cost and a high speed by a simple method. It is an issue.

[0004]

SUMMARY OF THE INVENTION According to the present invention, an image signal input as an analog signal from a camera is horizontally A / D converted.
A rancode forming step of expressing the start point and the end point in a continuous pixel row having a specific gradation value in each line by image data converted by the conversion by the pixel coordinates in each line; A block code forming step for evaluating the continuity of the run code created by the conversion and the run code in the line next to the line by the calculation processing of the run code, and the block code forming step. To evaluate the continuity between blocks that are evaluated as continuous, the final run
And a labeling step for evaluating the continuity between the leading runcode in the block and all the runcodes in the preceding line while evaluating the continuity between the runcode and all runcodes in the next line. The above-mentioned problems have been solved by a labeling method of image data characterized by the above.

[0005]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an image data according to the present invention.
FIG. 3 is a process chart showing a labeling method of a digital signal, which is a process of A / D converting an image signal input as an analog signal into a digital signal in the same manner as a conventional system (S2), and binarizing the digital signal. Then, a run-code forming step (S3) of expressing the start point and end point of a continuous pixel row having a specific gradation value in each line by pixel coordinates, and further,
The block coding process (S4) for evaluating the continuity of each run code to create a block, and the labeling process (S5) for evaluating the continuity of the blocks, The input camera image data is recognized as one continuous block (area) (S6). The run code referred to here is the one in which the start point and the end point of the designated color in the horizontal direction are expressed in pixel coordinates.

Since the binarized image data is composed of a combination of two gradation values, the start point and the end point of a continuous pixel row having a specific gradation value are represented by pixel coordinates ( It becomes possible to compress the image data as well as the run code conversion. That is, the camera image example (U-shaped is shown) shown in FIG. 1 will be described.
In the conventional method, 512 bytes of memory is required to represent the information on the line A ′, but in the method according to the present invention, two start points 1 and 2 and two end points 3 and 4 are used.
It can be seen that only four coordinate representation data consisting of

FIG. 2 is a block diagram showing a hardware configuration for executing in real time the process of binarizing the A / D-converted digital signal and converting it into a run code. The signal binarized in an LUT (Look Up Table) 11 is operated by inputting a sampling clock to a start point coordinate REG18 and an end point coordinate REG19 via a noise removal circuit 12 and a change point detection circuit 13. Output signal from horizontal counter 15 and runco
It is input together with the output signal from the read column counter 17 and then written in the run code memories 20 and 21 to be run coded.

FIG. 3 shows a runner in the method according to the present invention.
This is an example of an image (image in memory), and corresponds to the illustrated binarized image. There, the first (start point, end point) of runcode column number 0 is (200, 203)
And the vertical line numbers 100 to 200 to 203
It indicates that there are continuous pixels. In addition, the vertical line number 103 includes (2
01,204) and 2 of the runcodes shown in (211,213) of the runcode sequence number 1.
Indicates that there is one. It is the block coding process that evaluates the simple continuity in the downward direction (+ Y direction) of each of these run codes.

The continuity of two lancodes means that both lancodes contain the same X coordinate point. Therefore, the continuity can be evaluated by the following simple calculation. That is, (end point coordinate)-(next line start point coordinate) and (next line end point coordinate)-(start point coordinate)
Is positive when both are 0, it is evaluated as continuous. Referring to the example shown in FIG. 3, in the case of the run code 25 and run code 26, (end point coordinate)-(next line start point coordinate) is 203-201 = 2 and is positive, (next line end point coordinate) Since-(starting point coordinates) is positive at 203-201 = 2, the run code 25 and run code 26 are evaluated as having continuity.
However, regarding Run Code 26 and Run Code 27,
(End point coordinate)-(Next line start point coordinate) is 203-211
= Negative at -8, (Next line end point coordinate)-(Start point coordinate) is 2
Since 13-201 = 12 is positive, the run code 26 and run code 27 are evaluated as non-continuous.

At the same time as performing the continuity evaluation for each run code, the attribute evaluation of the block start point and the block end point is performed for each run code group (block) having continuity. The block start point is not continuous with the upward direction (-Y direction) at the same run code sequence number (marked with (*) in FIG. 3), and the block end point is down at the same run code sequence number. It is not continuous with the direction (+ Y direction) ((:) is attached in FIG. 3). Further, in this step, block attribute data such as a block number, a start point coordinate, an end point coordinate, a block total pixel number, a block maximum coordinate (X, Y), a block minimum coordinate (X, Y), etc. is further provided for each block. Created.

Then, labeling is performed. Labeling here means to evaluate the block code attribute and label blocks with continuity as areas. In labeling, all runcodes whose line numbers are incremented by 1 in the Y direction are targeted, and continuity with all runcodes of the next line (Y coordinate) at each end point is evaluated. That is, the end point 8 in the first block in FIG. 4 is evaluated as a complete end point because there is no run code in the next line, while the end point 10 in the second block is continuous with the first block in the next line. It is evaluated as the end point of the group because of its nature. As a result, the first block and the second block are continuous,
That is, it is recognized as one area.

Similarly, for each block, the continuity between the head runcode and all runcodes in the preceding line is evaluated to see if it is a group start point or a complete start point. Thus, when both the group start point and the group end point are satisfied, the block is recognized as a nuclear area, and otherwise, the block is merged into the nuclear area (FIG. 5). This process is performed for all blocks, and finally the total area area (total number of pixels),
Data such as maximum and minimum XY coordinates of the area and center coordinates of the area are created.

[0013]

According to the image data labeling method of the present invention, the number of memories handled by software is overwhelmingly reduced, image processing can be performed at high speed and low cost, and the algorithm is simple. Therefore, when using a high-speed frame camera or high-resolution camera that will become popular in the future, it will be cheaper than the current camera and the processing speed per pixel will be improved. , Its usefulness will increase in the future.

[Brief description of drawings]

FIG. 1 is a process chart showing a labeling method of image data according to the present invention.

FIG. 2 is a block diagram showing the configuration of hardware for lancoding according to the present invention.

FIG. 3 is an illustration of a rancode in a labeling method of image data according to the present invention.

FIG. 4 is an image diagram of the internal structure of the memory in the image data labeling method according to the present invention.

FIG. 5 is a flowchart showing a labeling process in the labeling method for image data according to the present invention.

FIG. 6 is a process diagram showing a conventional labeling method for image data.

Claims (1)

[Claims] 1. A start point and an end point in a continuous pixel row having a specific gradation value in each line of image data obtained by converting an image signal input as an analog signal from a camera into a digital signal by A / D conversion in the horizontal direction. Is represented by pixel coordinates in each line, a runcode created by the runcode, and the continuity of the runcode in the line next to the line is determined by the runcode. A block code conversion step to be evaluated by arithmetic processing, and a final run code of the block and the next line in order to evaluate the continuity between the blocks evaluated to be continuous in the block code conversion step. The continuity with all the run codes in the block is evaluated, and the run code in the block and all run codes in the preceding line are evaluated.
And a labeling process for evaluating the continuity with the image data.