JPH1079035A - Specific object extracting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a specific object extracting device speedily processing the centroid calculation of a specific pattern scattered in a picture. SOLUTION: The device is provided with a streak extraction part 101 for investigating the series of binary picture signals inputted by means of a raster scanning and extracting a streak with the series of not less than a fixed value, and an adjacency detection part 102 investigating contact between the extracted streak and a pattern extracted at a just preceding line. Then after extracting the streak, vertical adjacency is judged by the unit of a line, the specific object is defined when noticing line data adjacent to the pattern obtained at the just preceding line no longer exists and the coordinate adding value corresponding to the pattern and the number of the pixels of the pattern are defined to speedily execute the centroid calculation of the specific pattern.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly to a specific object extracting apparatus for extracting a region of a specific object (specific pattern) and a center of gravity scattered in an image.

[0002]

2. Description of the Related Art Conventionally, it has been known that a specific object scattered in an input image is separated from other backgrounds / areas and the center of the pattern is extracted as a method for recognizing the position of the specific object. Have been.

The extraction of the center of gravity of this pattern is shown in FIG.
It is generally realized by a processing configuration as shown in FIG. That is, provisional labeling is performed on the input image, and then actual labeling is performed based on the provisional label.
This is generally performed by four steps of processing of obtaining the coordinate value of the circumscribed rectangle of each pattern and further extracting the center of gravity value of the pattern in the area of the circumscribed rectangle.

More specifically, in the provisional labeling process performed in the primary labeling step 601 shown in FIG. 13, binary image data stored in a frame memory is sequentially read by a raster scan method, as shown in FIG. As shown by the primary labeling result 702, the input image binary data 701 is subjected to provisional labeling for each pixel. That is, for example, when a black pixel is a pixel where data exists,
The place where black is first found is assigned a provisional label “1”, and those that are in contact in the horizontal direction are given the label “1”. And where the next black was found,
A temporary label “2” is attached. However, in this case, if it is in contact with the upper run, the same temporary label as the upper run is attached.

When this provisional labeling process is performed for one screen, as shown in the primary labeling result 702, various label values are added to one touching pattern.
Therefore, in the label integration step 602 and the secondary labeling step 603, the same label value is used as the real label if the provisional labels are assigned even though they are in contact with each other. Label integration processing and secondary labeling processing as shown by the secondary labeling result 703 to be attached are performed.

After the correct label is attached in this way, in an area setting processing step 604, an area setting processing for obtaining a circumscribed rectangle for each pattern is performed based on the label value. Further, the image in the obtained circumscribed quadrangular region is rescanned, a weighting process and an area calculation are performed on the black or white pixels existing in the region, and a center of gravity calculating step 605 for obtaining the center of gravity of the pattern from the value is performed. Is performed.

[0007]

By the way, in the conventional specific pattern center-of-gravity extraction processing technology, the provisional labeling process and the label updating process are processes of one frame of each pixel. It takes two frame periods to output what is called. Next, the label image stored in the frame memory or the like is read out, a process of setting a circumscribed rectangle for the label value of each pixel is performed, and the center of gravity of the extracted circumscribed rectangle is further calculated for another two frame periods. Is required, and the barycentric value of the specific pattern finally obtained is also output with a delay of 4 frames or more. Thus, if there is a delay of 4 frames or more before the circumscribed rectangle is obtained,
In order to separate the specific object from the corresponding image using the circumscribed rectangle, a memory or the like for storing a plurality of images is required so that the original image can be held for at least four frames. Become. Furthermore, in the conventional method of calculating the center of gravity, since the adjacent object is examined in pixel units, if the specific object is divided or deformed due to scratches or dirt, the circumscribed rectangle of the specific object is extracted. There is a problem that can not be.

SUMMARY OF THE INVENTION The present invention has been made in order to solve the above-mentioned problems in the conventional centroid extracting apparatus for a specific pattern, and it is an object of the present invention to provide a specific object extracting apparatus which is faster and more resistant to noise. .

[0009]

According to a first aspect of the present invention, there is provided an image processing apparatus, comprising: a method for extracting a specific object defined in a predetermined shape as an object to be extracted from an image including the specific object; In the image processing apparatus having a specific object area setting means for setting the area of the specific object in the image, and a specific object gravity center calculating means for extracting the center of gravity value of the specific object, The processing of the specific object area setting means and the specific object gravity center calculating means is performed in parallel, and the specific object area and the center of gravity value are simultaneously extracted. Accordingly, it is possible to realize a specific object extracting device capable of processing the calculation of the center of gravity of the specific object scattered in the image at high speed.

According to a second aspect of the present invention, there is provided a specific object extracting means for extracting, from an image including the specific object, a specific object defined in a predetermined shape as an object to be extracted, A circumscribed rectangle extracting means for setting a circumscribed rectangle of the specific object extracted by the object extracting means, a specific object coordinate value calculating means for calculating a sum of coordinate values of respective pixels of the specific object, and the specific object coordinate A center-of-gravity extracting means for extracting the center-of-gravity coordinates of the specific object based on the result of the value calculating means, wherein the specific object extracting means comprises a run length of the input binary image data in the raster direction. A streak extracting means for extracting a binarized data string having a run length within the predetermined range as a streak constituting a part of the specific object only when is within a predetermined range; Adjacency detecting means for detecting the presence or absence of an adjacency between a streak extracted by the output means and a circumscribed rectangle based on a streak aggregate formed earlier than this, and the number of pixels of the circumscribed rectangle and the streak aggregate and Attribute storing means for storing the sum of the coordinate values of the pixels in the streak aggregate, wherein the specific object coordinate value calculating means calculates the sum of the coordinate values of each pixel of the streak. Calculating value updating means for updating the calculation result obtained by the coordinate value calculating means; effective pixel counting means for counting the effective pixels in the streak; and updating the number of effective pixels obtained by the effective pixel counting means. Effective pixel number updating means for generating a circumscribed rectangle including the streak in accordance with an adjacent result of the adjacent detection means, wherein the coordinate value calculating means and the effective pixel counting means The sum and the count value of the pixel of the coordinate values with the streak that contains the meta said circumscribed rectangle, is to configured to update by the effective pixel number updating means and the calculated value updating means. This allows
The calculation of the center of gravity of a specific object scattered in an image can be performed at high speed with simple processing, and the cost of a system incorporating the specific object extracting apparatus can be reduced.

According to a third aspect of the present invention, there is provided a specific object extracting means for extracting, from an image including the specific object, a specific object defined in a predetermined shape as an object to be extracted, A circumscribed rectangle extracting means for setting a circumscribed rectangle of the specific object extracted by the object extracting means, a specific object coordinate value calculating means for calculating a sum of coordinate values of respective pixels of the specific object, and the specific object coordinate A center-of-gravity extracting means for extracting the center-of-gravity coordinates of the specific object based on the result of the value calculating means, wherein the specific object extracting means comprises a run length of the input binary image data in the raster direction. A streak extracting means for extracting a binarized data string having a run length within the predetermined range as a streak constituting a part of the specific object only when is within a predetermined range; Adjacency detecting means for detecting the presence or absence of an adjacency between a streak extracted by the output means and a circumscribed rectangle based on a streak aggregate formed earlier than this, and the number of pixels of the circumscribed rectangle and the streak aggregate and Attribute storing means for storing the sum of the coordinate values of the pixels in the streak aggregate, wherein the specific object coordinate value calculating means calculates the sum of the coordinate values of each pixel of the streak. An operation updating unit for updating the operation result obtained by the coordinate value operation unit; an effective pixel counting unit for counting the effective pixels in the streak; and updating the number of effective pixels obtained by the effective pixel counting unit. Means for updating the number of effective pixels, wherein the adjacency detecting means detects the streak of the line of interest in contact with the circumscribed rectangle obtained in the immediately preceding line when the streak disappears. Deemed to confirm the square pattern, the center of gravity extraction means is configured to perform the centroid calculation streak aggregate contained in the outer contact rectangle immediately after the circumscribed rectangle has been determined. With such a configuration, it is possible to realize a high-speed specific object extracting device that can calculate the center of gravity of the specific object scattered in the image immediately after the specific object is determined.

According to a fourth aspect of the present invention, there is provided a specific object extracting means for extracting, from an image including the specific object, a specific object defined in a predetermined shape as an object to be extracted, A circumscribed rectangle extracting means for setting a circumscribed rectangle of the specific object extracted by the object extracting means, a specific object coordinate value calculating means for calculating a sum of coordinate values of respective pixels of the specific object, and the specific object coordinate A center-of-gravity extracting means for extracting the center-of-gravity coordinates of the specific object based on the result of the value calculating means, wherein the specific object extracting means comprises a run length of the input binary image data in the raster direction. A streak extracting means for extracting a binarized data string having a run length within the predetermined range as a streak constituting a part of the specific object only when is within a predetermined range; A neighbor detecting means for detecting the adjacent and circumscribing rectangle based on streak aggregate formed in advance than this and extracted streak by means output,
Attribute storage means for storing the number of pixels of the circumscribed rectangle and the streak aggregate and the sum of the coordinate values of the pixels in the streak aggregate; and a fixed pattern count for counting the number of circumscribed rectangles determined for each line Means, wherein the specific object coordinate value calculating means comprises: a coordinate value calculating means for calculating a sum of coordinate values of each pixel of the streak; and a calculation value updating means for updating a calculation result obtained by the coordinate value calculating means. Means, an effective pixel counting means for counting the effective pixels in the streak, and an effective pixel number updating means for updating the number of effective pixels obtained by the effective pixel counting means, the adjacent detection means,
When the streak of the line of interest in contact with the circumscribed rectangle determined in the immediately preceding line is eliminated, the circumscribed rectangle determined in the immediately preceding line is regarded as a determined pattern. It is configured to calculate the center of gravity of the streak aggregate for the counted value.

With this configuration, the center of gravity of the specific object scattered in the image is calculated during the horizontal blanking period immediately after the pattern is determined, so that the number of interrupts can be suppressed when outputting to a programmable IC. it can. Further, by calculating the center of gravity value during the horizontal blanking period, the arithmetic unit for processing the video signal and the arithmetic unit for calculating the center of gravity value can be shared, and the scale of the apparatus can be reduced.

According to a fifth aspect of the present invention, there is provided a specific object extracting means for extracting, from an image including the specific object, a specific object defined in a predetermined shape as an object to be extracted, A circumscribed rectangle extracting means for setting a circumscribed rectangle of the specific object extracted by the object extracting means, a specific object coordinate value calculating means for calculating a sum of coordinate values of respective pixels of the specific object, and the specific object coordinate A center-of-gravity extracting means for extracting the center-of-gravity coordinates of the specific object based on the result of the value calculating means, wherein the specific object extracting means comprises a run length of the input binary image data in the raster direction. A streak extracting means for extracting a binarized data string having a run length within the predetermined range as a streak constituting a part of the specific object only when is within a predetermined range; A neighbor detecting means for detecting the adjacent and circumscribing rectangle based on streak aggregate formed in advance than this and extracted streak by means output,
Attribute storage means for storing the number of pixels of the circumscribed rectangle and the streak aggregate and the sum of the coordinate values of the pixels in the streak aggregate; and a fixed pattern count for counting the number of circumscribed rectangles determined for each frame Means, wherein the specific object coordinate value calculating means comprises: a coordinate value calculating means for calculating a sum of coordinate values of each pixel of the streak; and a calculation value updating means for updating a calculation result obtained by the coordinate value calculating means. Means, an effective pixel counting means for counting the effective pixels in the streak, and an effective pixel number updating means for updating the number of effective pixels obtained by the effective pixel counting means, the adjacent detection means,
When the streak of the line of interest in contact with the circumscribed rectangle obtained in the immediately preceding line disappears, the circumscribed rectangle obtained in the immediately preceding line is regarded as a determined pattern, and the center of gravity extracting means performs the vertical blanking period by the confirmed pattern counting means in the vertical blanking period. It is configured to calculate the center of gravity of the streak aggregate for the counted value.

With this configuration, the center of gravity of the specific object scattered in the image is calculated in the vertical blanking period immediately after the pattern is determined, so that the number of interrupts can be further increased when outputting to a programmable IC. Can be suppressed. Further, by calculating the barycenter value during the vertical blanking period, the calculator for processing the video signal and the calculator for calculating the barycenter value can be shared, and the scale of the device can be reduced.

[0016]

Next, an embodiment will be described. First, a basic embodiment of the specific object extracting apparatus according to the present invention will be described based on FIG. 1 showing the concept of the present invention. In FIG. 1, reference numeral 101 denotes the horizontal run length of a binary image signal input by raster scanning, and data having a run length to be processed (called streak).
A streak extraction unit for extracting the streak, an adjacent detection unit 102 for investigating a contact between the streak and the pattern extracted in the previous line, a coordinate value addition unit 103 for adding a coordinate value during a period in which the streak exists, 104 Is an effective pixel counting unit for counting pixels in a period in which the streak exists, 105
Is a circumscribed rectangle extracting unit that generates a circumscribed rectangle including a streak of the line of interest according to the result of the adjacent detection unit 102;
6 is a circumscribed rectangle extraction unit 10 according to the result of the adjacent detection unit 102.
A coordinate addition value updating unit 107 for generating a coordinate addition value corresponding to the circumscribed rectangle extracted in step 5, and a value 107 of the effective pixel counting unit 104 according to the result of the adjacent detection unit 102.
The effective pixel number updating unit 108 updates the number of effective pixels corresponding to the circumscribed rectangle extracted in step 108. 108 stores the results of the circumscribed rectangle extraction unit 105, the coordinate addition value update unit 106, and the effective pixel number update unit 107 in line units as attributes. , An attribute storage unit for reading.

In the specific object extracting apparatus configured as described above, the input binary image signal is examined for white or black continuity in the streak extracting unit 101, and is determined to be a certain value or more and a certain value or less. Only those having a continuation of (streaks) are extracted. Next, in the adjacent detection unit 102, the contact between the processing target data and the pattern extracted in the previous line is determined. In order to weight the streaks extracted by the streak extracting unit 101, the coordinate value adding unit 103 adds the coordinate values of the pixels in the horizontal direction and the vertical direction of the pattern extracted as the streaks. The effective pixel counting unit 104 counts the number of pixels of the streak extracted by the streak extraction unit 101. The circumscribed rectangle extracting unit 105 extracts a circumscribed rectangle of a streak group including the streaks extracted by the streak extracting unit 101 based on the result of the adjacent detection unit 102. In the coordinate addition value update unit 106, the circumscribed rectangle extraction unit 10
An added value of the coordinate values of the streaks included in the circumscribed rectangle extracted in step 5 is generated based on the result of the coordinate value adding unit 103. On the other hand, the effective pixel number updating unit 107
The number of pixels included in the streak included in the circumscribed rectangle extracted by the circumscribed rectangle extracting unit 105 according to the result of 2 is generated based on the result of the effective pixel counting unit 104.

The data obtained by the circumscribed rectangle extracting unit 105, the coordinate addition value updating unit 106, and the effective pixel number updating unit 107 are stored in the attribute storage unit 108 as attributes of the pattern. The attribute data stored in the attribute storage unit 108 is controlled line by line, and the attribute data stored in the line of interest is read out in the next line and processed in each processing block. The above series of processing is performed for each line, and when there is no more processing target data for the target line in contact with the pattern extracted in the previous line, the pattern extracted in the previous line is regarded as a confirmed pattern.

As described above, after extracting the streak, the upper and lower adjacency is determined in units of lines, and when there is no more data of the line of interest adjacent to the pattern obtained in the immediately preceding line, the specific object is determined and the pattern is determined. By determining the corresponding coordinate addition value and the number of pixels of the pattern, the center of gravity of the specific pattern can be calculated at high speed.
In addition, by determining the adjacency on a streak basis instead of a pixel basis, even if there is an oblique flaw on the specific pattern, the specific pattern (object) is not divided and can be recognized as one pattern.

Next, a first specific embodiment will be described with reference to FIG. This embodiment is described in claims 1 and 2.
This corresponds to the described invention. In FIG. 2, 201
Is a streak extraction unit for examining white or black continuity of a binary image signal input in raster scan and extracting data having continuity of a certain value or more, 202 is a coordinate value indicating a position of a pixel of interest Is a coordinate value generation unit that calculates the coordinates of both ends of the streak, 203 is an edge coordinate storage unit that stores the edge coordinates of the streak extracted by the edge detection unit 203, and 205 is An effective pixel counting unit that counts the number of pixels of the streak extracted by the streak extraction unit 201, 206 is a coordinate value addition unit that extracts the sum of the coordinate values of the streaks extracted by the streak extraction unit 201, and 207 is the value obtained from the immediately preceding line. A coordinate value conversion unit for generating a binary signal indicating a horizontal area of the pattern from the coordinate values of the circumscribed rectangle of the pattern, and 208 a streak extracted by the streak extraction unit 201 An adjacency detection unit 209 for examining contact with the pattern extracted in the immediately preceding line, based on the detection result of the adjacency detection unit 208, extracts the coordinates of the streak and the immediately preceding line in order to extract a circumscribed rectangle including the streak of the line of interest. Comparison with the coordinate value of the pattern
The coordinate value comparing / updating unit 210 to be updated is the coordinate value comparing / updating unit.
An effective pixel number updating unit that calculates the number of pixels of the streak in the circumscribed rectangle extracted in 209, and 211 sums the sum of the coordinate values of the streak in the circumscribed rectangle extracted in the coordinate value comparing and updating unit 209 An addition value update unit 212 includes a coordinate value comparison update unit 209, an effective pixel number update unit 210, and an addition value update unit 21.
1 shows an attribute storage unit that stores the value extracted in 1 as a pattern attribute.

Next, the operation of the first embodiment shown in FIG. 2 will be described. FIG. 3 is a conceptual diagram showing a processing mode of the specific target object extracting device according to the first embodiment, wherein 221 indicates a specific pattern and 222 indicates a streak group thereof. FIGS. 4 to 6 are time charts showing the flow of data according to the state of the adjacency. FIG. 4 shows the case where the adjacency between the streak and the circumscribed rectangle extracted in the immediately preceding line is not detected and only the streak exists. FIG. 5 shows the processing flow when the adjacency between the streak and the circumscribed rectangle extracted in the immediately preceding line is detected, and FIG. 6 shows the detection flow when the adjacency between the streak and the circumscribed rectangle extracted in the immediately preceding line is detected. The flow of processing when only the circumscribed rectangle of the immediately preceding line exists is shown.

Now, assuming that the line of interest is H, the streak extraction unit 201 for processing data extracts streaks 2A. Since the streak 2A does not have an adjacency with the circumscribed rectangle extracted in the previous line, the end point coordinates {(a, H), (b, H)} of the streak extracted in the edge coordinate storage unit 204.
Is the information (coordinate value) of the circumscribed rectangle, the count value P = (b−a + 1) of the pixel extracted by the effective pixel counting unit 205 and the coordinate value adding unit 206, and the added value of the coordinate value (Mx, M
y) is selected as an attribute of the streak 2A and stored in the attribute storage unit 212. This process is a series of processes shown in FIG. 4. As can be seen from FIG. 4, during the period (H level) in which a streak exists, addition of coordinate values and counting of the number of effective pixels are performed, and the falling edge of the streak is obtained. And save the processed value. Here, when a signal indicating the adjacentness is not obtained, each saved value [P1
= (B−a + 1), Mx1 = {(a + b) / 2} × (b−
a + 1), My1 = H × (b−a + 1)] are output as attributes of the new circumscribed rectangle 2A. Further, at the rising edge of the streak, the left end coordinate value (a, H) of the streak is calculated as
Right edge coordinate value (b, H) at falling edge of streak
Is detected and output as information of the new circumscribed rectangle 2A.

Next, when the line of interest is H + 1, the adjacency detection unit 208 examines the contact between the circumscribed rectangle extracted by the H line and the streak existing in the line of interest. In this line, the streak extraction unit 201
B is extracted, and a contact investigation with the circumscribed rectangle 2A extracted in the previous line is performed. Streak 2B on H + 1 line
Is in contact with the circumscribed rectangle 2A, and a circumscribed rectangle 2B is generated. At the same time, the number of pixels P = (d−c + 1) of the streak 2B and the sum of coordinates [Mx = ｛(c +
d) / 2｝ × (d−c + 1), My = (H + 1) × (d
−c + 1)] is the effective pixel counting unit 205 and the coordinate value adding unit 20
6 and the number of pixels (P1) of the streak in the circumscribed rectangle 2A and the coordinate addition value (Mx1,
My1), and updates it with the streak in the circumscribed rectangle 2B. This processing is a series of processing shown in FIG. 5, and as shown in FIG. 5, the circumscribed rectangle 2A extracted in the immediately preceding line is converted into a binary signal indicating the area and compared with a streak.

The binary signal is obtained by storing the leftmost coordinate value of the circumscribed rectangle extracted first in the immediately preceding line in the circuit before the effective period of the line of interest, and during the scanning of the image during the effective period. Rises when the horizontal coordinate value of the pixel is matched. For example, if the first circumscribed rectangle detected in the immediately preceding line is the circumscribed rectangle 2A, the target line H
Before entering the +1 image valid period, the leftmost coordinate value a of the circumscribed rectangle 2A is read into the circuit. Then, the horizontal coordinate value of the target pixel is a in the image valid period of the target line H + 1.
Then, the binary signal rises. When the binary signal is activated, the attributes (P, Mx, My, b, H) of the circumscribed rectangle 2A are read. If another circumscribed rectangle is detected in the immediately preceding line, the left end coordinate value of the circumscribed rectangle detected next to the circumscribed rectangle 2A in the immediately preceding line is read.

On the other hand, the judging of the adjacency is performed when the binary signal and the streak overlap each other, and after the adjacency is recognized, when both the streak and the binary signal disappear, the attributes possessed by both are recognized. Add and update. The binary signal falls when the horizontal coordinate value of the pixel of interest becomes b. Now, streak 2B and circumscribed square 2A
Are adjacent to each other. In this case, the horizontal coordinate value (c, d) and the number of effective pixels (dc + d) for the streak 2B
1), addition update value [{(c + d) / 2} × (d−c +
1)], {(H + 1) × (d−c + 1)} is obtained when the horizontal coordinate value of the pixel of interest becomes d. Also,
Immediately after d is scanned, both the binary signal and the streak disappear, so that the recognition of the adjacency is completed and the circumscribed rectangle 2B including the circumscribed rectangle 2A and the streak 2B.
{(A, H), (b, H + 1)} is generated, the respective parameters of the circumscribed rectangle 2A and the streak 2B are updated, and the above coordinate values of the circumscribed rectangle 2B and the number of effective pixels P =
{(B−a + 1) + (d−c + 1)}, addition update value Mx
= [{(A + b) / 2} × (b-a + 1)] + {(c +
d) / 2｝ × (d−c + 1), My = H × (b−a +
1) + (H + 1) × (d−c + 1) is newly extracted.

If one of the streaks or the circumscribed quadrangular region exists after the recognition of the neighbor and one of the streaks or the circumscribed rectangular region exists, and the former exists again before the other disappears, the adjacent state is maintained. If the streak is interrupted once, the attributes are updated in multiple streaks, and in the case where the circumscribed rectangle area is interrupted, the attributes are updated in multiple circumscribed rectangles, and then the attributes are updated in the streak and the circumscribed rectangle .

Next, when the line of interest is H + 2, the streak extracting section 201 extracts streaks 2C. The extracted streak 2C is the circumscribed rectangle extracted in the previous line by the adjacent detection unit 208, that is, the circumscribed rectangle 2B
An investigation of the neighborhood with is conducted. Here, in the adjacency detecting unit 208, it is determined that the streak 2C is in contact with the circumscribed rectangle 2B, and the circumscribed rectangle 2C is generated. At the same time, the parameters of the circumscribed rectangle 2B and the streak 2C are updated, and The number of pixels and the sum of the coordinate values of the streak are calculated by the effective pixel number updating unit 210 and the added value updating unit 21.
Extracted by 1. The above processing is performed until the pattern disappears, that is, up to the H + 11 line, and when there is no streak and only the circumscribed rectangle of the previous line exists,
As the extracted specific pattern, the number of effective pixels P and the added and updated values Mx and My of the circumscribed rectangle of the immediately preceding line are output.
It is stored in a specific area of the attribute storage unit 212. This process is a process shown in FIG. 6. As shown in FIG. 6, when the circumscribed rectangle region signal is activated and the attribute of the circumscribed rectangle is read, if it is determined that only the circumscribed rectangle region exists, the read is performed. The attribute of the obtained circumscribed rectangle is output as the attribute of the fixed pattern. Here, if a pattern having a low height is deleted when the pattern is determined, only a specific pattern of interest is extracted.

By processing a video signal by the above processing, circular or elliptical shapes scattered in an image to be processed can be obtained.
Only the number of effective pixels and the coordinate addition value of a specific pattern such as a rhombus are extracted, and after extracting one frame, if a calculation of Mx / P and My / P is performed by a programmable IC such as a CPU, the center of gravity value of the specific pattern can be quickly obtained. You can ask.

Next, a specific second embodiment will be described. This embodiment corresponds to the first and third aspects of the present invention. FIG. 7 is a diagram showing a circuit configuration of the second embodiment, and FIG. 8 is a timing chart for explaining the processing operation. It is a chart. In FIG. 7, reference numeral 301 denotes a streak extracting unit for extracting a streak from an input binary image signal; 302, a coordinate value generating unit for extracting a coordinate value of a pixel of interest from a video synchronization signal; An edge detection unit that detects the falling coordinate value, 304 is an edge coordinate storage unit that stores both edge coordinate values of the streak extracted by the edge detection unit 303, and 305 is the number of pixels of the streak extracted by the streak extraction unit 301. The effective pixel counting unit for counting, 306 is the streak extraction unit 30
A coordinate value adder 307 adds a coordinate value for each pixel of the streak extracted in step 1, and a coordinate value converter 307 generates a binary signal indicating a circumscribed quadrilateral region from a parameter of a pattern of a previous line stored as a coordinate value. 308 is an adjacency detector that determines contact between the streak and the circumscribed rectangle of the previous line, and 309 is for extracting a circumscribed rectangle including the streak extracted by the streak extractor 301 according to the result of the adjacency detector 308. A coordinate value comparison / update unit 310 for comparing and updating the coordinate values of the streak and the circumscribed rectangle extracted in the immediately preceding line, and 310 extracts the number of pixels of the streak in the circumscribed rectangle extracted by the coordinate value comparison / update unit 309 An effective pixel number updating unit, 311 is an added value updating unit for extracting the sum of the coordinate values of the streaks in the circumscribed rectangle extracted by the coordinate value comparing and updating unit 309, and 312 is a coordinate value comparing and updating unit. An attribute storage unit 309 for storing the values extracted by the effective pixel number updating unit 310 and the added value updating unit 311 as attributes of the pattern; 313 stores the barycentric coordinate value of the pattern determined based on the detection result of the adjacent detection unit 308 5 shows a center-of-gravity extracting unit to be calculated.

Next, the operation of the second embodiment configured as described above will be described with reference to a timing chart shown in FIG. In the input binary image data, the streak extracting unit 301 examines the continuity of white or black in the horizontal direction, and extracts streaks having the continuation of a certain value or more and a certain value or less. The edge of the extracted streak is detected by the edge detection unit 303, and the coordinate value of the pixel of interest extracted by the coordinate value generation unit 302 using the video synchronization signal is stored in the edge coordinate storage unit 304 when the edge is detected.
Is stored in At the same time, the streak extracted by the streak extracting unit 301 is subjected to a process of counting the number of pixels of the streak by the effective pixel counting unit 305 during the streak existence period, and the streak is present by the coordinate value adding unit 306. The addition of the coordinate values is performed both in the horizontal and vertical directions. Further, this streak is input to the adjacency detecting unit 308, and an adjacency check is performed with the circumscribed rectangle extracted in the immediately preceding line. Since the circumscribed rectangle to be subjected to the adjacent investigation is extracted by the coordinate value, it is converted by the coordinate value conversion unit 307 into a binary signal indicating the area of the circumscribed rectangle.

The adjacency detecting unit 308 checks the adjacency between the streak and the circumscribed rectangle of the immediately preceding line. If there is no adjacency and only the streak exists, the adjacency detection unit 308 generates the circumscribed rectangle using only the streak. When there is only the circumscribed rectangle extracted in the previous line, the new parameter is set with the streak and the circumscribed rectangle extracted in the immediately preceding line so that the pattern becomes the final pattern in the circumscribed rectangle. Control the subsequent blocks to generate them. The coordinate value comparing / updating unit 309 extracts the coordinates of the new circumscribed rectangle according to the detection result of the adjacent detecting unit 308. Here, the coordinate value comparing / updating unit 309 determines the end point coordinates of the streak extracted in the edge coordinate storage unit 304 as a new circumscribed rectangle when the adjacency detecting unit 308 determines that only the streak exists.
If it is determined in 308 that there is an adjacency, the edge coordinate storage unit
Comparing the end point coordinate value of the streak extracted in 304 with the coordinate value of the circumscribed rectangle extracted in the immediately preceding line, the smaller coordinate value of each left coordinate, the larger coordinate value of each right coordinate value, The top coordinate value of the circumscribed rectangle,
The coordinate value in the vertical direction of the line of interest is selected and output as a coordinate value indicating a new circumscribed rectangle. If only the circumscribed rectangle of the immediately preceding line exists, the circumscribed rectangle is used as the fixed pattern.

The number-of-effective-pixels updating section 310 includes an adjacency detecting section 308
If it is determined that only the streak exists, the number of pixels of the streak extracted by the effective pixel counting unit 305 is output as one of the attributes of the new circumscribed rectangle. on the other hand,
If it is determined that the streak is adjacent to the circumscribed rectangle extracted in the immediately preceding line by the adjacent detection unit 308, the value extracted in the effective pixel counting unit 305 is added to the number of pixels of the circumscribed rectangle extracted in the immediately preceding line. Is output as one of the attributes of the new circumscribed rectangle extracted by the coordinate value comparing / updating unit 309. If the adjacent detection unit 308 determines that only the circumscribed rectangle extracted in the immediately preceding line is present, it outputs the coordinate added value extracted in the immediately preceding line as the coordinate added value of the determined pattern.

The attribute storage unit 312 stores a circumscribed rectangle, a coordinate addition value, and the number of effective pixels of a pattern extracted for each line, and reads out the extracted lines in the extraction order in the next line. If the adjacent detection unit 308 determines that the pattern has been decided, the coordinate value comparison and update unit 309 and the effective pixel number update unit 310
, The added value updating unit 311 and the center of gravity extracting unit 313 in synchronization with the information. Center of gravity extractor 313
When notified of the information of the pattern determination, the effective pixel number updating unit 31 that is output as the number of effective pixels of the determined pattern
0 (P) and the coordinate addition value (Mx
, My), the value of the added value updating unit 311 is taken in, and the calculation of (Mx / P) and the calculation of (My / P) are performed. The two values obtained here are the center of gravity values (Mx / P, My / P) of the determined pattern. By performing the above series of processing, it is possible to extract the specific pattern scattered in the image and, at the same time, obtain the barycentric coordinates of the pattern.

Next, a third embodiment of the present invention will be described. This embodiment corresponds to the first and fourth aspects of the present invention. FIG. 9 is a diagram showing a circuit configuration of the present embodiment. In FIG. 9, reference numeral 401 denotes a streak extracting unit for extracting a streak from input binary image data;
Reference numeral 402 denotes a coordinate value generation unit that calculates the position of the pixel being scanned from the video synchronization signal, 403 denotes an edge detection unit that detects the edge of the streak, and 404 denotes the coordinate value of the streak edge extracted by the edge detection unit 403. An edge coordinate storage unit to be stored, 405 is an effective pixel counting unit that counts the number of pixels in the streak extracted by the streak extraction unit 401, and 406 adds the coordinate values of each pixel of the streak extracted by the streak extraction unit 401 A coordinate value adding unit, 407 is a coordinate value converting unit for converting a circumscribed rectangle of a streak group extracted by coordinate values into a binary signal indicating an area, and 408 is for examining an adjacency between the streak and the circumscribed rectangle extracted in the immediately preceding line. 409 is a coordinate value comparing / updating unit for generating a new circumscribed rectangle according to the detection result of the adjacent detecting unit 408, and 410 is a circumscribing quadrangle extracted by the coordinate value comparing / updating unit 409. Effective pixel number updating unit 411 for extracting the number of pixels of a streak group in a square
Is an addition value update unit that generates a coordinate addition value of the streak group in the circumscribed rectangle extracted by the coordinate value comparison / update unit 409,
412 is a coordinate value comparing / updating unit 409, an effective pixel number updating unit 410,
An attribute storage unit that stores the value extracted by the addition value update unit 411 as an attribute of the pattern, 413 is a centroid extraction unit that calculates the centroid value of the determined pattern, and 414 is the number of patterns determined in that line for each line. 5 shows a confirmed pattern counting unit for counting.

In the present embodiment, the determined pattern is counted for each line, and the center of gravity of the determined pattern counted in the immediately preceding line is calculated during the horizontal blanking period. FIG. 4 is a timing chart showing the concept of the operation of the embodiment. The input binary image signal is decomposed into streaks by a streak extracting unit 401. Next, the edge of the streak extracted by the streak extraction unit 401 is detected by the edge detection unit 403, and the edge coordinate storage unit 404
Stores the edge coordinates. Simultaneously with the above processing, the effective pixel counting unit 405 counts the number of pixels of the streak extracted by the streak extraction unit 401, and the coordinate value addition unit 406 adds the coordinate value of each pixel of the streak. The adjacency detection unit 408 examines the adjacency with the circumscribed rectangle extracted in the immediately preceding line. The circumscribed rectangle extracted in the immediately preceding line, which is an object to be adjacent to the streak in the adjacency detecting unit 408, is extracted by the coordinate value in the immediately preceding line, so that the horizontal region of the circumscribed rectangle converted by the coordinate value converting unit 407 is extracted. The binary signal shown is used.

The adjacency detection processing performed by the adjacency detection unit 408 is similar to the adjacency detection processing performed in the first and second embodiments, in that only the streak exists, and the streak and the circumscribed rectangle extracted in the immediately preceding line are used. Are adjacent,
The three states that only the circumscribed rectangle extracted in the immediately preceding line exists are detected. When the adjacent detection unit 408 detects a state in which only a streak exists, the coordinate value comparison update unit 40
9 outputs the coordinate values of both ends of the streak stored in the edge coordinate storage unit 404 as new circumscribed rectangle coordinate values, and the effective pixel number updating unit 410 selects a pixel value of the streak extracted by the effective pixel counting unit 405. Output and add value update unit 41
In 1, the coordinate addition value of the pixel of the streak extracted by the coordinate value addition unit 406 is selectively output.

On the other hand, when the adjacency detecting unit 408 determines that the streak is adjacent to the circumscribed rectangle extracted in the immediately preceding line, the coordinate value comparing / updating unit 409 determines the horizontal left of the streak and the circumscribed rectangle extracted in the immediately preceding line. The coordinate value and the right coordinate value are compared with each other, and the smaller left coordinate value and the larger right coordinate value are selected, and the vertical coordinate value of the upper end of the circumscribed rectangle extracted in the immediately preceding line, and the vertical direction of the line of interest The direction coordinate value is selected, the coordinate value as a new circumscribed rectangle is output, and the effective pixel number update unit 410 extracts the pixel value of the streak group in the circumscribed rectangle extracted in the immediately preceding line and the effective pixel count unit 405 The pixel value of the streak is added,
The coordinate value comparison / update section 409 outputs the number of pixels of the streak group in the new circumscribed rectangle extracted by the coordinate value update section 409, and the addition value update section 411 adds the coordinate addition value of the streak group in the circumscribed rectangle extracted in the immediately preceding line to the coordinate value addition The coordinate addition value of the streak extracted by the unit 406 is added to the coordinate value comparison update unit 40.
It is output as the coordinate addition value of the streak group in the new circumscribed rectangle extracted in 9.

When the adjacency detection unit 408 determines that only the circumscribed rectangle extracted in the immediately preceding line is present, the coordinate value comparing / updating unit 409 converts the coordinate value of the circumscribed rectangle extracted in the immediately preceding line into an effective pixel number updating unit. Reference numeral 410 denotes the pixel value of the streak group in the circumscribed rectangle extracted in the immediately preceding line, and the addition value update unit 411 selects and outputs the coordinate addition value of the streak group in the circumscribed rectangle extracted in the immediately preceding line. If the adjacent detection unit 408 determines that only the circumscribed rectangle extracted in the immediately preceding line exists, the circumscribed rectangle is regarded as a fixed pattern. At this time, the counter inside the fixed pattern counting unit 414 is incremented. Confirmed pattern counting unit 41
Reference numeral 4 denotes a configuration in which the value of the counter is incremented each time a pattern is determined, and is reset in units of one line.

On the other hand, the attribute storage unit 412 stores the parameters extracted by the coordinate value comparison / update unit 409, the effective pixel number update unit 410, and the added value update unit 411 for each line as attributes of the pattern. The attribute of the pattern is stored in an area different from the area where the attribute of the undetermined pattern is stored. When the video effective period of the line ends and the horizontal blanking period starts, the value of the fixed pattern counting unit 414 is transferred to the center of gravity extraction unit 413, and the center of gravity extraction processing is started. Immediately after receiving the information on the number of patterns extracted during the immediately preceding horizontal image validity period from the confirmed pattern counting unit 414, the centroid extracting unit 413 performs centroid extracting processing for the number of patterns.

More specifically, the pixel addition value and the effective pixel value of the first or last pattern extracted in the immediately preceding image effective period in the horizontal direction are read from the attribute storage unit 412, and the centroid is calculated from the value. Extract coordinate values. Next, the pixel addition value and the effective pixel value extracted from the second or the second from the last in the immediately preceding image effective period in the horizontal direction are stored in the attribute storage unit 412.
, And a barycentric coordinate value is extracted from the value. When this process has been performed the number of times given by the fixed pattern counting unit 414, the barycentric coordinate extraction process for that line is completed. More specifically, as shown in FIG. 10, when patterns are determined in the order of pattern 1, pattern 2,... Pattern n, the attribute storage unit 412 stores a determined pattern different from the undetermined pattern storage area. The pattern information (pixel addition value and effective pixel value) of the determined pattern 1, pattern 2,..., Pattern n is stored in the storage area. Then, when obtaining the barycentric value of each pattern, the barycentric value calculation is performed in order from pattern 1 to pattern n or from pattern n to pattern 1 in order to facilitate readout control. This means that the center of gravity of the patterns found sequentially from the right side or the left side of the image is calculated. By performing the above series of processing for one frame for each line, it becomes possible to extract the center of gravity coordinate value of the specific pattern in one frame period simultaneously with the extraction processing of the specific pattern.

Further, by performing the process of extracting the coordinates of the center of gravity during the horizontal blanking period, the value of the center of gravity can also be obtained for each line.
In the case of a system in which the subsequent processing is performed by a programmable IC such as a CPU, data output to the subsequent processing is always performed during the horizontal blanking period, which makes it easy to control the processing sequence on the CPU side, and When the number of confirmed patterns is large, it is possible to reduce the end processing of the processing being executed by the CPU, and the useless processing time is reduced.

Next, a fourth embodiment of the present invention will be described. This embodiment corresponds to the first and fifth aspects of the present invention. FIG. 11 is a diagram showing a circuit configuration of the present embodiment, and FIG. 12 is a timing chart showing an outline of the operation of the present embodiment. FIG. In FIG. 11, processing blocks denoted by reference numerals 501 to 513 are similar to the processing blocks denoted by reference numerals 401 to 413 in the third embodiment shown in FIG. In the present embodiment, each process is performed in the same manner as in the third embodiment, but the fixed pattern counting unit 514 resets the internal counter in units of one frame. The number of confirmed patterns counted by the confirmed pattern counting unit 514 is transferred to the centroid extracting unit 513 when the vertical blanking period starts, and thereafter, the centroid extracting process of the pattern decided by the centroid extracting unit 513 is executed. You. When the number of confirmed patterns is transferred from the confirmed pattern counting unit 514, the center of gravity extraction unit 513 starts the center of gravity extraction processing if the value is not 0. When the center-of-gravity extraction processing is started, the center-of-gravity extraction unit 513 determines the first to last one of the final patterns extracted from the final vertical image validity period among the final patterns stored in the attribute storage unit 512. The effective pixel and the coordinate addition value are read, and the calculation for calculating the center of gravity is performed.

If there are more confirmed patterns, the effective pixels of the second or last extracted final pattern in the immediately preceding vertical image validity period among the confirmed patterns stored in the attribute storage unit 512 next. And the coordinate addition value are read, and the calculation for calculating the center of gravity is performed. The center-of-gravity calculation process is performed after performing the number of times given by the fixed pattern counting unit 514,
The center of gravity calculation processing ends. As a result, as in the other embodiments, it becomes possible to extract a specific pattern included in an image during one frame period of the image, and to obtain the barycentric coordinate value.

Further, according to the present embodiment, the center of gravity calculation processing is performed during the vertical blanking period, and its value is also output during the vertical blanking period, so that the processing is performed using a programmable IC such as a CPU at the subsequent stage. When the present invention is applied to such a system, not only the sequence control of the programmable IC becomes easy, but also the processing interruption processing inside the programmable IC for the programmable IC to receive the barycentric coordinate value signal is performed once per frame. Wasteful processing time of the entire system is greatly reduced.

[0045]

As described above with reference to the embodiments, according to the present invention, a specific object area setting means for setting an area of a specific object in an image, and a center of gravity value of the specific object are determined. The processing of the specific target object center of gravity calculation means to be extracted is performed in parallel, and the area of the specific target object and the center of gravity value are simultaneously extracted, so that the center of gravity calculation of the scattered specific object in the image can be processed at high speed. A specific object extracting device can be realized.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram showing a basic configuration of a specific object extracting apparatus according to the present invention.

FIG. 2 is a block diagram showing a first specific embodiment of the present invention.

FIG. 3 is a conceptual diagram showing a processing mode according to the first embodiment shown in FIG. 2;

FIG. 4 is a timing chart for explaining a processing operation in an adjacent situation where only a streak exists in the first embodiment shown in FIG. 2;

FIG. 5 is a timing chart for explaining a processing operation when an adjacency between a streak and a circumscribed rectangle is detected in the first embodiment shown in FIG. 2;

FIG. 6 is a timing chart for explaining a processing operation in an adjacent situation where only a circumscribed rectangle exists in the first embodiment shown in FIG. 2;

FIG. 7 is a block diagram showing a second embodiment of the present invention.

FIG. 8 is a timing chart for explaining the processing operation of the second embodiment shown in FIG. 7;

FIG. 9 is a block diagram showing a third embodiment of the present invention.

10 is a timing chart for explaining the processing operation of the third embodiment shown in FIG.

FIG. 11 is a block diagram showing a fourth embodiment of the present invention.

FIG. 12 is a timing chart for explaining a processing operation of the fourth embodiment shown in FIG. 11;

FIG. 13 is a diagram showing a processing flow of a conventional method of extracting a center of gravity of a specific object.

14 is an explanatory diagram illustrating a processing mode of the center of gravity extraction method illustrated in FIG. 13;

[Explanation of symbols]

 101 Streak extraction unit 102 Adjacency detection unit 103 Coordinate value addition unit 104 Effective pixel counting unit 105 Circumscribed rectangle extraction unit 106 Coordinate addition value update unit 107 Effective pixel number update unit 108 Attribute storage unit 201,301,401,501 Streak extraction unit 202,302,402,502 Coordinate value generation unit 203,303,403,503 Edge Detection unit 204, 304, 404, 504 Edge coordinate storage unit 205, 305, 405, 505 Effective pixel counting unit 206, 306, 406, 506 Coordinate value addition unit 207, 307, 407, 507 Coordinate value conversion unit 208, 308, 408, 508 Adjacency detection unit 209, 309, 409, 509 Coordinate value comparison and update unit 210, 310, 410, 510 Unit 414,514 Fixed pattern counting unit

Claims (5)

[Claims] 1. A specific object for extracting a specific object defined in a predetermined shape as an object to be extracted from an image including the specific object and setting an area of the specific object in the image In an image processing apparatus having an object area setting means and a specific object gravity center calculating means for extracting the center of gravity value of the specific object, the processing of the specific object area setting means and the processing of the specific object gravity center calculating means are performed in parallel. And extracting the specific object area and the center of gravity value simultaneously. 2. A specific object extracting means for extracting a specific object prescribed in a predetermined shape as an object to be extracted from an image including the specific object, and a specific object extracting means for extracting the specific object by the specific object extracting means. A circumscribed rectangle extracting means for setting a circumscribed rectangle of the specific object, a specific object coordinate value calculating means for calculating a sum of coordinate values of respective pixels of the specific object, and a result of the specific object coordinate value calculating means. A center-of-gravity extracting means for extracting barycentric coordinates of the specific object based on the specific object extracting means, wherein the run length in the raster direction of the input binary image data is within a predetermined range. Only when there is a streak extracting means for extracting a binarized data string having a run length within the predetermined range as a streak constituting a part of the specific object, Detecting means for detecting the presence / absence of an adjacent streak and a circumscribed rectangle based on a streak aggregate formed earlier than this, and the number of pixels of the circumscribed rectangle and the streak aggregate and the number of pixels in the streak aggregate Attribute storage means for storing the sum of the coordinate values of the pixels, wherein the specific object coordinate value calculation means comprises: a coordinate value calculation means for calculating a sum of the coordinate values of the pixels of the streak; Calculation value updating means for updating the calculation result obtained by the value calculation means; effective pixel counting means for counting the effective pixels in the streak; and effective pixel number for updating the number of effective pixels obtained by the effective pixel counting means. Updating means for generating a circumscribed rectangle including the streak in accordance with an adjacent result of the adjacency detecting means, and calculating the circumscribed rectangle obtained by the coordinate value calculating means and the effective pixel counting means. The count value of the sum of the pixel of the coordinate values with the streak that contains the calculated value updating means that it is configured to update by said effective pixel number updating means specific object extracting apparatus according to claim. 3. A specific object extracting means for extracting, from an image including the specific object, a specific object defined in a predetermined shape as an object to be extracted, and a specific object extracted by the specific object extracting means. A circumscribed rectangle extracting means for setting a circumscribed rectangle of the specific object, a specific object coordinate value calculating means for calculating a sum of coordinate values of respective pixels of the specific object, and a result of the specific object coordinate value calculating means. A center-of-gravity extracting means for extracting barycentric coordinates of the specific object based on the specific object extracting means, wherein the run length in the raster direction of the input binary image data is within a predetermined range. Only when there is a streak extracting means for extracting a binarized data string having a run length within the predetermined range as a streak constituting a part of the specific object, Detecting means for detecting the presence / absence of an adjacent streak and a circumscribed rectangle based on a streak aggregate formed earlier than this, and the number of pixels of the circumscribed rectangle and the streak aggregate and the number of pixels in the streak aggregate Attribute storage means for storing the sum of the coordinate values of the pixels, wherein the specific object coordinate value calculation means comprises: a coordinate value calculation means for calculating a sum of the coordinate values of the pixels of the streak; Calculation updating means for updating the calculation result obtained by the value calculation means; effective pixel counting means for counting the effective pixels in the streak; and effective pixel number updating for updating the number of effective pixels obtained by the effective pixel counting means. Means for determining the circumscribed rectangle determined on the immediately preceding line when there is no streak on the line of interest in contact with the circumscribed rectangle determined on the immediately preceding line. Regarded as down, the center of gravity extraction means, the specific object extracting apparatus characterized by being configured to perform the centroid calculation streak aggregate contained in the outer contact rectangle immediately after the circumscribed rectangle has been determined. 4. A specific object extracting means for extracting a specific object previously defined in a predetermined shape as an object to be extracted from an image including the specific object, and a specific object extracting means for extracting the specific object by the specific object extracting means. A circumscribed rectangle extracting means for setting a circumscribed rectangle of the specific object, a specific object coordinate value calculating means for calculating a sum of coordinate values of respective pixels of the specific object, and a result of the specific object coordinate value calculating means. A center-of-gravity extracting means for extracting barycentric coordinates of the specific object based on the specific object extracting means, wherein the run length in the raster direction of the input binary image data is within a predetermined range. Only when there is a streak extracting means for extracting a binarized data string having a run length within the predetermined range as a streak constituting a part of the specific object, Detecting means for detecting the adjacency between the extracted streak and the circumscribed rectangle based on the streak aggregate formed earlier than this, the number of pixels of the circumscribed rectangle and the streak aggregate, and the pixels in the streak aggregate Attribute storage means for storing the sum of the coordinate values of
A fixed pattern counting means for counting the number of circumscribed rectangles fixed for each line; the specific object coordinate value calculating means; a coordinate value calculating means for calculating a sum of coordinate values of each pixel of the streak; A calculation value updating means for updating the calculation result obtained by the coordinate value calculation means; an effective pixel counting means for counting the effective pixels in the streak; and an effective pixel for updating the number of effective pixels obtained by the effective pixel counting means. Number updating means, wherein the adjacent detection means regards the circumscribed rectangle determined in the immediately preceding line as a determined pattern when the streak of the line of interest in contact with the circumscribed rectangle determined in the immediately preceding line is eliminated, and extracts the center of gravity. The means is configured to perform the center of gravity calculation of the streak aggregate for the count value counted by the confirmed pattern counting means during the horizontal blanking period. Object extraction device. 5. A specific object extracting means for extracting a specific object defined in a predetermined shape as an object to be extracted from an image including the specific object, and a specific object extracted by the specific object extracting means. A circumscribed rectangle extracting means for setting a circumscribed rectangle of the specific object, a specific object coordinate value calculating means for calculating a sum of coordinate values of respective pixels of the specific object, and a result of the specific object coordinate value calculating means. A center-of-gravity extracting means for extracting barycentric coordinates of the specific object based on the specific object extracting means, wherein the run length in the raster direction of the input binary image data is within a predetermined range. Only when there is a streak extracting means for extracting a binarized data string having a run length within the predetermined range as a streak constituting a part of the specific object, Detecting means for detecting the adjacency between the extracted streak and the circumscribed rectangle based on the streak aggregate formed earlier than this, the number of pixels of the circumscribed rectangle and the streak aggregate, and the pixels in the streak aggregate Attribute storage means for storing the sum of the coordinate values of
Fixed pattern counting means for counting the number of circumscribed rectangles fixed for each frame; the specific object coordinate value calculating means; a coordinate value calculating means for calculating a sum of coordinate values of each pixel of the streak; A calculation value updating means for updating the calculation result obtained by the coordinate value calculation means; an effective pixel counting means for counting the effective pixels in the streak; and an effective pixel for updating the number of effective pixels obtained by the effective pixel counting means. Number updating means, wherein the adjacent detection means regards the circumscribed rectangle determined in the immediately preceding line as a determined pattern when the streak of the line of interest in contact with the circumscribed rectangle determined in the immediately preceding line is eliminated, and extracts the center of gravity. The means is configured to perform the center of gravity calculation of the streak aggregate for the count value counted by the confirmed pattern counting means during the vertical blanking period. Constant object extraction device.