JPH087109A - Image division processing method - Google Patents

Abstract

(57) [Summary] [Object] The present invention relates to a method for performing processing for dividing a region of an image developed on a memory to obtain a barycentric position of each region by software, and performing a region dividing process of an image at high speed. The purpose is to provide a method that can be performed. [Structure] Each line of the image memory 10 is sequentially scanned in sequence (step 100), and each pixel data of the scanned line is paired with a pair of line buffers 12_1 and 12_.
Information for identifying each area by dividing the image portions of both line buffers 12_1 and 12_2 into groups into groups in which the result of binarization is the same and the adjacent pixel data is a member which is stored alternately (step 102). And the information indicating the spread of each area are stored (step 104), and when all the lines of the image memory 10 are scanned, adjacent areas are integrated and the barycentric position of each area is obtained from the stored information. (Step 106).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for performing a process of dividing an image developed on a memory into regions and obtaining a barycentric position of each region by software.

When satellite capture work is performed in space,
The mark plate portion on the satellite side is photographed by the camera, and the photographed image is developed on the memory. Then, the image on the memory is divided into regions, and the area and the position of the center of gravity of each region are sequentially calculated, and the distance to the mark plate and the posture of the mark plate are obtained using these calculation results.

[0003]

2. Description of the Related Art An image obtained by photographing a mark plate (for example, four circle marks are drawn in a square arrangement) provided on a satellite to be captured is binarized and developed on a memory (image memory). It

Then, the binarized image developed on the memory is divided into a plurality of areas (in the above case, the binarized image of the mark plate is divided into four circle mark portions), and each divided area is divided into four areas. Numbers are added (labeling process).

Further, the area and barycentric position of each divided area are calculated in order (in the above case, the area and barycentric position of four circle marks are calculated in order), and the calculated result of each divided area is used up to the mark plate. The distance and the posture are required.

Since physical resources are limited in space, it is impossible to perform the above processing by hardware. Therefore, all the processes are sequentially performed on software by using one image memory.

[0007]

However, in the prior art, since all the processing is sequentially performed by software using one image memory, the image memory slower than the normal memory is frequently accessed. Therefore, it becomes difficult to increase the processing speed, which is particularly problematic when capturing slow spin satellites.

The present invention has been made in view of the above conventional circumstances, and an object of the present invention is to provide a method capable of performing image area division processing by software at high speed.

[0009]

In the method according to the present invention in FIG. 1, each line of the image memory 10 is sequentially scanned in sequence (step 100), and each pixel data of the scanned line is paired with a pair of line buffers 12_1. , 12_
Information for identifying each area by dividing the image portions of both line buffers 12_1 and 12_2 into groups into groups in which the result of binarization is the same and the adjacent pixel data is a member which is stored alternately (step 102). And the information indicating the spread of each area are stored (step 104), and when all the lines of the image memory 10 are scanned, adjacent areas are integrated and the barycentric position of each area is obtained from the stored information. (Step 106) is characterized in that

[0010]

In the present invention, the lines of the image memory 10 are sequentially taken out, and the area is divided into two lines (the data of each pixel is binarized when the area is divided).
When all the lines are fetched from the image memory 10, the adjacent divided areas are integrated.

Therefore, the low-speed access to the image memory 10 can be limited to only once for each line extraction, and an extremely high speed other memory (main memory) can be accessed at the time of area division and area integration.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In FIG. 2, a CPU board 20 is connected to an image memory 10 via a bus 22, and a microprocessor unit 23 and a ROM are connected to the CPU board 20.
24 and a main memory 26 are mounted.

The image memory 10 stores, for example, image data obtained by photographing the mark plate of the slow spin satellite, and the microprocessor unit 23 executes the program stored in the ROM 24 using the main memory 26. It

The program to be executed by the microprocessor unit 23 may be transmitted from the ground, stored in the main memory 26, and the program in the main memory 26 may be activated.

In FIG. 3, 512 is stored in the image memory 10.
A size of pixels × 512 pixels is used, and each line (horizontal direction) of the image memory 10 is sequentially scanned in the order of arrangement (vertical direction).

Then, each pixel data of the scanned line is alternately stored in the pair of line buffers 12_1 and 12_2. Therefore, the line buffers 12_1, 1
2_2 having a length of 512 pixels is prepared (in C language, it can be represented by Li1 [512] and Li2 [512]).

A feature quantity storage buffer (1000 in total) is also prepared for calculating the area and the position of the center of gravity of each divided area, and this buffer can be defined by the structure representation in C language as follows.

[0018]

Further, each divided area is identified by a label number, and a connection relation buffer (total 100 pieces) for storing connection information of the divided areas with this label number is prepared, and this buffer is a structure of C language. The expression can be defined as follows.

[0020]

In FIG. 4, the processing procedure of this embodiment is explained by using a flow chart, and the line buffer 12_
1, 12_2, the feature quantity storage buffer, the connection relation buffer, and the variables are initialized in advance (step 400).

Then, each line of the image memory 10 is sequentially scanned in the order of arrangement, and each pixel data of the scanned line is converted into a pair of line buffers 12_ as shown in FIG.
1, 12_2 are alternately stored (steps 402, 4)
06, 408, 412, 414).

Each time the pixel data of the scanned line is stored in the line buffer 12_1 or 12_2, the data of each pixel in the buffer is binarized,
Furthermore, both line buffers 12_1 and 1 are added to a group in which the binarization result is the same and adjacent pixel data are members.
The image portion of 2_2 is divided into areas (labeling), and identification information (label number) of each area and information indicating the spread of each area (area of area, horizontal and vertical address added value) are stored. (Steps 404, 410).

After that, when all the lines of the image memory 10 are scanned (YES in step 414),
The information is obtained by using the information in which the adjacent regions are integrated and the position of the center of gravity of each region is stored (step 416).

In FIGS. 5 and 6, the line buffers 12_1,
12_2 connection relation detection processing (step 404,
The contents of (410) are explained using a flow chart, and the operation thereof is explained in FIG. 7.

The line buffer 12 shown in FIGS.
_1, and in the case of the line buffer 12_2, Li1 [i] and Li2 [i] are exchanged. In FIG. 5, the variable i indicating the position of the line buffer scan is incremented from the value 1 to the value 512 (steps 500, 606, 608).

During this scan, the line buffer value Li1
[I] (pixel value) is the threshold value for the binarization.
If it is confirmed that the number of hold-bels has been exceeded (YES in step 502), both line buffers 12_1, 1
It is checked in the order of FIG. 7 whether or not there are adjacent pixels in the image portion formed by 2_2 that have the same binarization result (steps 504, 512, 5).
16,600).

At this time, when it is confirmed that adjacent pixels have the same binarization result (the current pixel belongs to the same region as the adjacent pixel: both are connected). , The value of the adjacent pixel is substituted for the value of the pixel of interest at present, the number of pixels number indicating the area area of the feature amount storage buffer Lbtable is incremented, and the X address added value samX and the Y address added value samY are updated. (Steps 506, 514, 518, 602).

However, the pixel of interest (line buffer value Li
1 [i]) and line buffer values Li1 [i-1], Li
When the combination with the adjacent pixel of 2 [i-1] (the first and third pixels in FIG. 7) is confirmed, the adjacent pixel of the line buffer value Li2 [i + 1] (the fourth pixel in FIG. 7 is obtained. When the combination with the pixel is also confirmed (YES in steps 508 and 520), the line buffer value Li1
The pixel of interest of [i] and the line buffer value Li2 [i + 1]
The combination of the areas each including the adjacent pixel of is registered in the connection relation buffer ct (steps 510 and 522).

The line buffer value Li1 [i-
1], Li2 [i], Li2 [i-1], Li2 [i +
1] is confirmed not to be combined with each adjacent pixel (NO in step 600), the label of the next area is prepared (step 604).

In FIG. 8, the connection relation integration processing (step 41
The contents of 6) are explained using a flow chart,
First, a structure array Lbinteg identical to the feature quantity storage buffer is prepared (step 800).

Further, the variable i is the number of divided areas (the number of labels)
Until (steps 802 and 820) are reached, the divided areas in which the connection relation is indicated in the connection relation buffer ct are integrated to update the flag ConnectTo, and the number of pixels in each divided area number
(Indicates the area of the divided area), X address added value samX, Y
The process of recalculating the address addition value samY is a structure array Lbin.
performed using teg (steps 804, 806, 8
08,810,812,814,816,818).

Then, the feature amount buffer of the divided areas that are not combined with the integrated divided areas from the value of the flag ConnectTo
The Lbtable is selected (step 822) and the barycentric position of each divided area is calculated by dividing the X address added value samX and the Y address added value samY by the number of pixels number (indicating the area of the divided area) in these buffers Lbtable. (Step 824).

As described above, the images obtained by photographing with a camera or the like and written in one image memory 10 are sequentially read into the buffer for two lines one by one and binarized, and the image parts for the two lines are read. A divided area in which pixels that are adjacent to each other and have the same binarization result as group members are defined, and adjacent areas are integrated when all lines are read out, integrated area and unintegrated area The process of calculating the position of the center of gravity of all the independent regions is performed by software.

Therefore, according to this embodiment, the low-speed image memory 10 is accessed only once. Therefore, the speed of software processing for calculating the position of the center of gravity of the area is significantly increased. As a result, even when the slow spin satellite is captured, the distance to the mark plate of the satellite and the attitude of the mark plate can be obtained at a sufficient speed.

In order to reduce noise on the photographed image and obtain a more accurate result (distance to the mark plate and posture of the mark plate), it is preferable to prepare pixel histograms in the horizontal and vertical directions. Is.

In this case, as a feature quantity storage buffer The structure array of Lbtable [Li1 [i-1]]. HistX [i] + = 1; Lbtable [Li1 [i-1]]. HistY [Y] + = 1; Are added to steps 506 and 510.

Further, a square window having (XUP, YUP) in FIG. 9 as the upper left corner and (Xdown, Ydown) as the lower right corner is set on the photographed image, and the processing is limited to that window, thereby achieving higher speed. It is possible to perform various image division processes, and the robot hand can stably follow the object to be grasped (for example, slow spin satellite) at high speed.

In this case, Y = 1 in step 400 of FIG. 4 is set to Y = YUP, and Y <Y in step 414.
512 is set to Y <YDOWN, and the range of the variable i is set from XUP to XDOWN in the processing of detecting the connection relation.

[0040]

As described above, according to the present invention, the lines of the image memory are sequentially fetched and alternately written into the buffer for two lines, and the region division is performed by using the two lines as a unit. Since the adjacent divided areas are integrated when each line is fetched from the image memory, the slow image memory is accessed only once for each line fetch.

Therefore, an extremely high-speed memory (main memory etc.) can be accessed at the time of area division and area integration. Therefore, even when the slow spin satellite is captured, the distance to the mark plate of the satellite and the mark plate The posture can be calculated at a sufficient speed.

[Brief description of drawings]

FIG. 1 is a diagram illustrating the principle of the invention.

FIG. 2 is an explanatory diagram of a configuration of an embodiment.

FIG. 3 is an explanatory diagram of a configuration of an image memory and a line buffer.

FIG. 4 is a flowchart illustrating a schematic processing procedure of an embodiment.

FIG. 5 is a flowchart showing the processing contents of connection relation detection.

FIG. 6 is a flowchart showing the processing contents of connection relation detection.

FIG. 7 is an explanatory diagram of an operation of detecting a connection relationship.

FIG. 8 is a flowchart showing the contents of a combined relation integration process.

FIG. 9 is an explanatory diagram of an operation of window setting.

[Explanation of symbols]

 10 image memory 20 CPU board 22 bus 23 microprocessor unit 24 ROM 26 main memory

Claims (1)

[Claims] 1. Each line of an image memory (10) is sequentially scanned in sequence (step 100), and each pixel data of the scanned line is alternately stored in a pair of line buffers (12_1, 12_2) (step 102). ), Both line buffers (12_1 and 12_) are grouped into a group in which the binarized result is the same and adjacent pixel data is a member.
The image portion of 2) is divided into areas, information for identifying each area and information indicating the spread of each area are stored (step 104), and when all lines of the image memory (10) are scanned. An image segmentation processing method characterized in that adjacent regions are integrated to obtain the barycentric position of each region from the stored information (step 106).