JP2009176083A - Noise removing/grouping circuit for star tracker - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To adequately perform the grouping of stars scattered in a cosmic space where continuity of image data is extremely deteriorated. <P>SOLUTION: A noise removing/grouping circuit for a star tracker includes: noise removal logic 4 for determining that the respective pixels of binarization data of the image data are invalid data when being two vertical/horizontal connections or less and valid data when being three vertical/horizontal connections or more; and grouping logic 5 for assigning grouping label numbers, which are allocated to the respective stars, by each pixel of the valid data in response to connection states to the peripheral pixels. When assigning the grouping label numbers, the plurality of pixels adjacent to the pixel of interest may have the different grouping label numbers. In this case, the smallest grouping label number among the grouping label numbers of the adjacent pixels is defined as the grouping label number of the pixel of interest. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a noise removal / grouping circuit for a start tracker, and more particularly, a noise removal / grouping circuit for a start tracker that is mounted on an artificial satellite and performs star grouping from image data by a start tracker (star tracker) for attitude control. About.

  As this type of conventional technology, in addition to the data of only the portion of the CCD image signal that exceeds a predetermined threshold, all the data of only the portion of the preset window area (measurement area) of the CCD image signal is extracted. "Star detection circuit for star sensor and its star data extraction method" are known (for example, see Patent Document 1). The purpose of this is to make it possible to extract star data necessary for high-precision star position calculation without greatly increasing the amount of data to be handled or increasing the load on the arithmetic processing circuit.

  In addition, it has multiple systems of sensors to capture data from different fields of view, picks up the image data at a predetermined discrete level, stores it in the line buffer, checks the continuity of the image data, and calculates the line bias level An “image reading method” is also known in which a new discrete level is determined (see, for example, Patent Document 2). In this method, when the data stored in the line buffer is continuously equal to or greater than the predetermined number of data for one line, it is determined as abnormal data and only other normal systems are processed.

  Although it is not a technique for grouping stars, pay attention to each pixel included in the image data, and the target pixel is a new pixel that forms a new figure based on the labels of surrounding pixels, or an existing figure Also known is an “image processing apparatus” that determines whether a pixel is an additional pixel, and if it is an additional pixel, whether or not the pixel of interest is a connected pixel that connects a plurality of different figures. Yes (for example, see Patent Document 3). If it is a connected pixel, the larger label number is rewritten to the smaller label number, and the area of the figure with the larger label number is added to the area of the figure with the smaller label number.

JP2000-180201 (2nd page-3rd page, FIG. 1) JP-A-8-65468 (page 3 to page 4, FIG. 2) JP2007-122594 (5th page, 14th page to 15th page, FIG. 7)

  However, in the technique described in Patent Document 1 described above, all the data of only the preset window area portion is also extracted. However, for the portions other than the window area portion, for each pixel. Since the signal level is extracted by comparing with the threshold value, the star data with a small brightness even though it is large is discarded, and as a result, the star detection accuracy is lowered.

  Further, in the technique described in Patent Document 2, the continuity of the imaging data is checked, but this is only about the connectivity of the pixel columns read from any one column of the image sensor, and an image composed of a plurality of columns. Continuity between data strings is not subject to processing at all. This is considered to be a general function that only copes with stray light processing such as solar reflected light and moon interference in the line buffer, that is, a smear phenomenon occurring in the image sensor. Therefore, there is a first problem that image data as two-dimensional data cannot be correctly processed.

  In addition, since image processing is performed by calculation processing via software, the processing time is long and performance is degraded compared to hardware calculation processing, so the imaging cycle cannot be shortened and the posture must be changed at high speed. There is also a second problem that the attitude cannot be controlled with high accuracy in the case of an artificial satellite.

  The technique described in Patent Document 3 is an image processing technique for calculating a feature amount (area) of a graphic composed of connected pixel groups included in image data captured by a camera, and in particular, performs star grouping. It is not intended. For that reason, there is no description of logic that assumes the case of connecting figures with different label numbers, so images of a group of point light sources (stars scattered in outer space) that significantly reduce the continuity of image data are used in the radiation environment. It is not suitable for application to an image processing apparatus that performs grouping of stars below.

  Accordingly, an object of the present invention is to speed up the imaging cycle (frame cycle) by reducing the processing time of noise removal processing and star grouping considering the vertical connection of image data, and in space. It is an object of the present invention to provide a noise removal / grouping circuit for a start tracker suitable for performing star grouping in a radiation environment.

  The noise removal / grouping circuit of the present invention is a noise removal / grouping circuit for a star tracker that performs star grouping from image data by a star tracker for attitude control of an artificial satellite or the like. Noise removal logic (4 in FIG. 1) for determining invalid data if the vertical / horizontal connection is 2 or less, and valid data if the vertical / horizontal connection is 3 or more, and the valid data in the binarized data from the noise removal logic. Each pixel includes grouping logic (5 in FIG. 1) for assigning a grouping label number assigned to each star according to the connection state with surrounding pixels.

  When assigning grouping label numbers, there may be cases where a plurality of pixels adjacent to the pixel of interest have different grouping label numbers, the smallest grouping label number among the grouping label numbers of the adjacent pixels is assigned to the pixel of interest. The grouping label number is used.

  More specifically, the noise removal / grouping circuit of the present invention is a noise removal / grouping circuit for a star tracker that performs star grouping from image data by a star tracker for attitude control of an artificial satellite or the like. A line buffer (FIFO) for noise removal (1 in FIG. 1) that holds binarized data of 1 LineDelay to n (even) LineDelay from the binarized data of the image, and binarized data of the current read line of the image data; The binarized data for 1 LineDelay to nLineDelay input from the noise removal line buffer is developed on the register in the form of a matrix of (n + 1) × (n + 1) on a pixel basis, and (n / 2) LineDelay If the pixel is less than 2 vertical / horizontal connections, it will be judged as invalid data. Noise removal logic (4 in FIG. 1) that determines that the data is valid if the vertical / horizontal connection is 3 or more, and for each valid data pixel in the binarized data from the noise removal logic, depending on the connection state with surrounding pixels A grouping logic (5 in FIG. 1) for assigning the grouping label number assigned to each star, and a grouping label management table (RAM) for storing grouping label numbers for each (n / 2) LineDelay from the noise removal logic (6 in FIG. 1) and a grouping label information line buffer (FIFO) for storing grouping label information (grouping label number and binarized data) for (n / 2) LineDelay from the noise removal logic (FIG. 1). 2).

  The grouping logic then assigns the grouping label number from the grouping label management table and the grouping label information line buffer for the binarized data for (n / 2) LineDelay from the noise removal logic when assigning the grouping label number. ((N + 1) / 2) With reference to the grouping label information for LineDelay, when a plurality of pixels adjacent to the target pixel may have different grouping label numbers, the grouping label numbers of the adjacent pixels The smallest grouping label number among these is the grouping label number of the target pixel.

  Note that the case of having different grouping label numbers means that the pixel of interest is adjacent to two discrete pixels, two adjacent pixels adjacent in an oblique direction, or three or more pixels. It is.

  The first effect of the present invention is that, for each pixel, real-time processing is performed in which the correlation between the front and rear in the line direction (horizontal direction) and the vertical direction (vertical direction) is recognized, and noise removal is performed in consideration of vertical connection. Therefore, it is possible to detect star data with low brightness, widen the limit grade width of stars, and improve star detection accuracy.

  The second effect of the present invention is that the star grouping process is shortened by performing the star grouping process on the continuously input star data in real time by the grouping determination logic, thereby reducing the processing time. As a result, the imaging cycle (frame cycle) can be increased, and the star data can be captured even for a high-speed maneuver of the satellite.

  The third effect of the present invention is that when a plurality of pixels adjacent to the target pixel may have different grouping label numbers during grouping label number assignment, the grouping label numbers of the adjacent pixels may be different. Since the smallest grouping label number is the grouping label number of the pixel of interest, even image data of a group of point light sources (stars scattered in outer space) whose continuity of image data is significantly reduced This means that star groupings can be performed accurately.

  Next, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram showing a star data processing apparatus having a noise removal / grouping circuit 9 according to the present invention. This star data processing apparatus comprises a CCD 10, a signal processing circuit 11, a noise removal / grouping circuit 9, and a star position calculation software 12.

  The CCD 10 is mounted on an artificial satellite or the like and images a star. The signal processing circuit 11 performs analog processing on the analog data captured by the CCD 10 and extracts only the image data. The extracted image data is converted from analog to digital, and binarized data and effective image data (star data) are output to the noise removal / grouping circuit 9. The noise removal / grouping circuit 9 performs noise removal and star data grouping processing in real time on the binarized data and the star data continuously transferred from the signal processing circuit 11. The star position calculation software 12 calculates the position of the star (center of gravity position) based on the processing result in the noise removal / grouping circuit 9.

  The noise removal / grouping circuit 9 includes a noise removal line buffer 1, a grouping label information line buffer 2, a luminance information line buffer 3, a noise removal logic 4, a grouping logic 5, and a RAM configuration each having a FIFO configuration. It is composed of a grouping label management table 6, a star data memory 8, and a star data write control logic 7.

  The binarized data from the signal processing circuit 11 is input to the noise removal line buffer 1 and the noise removal logic 4. Each bit of the binarized data indicates “1” as an effective pixel, and “0” indicates an invalid pixel. The effective pixel means star data, and the invalid pixel means data that is not star data, that is, noise. The star data from the signal processing circuit 11 is written into the luminance information line buffer 3. This star data includes address information and luminance information indicating the brightness of the star. The address information is an address on the star data processing device corresponding to the position of the star.

  The noise removal line buffer 1 is a FIFO for 4 lines (5K bits or more) for storing the binarized data from the signal processing circuit 11, and each binarized data for 1 LineDelay to 4LineDelay from the current read Line. Output to noise removal logic 4.

  The luminance information line buffer 3 is a FIFO (4 Kbytes or more) for storing star data delayed by two lines from the current read line, and outputs image data for 2 LineDelay to the star data write control logic 7.

  The noise removal logic 4 uses the current read binarized data from the signal processing circuit 11 and the binarized data for 1 LineDelay to 4LineDelay from the noise removal line buffer 1 for pixel data for 2 LineDelay. Is determined to be invalid data if it is less than 2 vertical / horizontal connections, and valid data if it is 3 or more vertical / horizontal connections. Then, the binarized data for 2 LineDelay reflecting the result is output to the grouping logic 5 and the star data write control logic 7. The noise removal logic 4 can be realized by a normal digital IC.

  The grouping logic 5 assigns a grouping label number assigned to each star for each pixel of binary data for 2 LineDelay from the noise removal logic 4. At this time, the grouping label information one line before (3LineDelay) from the grouping label information line buffer 2 and the grouping label numbers for 2LineDelay from the grouping label management table 6 are referred to. Grouping label information refers to a grouping label number and binarized data. The grouping label numbers for 2 LineDelay thus distributed are output to the star data write control logic 7. The grouping logic 5 can be realized by a normal digital IC.

  Further, the grouping logic 5 outputs the grouping label information for 2 LineDelay composed of the grouping label numbers for 2 LineDelay thus assigned and the binarized data for 2 LineDelay from the noise removal logic 4 to the line buffer 2 for grouping label information. Then, the grouping label management table 6 is updated with the grouping label numbers corresponding to the distributed 2LineDelay.

  The grouping label information line buffer 2 is a FIFO for one line storing the grouping label information for 2 LineDelayed by the grouping logic 5. When the grouping label information for 2LineDelay is input from the grouping logic 5, the line buffer 2 stores the grouping label information. The grouping label information for 2 LineDelay is output to the grouping logic 5 as grouping label information for 3 LineDelay.

  The grouping label management table 6 is an SRAM (512 bytes or more) that stores grouping label numbers for each pixel of 2 LineDelay from the grouping logic. The grouping label number is information indicating whether or not they are finally connected to each other depending on the difference with other assigned grouping label numbers.

  On the other hand, the luminance information line buffer 3 is a FIFO for one line storing the address information and luminance information of the star data from the signal processing circuit 11, and controls the star data of 2 LineDelay from the current read line to the star data writing control. Output to logic 7.

  The stellar data write control logic 7 includes binary data for 2 LineDelay from the noise removal logic 4, grouping label number for 2 LineDelay from the grouping logic 5, and address information for 2 LineDelay from the luminance information line buffer 3. And the level information are input, and a write control signal is output to the star data memory 8. The star data write control logic 7 can be realized by a normal digital IC.

  The star data memory 8 is an SRAM (64 Kbytes or more) that stores grouping label numbers, vertical / horizontal addresses, and level information for each pixel of star data. The star position calculation software 12 calculates the position (center of gravity) of the star from the contents of the star data memory 8.

  FIG. 2 shows an example of a memory map of the noise removal line buffer 1. In this memory map, each pixel 1 to pixel 1000 is a current read line, and binary data that is delayed by four lines from the current read line is shown. The binarized data of the current read Line to 3 LineDelay is sequentially output to the noise removal logic 4 as the binarized data of 1 LineDelay to 4LineDelay.

  FIG. 3 shows an example of a memory map of the luminance information line buffer 3. In this memory map, star data (address information and luminance information) is shown in one-to-one correspondence with each pixel of the current reading Line to 2 LineDelay in the memory map of the noise removal line buffer 1. The star data of the current read Line to 1 LineDelay is sequentially changed to star data of 1 LineDelay to 2LineDelay, and the star data of 2LineDelay is output to the star data write control logic 7.

  FIG. 4 shows an example of the contents of the grouping label management table 6. The grouping label management table 6 is a table in which adjacent grouping label numbers are registered for each grouping label number having a one-to-one correspondence with 2 LineDelay pixels. In the initial state, the grouping label number and the “adjacent grouping label number” are equal, and when the grouping label number is input from the grouping label logic 7, the “adjacent grouping label number” is updated. In FIG. 4, the “adjacent grouping label numbers” of the grouping label numbers 2 and 4 are updated to “1”, and as a result, the pixels of the grouping label numbers 1, 2, and 4 are connected. The “adjacent grouping label number” of the grouping label number maintains “3” in the initial state.

  FIG. 5 shows a specific example of determination of valid data and invalid data in the noise removal logic 4. Each block indicates a pixel, a solid block indicates an effective pixel, and a white block indicates an invalid pixel. The binarized data of the current read line input from the signal processing circuit 11 and the binarized data for 1 LineDelay to 4LineDelay input from the noise removal line buffer 1 are 5 × 5 on the register in the noise removal logic 4. It is expanded in pixel form in a matrix. Note that the size of the matrix is not limited to 5 × 5, and may be 3 × 3 or more so that three connected pixels can be detected.

  Then, based on the correlation between the binarized data of the center pixel (center data of 2LineDelay) and other binarized data, it is determined whether the binarized data of the center pixel is valid data or invalid data. Judgment is made by looking at the connected state in the direction, vertical direction and diagonal direction. Here, if the number of connected effective pixels is 3 or more, the data composed of these pixels is valid data, and if it is 2 or less, it is invalid data.

  As described above, in the present invention, since the correlation between the line direction (horizontal direction) and the longitudinal direction (vertical direction) is recognized and noise removal is performed in consideration of the connection in the vertical direction, a star with low brightness is used. Data can be detected, the limit grade range of stars is expanded, and the accuracy of star detection can be improved.

  FIG. 5A shows the effective pixel (dot-filled block) and (i) of (i-1, i-3) with respect to the effective pixel (shaded block) at the center (i, i-2) of interest. −1, i−4) effective pixels (dot-filled blocks) are connected, so three effective pixels are connected, and thus the pixel of interest is determined to belong to effective data. . The process to reach here is shown in time series in FIG.

  In FIG. 6A, binarized data including a pixel of interest is input at t = 0, binarized data of 1 LineDelay including a pixel connected to the pixel of interest is input at t = −1, and t = −2. 2LineDelay binarization data including a pixel connected to the target pixel is input. 6B shows the state after 1 Line Delay in FIG. 6A, and FIG. 6C shows the state after 2 Line Delay in FIG. The carry-down of the binarized data from the noise removal line buffer 1 is due to the FIFO action.

  On the other hand, FIG. 5B shows a case where it is determined that the central pixel belongs to invalid data because there is no effective pixel connected to the central pixel. The process to reach here is shown in time series in FIG. FIG. 7 should be understood in the same manner as FIG.

  When the noise removal logic 4 determines that the data is valid, the grouping logic 5 performs grouping label determination on the valid data and assigns a grouping label number. FIG. 8 shows this state. The grouping logic 5 includes binary data for 2 LineDelay from the noise removal logic 4, grouping label numbers for 2 LineDelay from the grouping label management table 6, and grouping labels for 3 LineDelay from the grouping label information line buffer 2. Information (binarized data and grouping label number) is input.

  In FIG. 8, the pixel X in the binarized data for 2 LineDelay is about to be subjected to grouping label determination and grouping label number as a target pixel. The pixel D is a pixel immediately before the pixel X, and the pixels A, B, and C are pixels adjacent to the pixel X in the binarized data for 3 LineDelay. Grouping label numbers ADT, BDT, CDT, and DDT indicate grouping label numbers corresponding to the pixels A, B, C, and D, respectively.

  FIG. 9 is a table for determining grouping label numbers, and shows the output of the grouping label number XDT for the pixel X and the grouping label number to be updated using the binarized data of the pixels X, A, B, C, and D as input conditions. ing. The grouping label number is a series number of 1 origin. The grouping label number and grouping label number XDT to be updated correspond to the grouping label number and “adjacent grouping label number”, respectively, in the grouping label management table 6.

  In FIG. 9, if the pixel X is invalid (“0”), the pixel X is “0” and there is no grouping label number to be updated. If the pixel X is valid (“1”) and the adjacent pixels A, B, C, and D are invalid, the pixel X is a pixel of new valid data, and therefore a new grouping label number is assigned as XDT. Is written in the “adjacent grouping label number” field of the grouping label management table 6. The grouping label number assigned at this time is obtained by adding 1 to the largest grouping label number registered in the grouping label management table 6.

  When at least one of the adjacent pixels A, B, C, and D is valid (11000, 10100, 10010, 10001), the pixel X is connected to the pixel, so the grouping label number of the pixel is XDT. There is no grouping label number to be updated.

  When adjacent to two adjacent pixels A, B, C, and D in the horizontal or vertical direction (11100, 11001, 10110), the grouping label number of the “latest” pixel is XDT. This is because two pixels adjacent in the horizontal or vertical direction always have the same grouping label number. For example, in the case of 11100, A and B adjacent to the horizontal direction and the pixel X are adjacent, and A, B, and X have the same grouping label number.

  When adjacent to two discrete pixels among the adjacent pixels A, B, C, and D (11010, 10011), or adjacent to two pixels adjacent in an oblique direction (10101), or three or more When adjacent to the pixel (11110, 10101, 1101, 10111, 11111), the smallest grouping label number among the grouping label numbers of the adjacent pixels is XDT. This is because in these cases, adjacent pixels may have different grouping label numbers. FIG. 10 is a diagram illustrating such a case, in which the pixel of interest X is adjacent to two pixels B and D that are adjacent in the oblique direction (10101). Pixel B has a grouping label number “4”, and pixel D has a grouping label number “5”. When the pixel B is the target pixel, the grouping label number “5” is assigned because it corresponds to the case of 10010, and when the pixel D is the target pixel, the grouping label number “ 4 "is given. In this case, the target pixel X corresponds to the case of 10101, XDT is the minimum grouping label number “4”, and the grouping label number to be updated is “5”.

  In this way, in the present invention, since logic is provided assuming that figures with different label numbers are connected, a group of point light sources (stars scattered in outer space) that significantly reduce the continuity of image data. It is possible to accurately group stars even in a radiation environment.

  The XDT obtained as described above is written in the “adjacent grouping label number” column of the grouping label management table 6 as a grouping label number for updating the largest grouping label number among the grouping label numbers of adjacent pixels. .

  FIGS. 11 to 13 show three typical examples for obtaining the XDT based on the grouping label number determination table (FIG. 9). FIG. 11 shows the case of 11000. Since ADT = 3, XDT = ADT = 3. There is no writing to the grouping label management table 6.

  FIG. 12 shows the case of 10000. Since grouping label number “4” is registered in the grouping label management table 6, XDT = 5 is registered in the grouping label management table 6.

  FIG. 13 shows the case of 11010. Since ADT = 1 and CDT = 2, XDT = min (ADT, CDT) = 1 and “adjacent grouping label number” = max (ADT, CDT) = 2. Accordingly, “2” in the “adjacent grouping label number” column of the grouping label number “2” in the grouping label management table 6 is updated to “1”. Therefore, it can be seen that the pixel with the grouping label number “1” and the pixel with the grouping label number “2” are the same grouping data.

  The grouping label number obtained by the grouping logic 5 as described above is also input to the stellar data write control logic 7. Further, it is written into the grouping label information line buffer 2 as grouping label information together with the binarized data for 2 LineDelay.

  The star data writing control logic 7 uses the grouping logic 5 to output the two-line delayed star data (address information and level information) from the luminance information line buffer 3 corresponding to the binarized data from the noise removal logic 4. The determined grouping label number is written in the star data memory 8. The software 9 calculates the position of the star (center of gravity position) from the address information, level information and grouping label number written in the star data memory 8.

  As described above, in the present invention, star grouping processing is performed in real time by hardware on star data that is continuously input, so that the star grouping processing can be shortened and the imaging cycle (frame cycle) is shortened. Therefore, it is possible to capture star data even for high-speed maneuvers of satellites.

1 is a block diagram showing a star data processing apparatus having a noise removal / grouping circuit according to the present invention. The figure which shows an example of the memory map of the line buffer 1 for noise removal The figure which shows an example of the memory map of the line buffer 3 for luminance information The figure which shows the example of the content of the grouping label management table 6 The figure which shows the specific example of the effective data in the noise removal logic 4, and invalid data The figure which shows the process of the effective data in FIG. 5 in time series The figure which shows the process of invalid data in FIG. 5 in time series The figure which shows the structure for grouping label number assignment Figure showing a grouping label number determination table The figure which illustrates the case where an adjacent pixel has a different grouping label number The figure which shows the 1st example which calculates | requires grouping label number XDT The figure which shows the 2nd example which calculates | requires grouping label number XDT The figure which shows the 3rd example which calculates | requires grouping label number XDT

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Line buffer for noise removal 2 Line buffer for grouping label information 3 Line buffer for luminance information 4 Noise removal logic 5 Grouping logic 6 Grouping label management table 7 Stellar data write control logic 8 Stellar data memory 9 Noise removal / grouping circuit 10 CCD
11 Signal processing circuit 12 Star position calculation software

Claims (3)

In the noise removal / grouping circuit for the star tracker that performs stellar grouping from image data by a star tracker for attitude control of artificial satellites, etc.
Noise removal logic for determining invalid data if each pixel of the binarized data of the image data is vertical / horizontal 2 connection or less, and valid data if vertical / horizontal 3 connection or more,
Grouping logic for assigning a grouping label number assigned to each star according to the connection state with surrounding pixels for each pixel of the effective data in the binarized data from the noise removal logic,
When a plurality of pixels adjacent to the target pixel may have different grouping label numbers when assigning the grouping label number, the smallest grouping label number among the grouping label numbers of the adjacent pixels is grouped with the target pixel. A noise removal / grouping circuit for a start tracker, characterized in that the label number is used. In the noise removal / grouping circuit for the star tracker that performs stellar grouping from image data by a star tracker for attitude control of artificial satellites, etc.
A noise removal line buffer (FIFO) that holds binarized data of 1 LineDelay to n (even) LineDelay from the binarized data of the current read line of the image data;
The binarized data of the current read line of the image data and the binarized data for 1 LineDelay to nLineDelay input from the noise removal line buffer are in the form of a matrix of (n + 1) × (n + 1) on the register. In (n / 2) LineDelay binarized data, it is determined that the data is invalid if the pixel is less than 2 vertical / horizontal connections, and valid data if the pixel is more than 3 vertical / horizontal connections. Logic and
A grouping logic for assigning a grouping label number assigned to each star according to a connection state with surrounding pixels for each pixel of the effective data in the binarized data from the noise removal logic;
A grouping label management table (RAM) for storing grouping label numbers for each pixel of (n / 2) LineDelay from the noise removal logic;
A grouping label information line buffer (FIFO) for storing (n / 2) LineDelay grouping label information (grouping label number and binarized data) from the noise removal logic,
The grouping logic, when assigning the grouping label number, for the binarized data for (n / 2) LineDelay from the noise removal logic, the grouping label number from the grouping label management table and the grouping label information line With reference to the grouping label information for ((n + 1) / 2) LineDelay from the buffer, when a plurality of pixels adjacent to the target pixel may have different grouping label numbers, A noise removal / grouping circuit for a start tracker, wherein the smallest grouping label number among the grouping label numbers is used as the grouping label number of the target pixel.   The case of having a different grouping label number is a case where the pixel of interest is adjacent to two discrete pixels, adjacent to two pixels adjacent in an oblique direction, or adjacent to three or more pixels. 3. A noise removal / grouping circuit for a start tracker according to claim 1, wherein the noise removal / grouping circuit is used.

Images