Image segmentation method, image segmentation apparatus, image processing method, and image processing apparatus

• US 7,298,899 B2
• Filed: 05/27/2003
• Issued: 11/20/2007
• Est. Priority Date: 05/27/2002
• Status: Active Grant

• Current Assignee: Rakuten Group, Inc.
• Original Assignee: Hiroshima University
• Inventors: Koide, Tetsushi, Morimoto, Takashi, Mattausch, Hans Juergen, Harada, Youmei
• Primary Examiner(s): Chen; Wenpeng
• Assistant Examiner(s): Ge; Yuzhen
• Application Number:
  US10/445,247
• Publication Number:
  US 20030219157A1
• Time in Patent Office: 1,638 Days
• Field of Search: 382155-161, 382/173
• US Class Current: 382/173
• CPC Class Codes: G06T 2200/28 involving image processing hardware
• CPC Class Codes: G06T 2207/20084 Artificial neural networks [ANN]
• CPC Class Codes: G06T 5/20 using local operators
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51
• CPC Class Codes: G06V 10/267 by performing operations on regions, e.g. growing, shrinking or watersheds
• CPC Class Codes: G06V 10/955 using specific electronic processors

Abstract

Cells i corresponding to pixels are initialized into a non-excitation state, to calculate coupling weights W_ik between the eight cells k adjacent to the cells i, thereby determining leader cells p_i=1 based on calculation results. Next, one leader cell yet to be excited is selected as a self-excitable cell. The selected cell is put into the excitation state, the excitable cells are selected based on the coupling weights between the adjacent cells, and the selected cells are put into the excitation state. These operations are repeated until no excitable cell is detected any more and, if there no excitable cell is detected any more, inhibition processing is performed, thereby completing image segmentation of one region. These operations are repeated until there is no non-excited and non-inhibited leader cell any more, thereby pinpointing regions belonging to the same category from an input image and identifying them as an image segmentation regions.

Claims

1. An image segmentation method which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region, said method comprising:

a preparation step including,

an initialization step of putting, into a non-excitation state, a cell which is an individual image segmentation unit corresponding to pixels of the input image,

a taking step of pixel values of the pixels corresponding to the cell, and calculating each coupling weight between a plurality of adjacent cells, and

a determination step of determining leader cells based on each calculation result in the taking step;

a self-excitable cell detection step of selecting one of the leader cells determined by the determination step to detect the one leader cell as a self-excitable cell;

a self-excitation step of putting, into an excitation state, the self-excitable cell detected in the self-excitable cell detection step;

an excitable cell detection step of detecting an excitable cell from adjacent cells based on said coupling weights between cells in the excitation state including the leader cells and the adjacent cells;

an excitation step of putting, into an excitation state, the excitable cell detected in the excitable cell detection step; and

an inhibition step of putting, into an inhibition state, a cell in the excitation state if no cell is detected in the excitable cell detection step, wherein;

image segmentation of one region is completed by repeating the excitation step until no cell is detected any more in the excitable cell detection step; and

the image segmentation of all the regions is completed by repeating the respective steps until no leader cell in the non-excitation state is detected any more in the self-excitable cell detection step,

wherein I_i is a pixel value of the cell i (i indicates a cell number) which is an individual image segmentation unit corresponding to the pixels of the input image, x_i is a variable which indicates whether the cell i is in the excitation or non-excitation state, p_i is a variable which indicates whether the self-excitation is permitted or not, W_ik is the coupling weight between the adjacent cells i and k, φ

_p is a threshold value which decides whether the self-excitation is permitted or not, φ

_z is a threshold value which decides whether the self-excitation is permitted or not, z_i is a variable which indicates whether the state of cell i is changed or not, and z is a variable of a global suppressor which decides whether are cells having changed to the excitation state based on a logical sum of the values z_i'"'"'s of all the cells;

in the preparation step, the variable x_i of the image segmentation cell i is set to 0 (x_i=0;

non-excitation), the pixel values I_i corresponding to the cells i are taken in to calculate the coupling weight W_ik between the adjacent plural cells i and k, so that if a total sum of calculation results is larger than the threshold value φ

_p, p_i=1 (self-excitable state) is set and if the total sum is equal to or smaller than the threshold value φ

_p, p_i=0 (non-self-excitable state) is set for initialization, and z=0 where the variable z is the global suppressor is set for initialization;

in each self-excitable cell detection step, at least one non-excited leader cell is selected from among the leader cells of p_i=1 determined in the determination step to detect the selected cells as the self-excitable cells;

in each self-excitation step, the variables x_i and z_i of the self-excitable cells i detected in the self-excitable cell detection step are set to 1 (x_i=1;

self-excitation, z_i=1;

changed state), respectively;

in the excitable cell detection step, if a total sum of the coupling weights W_ik between the excited cells k (x_k=1) adjacent to the cells i in the non-excitation state is larger than the threshold value φ

_z the cells i are detected to be the excitable cells;

in each excitation step, the variables x_i and z_i of all the cells i detected in the excitable cell detection step are set to 1 (x_i=1;

excitation state, z_i=1;

changed state), respectively, and the variable z_i of the cells i already in the excitation state (x_i=1 and z_i=1) is set to 0 (z_i=0;

unchanged state); and

in each inhibition step, if no cell is detected in the excitable cell detection step, the cells of x_i=1 (excitation state) are set to x_i=0 (non-excitation state) and z_i=0 (unchanged state), and if p_i=1, the cells are set to p_i=0 (inhibition state).

2. The image segmentation method according to claim 1, wherein:

when the input image is a grayscale image, a brightness value is used as the pixel value.

3. The image segmentation method according to claim 1, wherein:

when the input image is a color image, color information is used as the pixel value.

4. An image segmentation apparatus which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region to selectively output an image of the arbitrary image segmentation region, said apparatus comprising:

an input image memory which stores pixel values of the input image;

a coupling weight calculation circuit which reads out the pixel values from the input image memory to calculate a coupling weight between each image segmentation cell corresponding to each pixel and an adjacent cell;

a leader cell determination circuit which determines, based on the coupling weights calculated by the coupling weight calculation circuit, as a leader cell, the cell in which a total sum of the coupling weights with the adjacent cells is in excess of a reference value;

an image segmentation cell network having decision means in which there are alternately arranged in an array state the image segmentation cells which transit over a non-excitation state, a self-excitable state and an excitation state in accordance with each pixel of the input image and coupling weight registers which hold the inter-cell coupling weights obtained by the coupling weight calculation circuit, the decision means deciding whether each cell is excitable or not based on values held in the coupling weight registers in which the cells are arranged adjacent to each other, the decision means putting, into the excitation state, the leader cell determined by the leader cell determination circuit and putting, into the excitation state, an excitable cell selected from the adjacent cells to expand an excitation region, thereby deciding the image segmentation region;

a segmentation region storage circuit which stores information of all the cells in the image segmentation region decided by the image segmentation cell network; and

an output image memory which stores the pixel value corresponding to each cell in an arbitrary image segmentation region based on contents stored in the segmentation region storage circuit,

wherein the coupling weight calculation circuit takes in the pixel values corresponding to the cells from the input image memory, to calculate the coupling weights between the adjacent cells in column direction by parallel processing and to calculate the coupling weights between the adjacent cells in row direction by pipeline processing.

5. The image segmentation apparatus according to claim 4, wherein:

the coupling weight calculation circuit is provided with an encoder which reduces a number of bits of a calculation result of the coupling weights.

6. The image segmentation apparatus according to claim 4, wherein:

each row of the image segmentation cell network is provided with a bus which transfers data to all the cells and the coupling weight registers.

7. The image segmentation apparatus according to claim 4, wherein:

the coupling weight register reduces a number of bits of the coupling weight data from the coupling weight calculation circuit to a permitted number of bits and stores the reduced data.

8. The image segmentation apparatus according to claim 4, wherein:

the cells of the image segmentation cell network carry out additions and subtractions required in a decision of the excitable state, serially at k number of adder/subtractors (k is an integer number smaller than the number of the adjacent cells) and at the corresponding registers.

9. The image segmentation apparatus according to claim 4, wherein:

a brightness value is used as the pixel value if the input image is a grayscale image.

10. The image segmentation apparatus according to claim 4, wherein:

color information is used as the pixel value if the input image is a color image.

11. An image segmentation apparatus which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region to selectively output an image of the arbitrary image segmentation region, said apparatus comprising:

an input image memory which stores pixel values of the input image;

a coupling weight calculation circuit which reads out the pixel values from the input image memory to calculate a coupling weight between each image segmentation cell corresponding to each pixel and an adjacent cell by pipeline processing;

a leader cell determination circuit which determines, based on the coupling weights calculated by the coupling weight calculation circuit, as a leader cell, the cell in which a total sum of the coupling weights with the adjacent cells is in excess of a reference value;

an image segmentation cell network having decision means in which there are alternately arranged in an array state the image segmentation cells which transit over a non-excitation state, a self-excitable state and an excitation state in accordance with each pixel of the input image and coupling weight registers which hold the inter-cell coupling weights obtained by the coupling weight calculation circuit, the decision means deciding whether each cell is excitable or not based on values held in the coupling weight registers in which the cells are arranged adjacent to each other, the decision means putting, into the excitation state, the leader cell determined by the leader cell determination circuit and putting, into the excitation state, an excitable cell selected from the adjacent cells to expand an excitation region, thereby deciding the image segmentation region;

a segmentation region storage circuit which stores information of all the cells in the image segmentation region decided by the image segmentation cell network; and

an output image memory which stores the pixel value corresponding to each cell in an arbitrary image segmentation region based on contents stored in the segmentation region storage circuit, wherein;

the leader cell determination circuit takes in the coupling weights corresponding to the cells from the coupling weight calculation circuit, to sequentially determine the leader cells by column-parallel pipeline processing.

12. An image segmentation apparatus which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region to selectively output an image of the arbitrary image segmentation region, said apparatus comprising:

an input image memory which stores pixel values of the input image;

a coupling weight calculation circuit which reads out the pixel values from the input image memory to calculate a coupling weight between each image segmentation cell corresponding to each pixel and an adjacent cell by pipeline processing;

a leader cell determination circuit which determines, based on the coupling weights calculated by the coupling weight calculation circuit, as a leader cell, the cell in which a total sum of the coupling weights with the adjacent cells is in excess of a reference value;

an image segmentation cell network having decision means in which there are alternately arranged in an array state the image segmentation cells which transit over a non-excitation state, a self-excitable state and an excitation state in accordance with each pixel of the input image and coupling weight registers which hold the inter-cell coupling weights obtained by the coupling weight calculation circuit, the decision means deciding whether each cell is excitable or not based on values held in the coupling weight registers in which the cells are arranged adjacent to each other, the decision means putting, into the excitation state, the leader cell determined by the leader cell determination circuit and putting, into the excitation state, an excitable cell selected from the adjacent cells to expand an excitation region, thereby deciding the image segmentation region;

a segmentation region storage circuit which stores information of all the cells in the image segmentation region decided by the image segmentation cell network; and

an output image memory which stores the pixel value corresponding to each cell in an arbitrary image segmentation region based on contents stored in the segmentation region storage circuit, wherein;

each row of the image segmentation cell network is provided with a shift register which transfers data to all the cells and the coupling weight registers.

13. An image segmentation apparatus which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region to selectively output an image of the arbitrary image segmentation region, said apparatus comprising:

an input image memory which stores pixel values of the input image;

a coupling weight calculation circuit which reads out the pixel values from the input image memory to calculate a coupling weight between each image segmentation cell corresponding to each pixel and an adjacent cell by pipeline processing;

a leader cell determination circuit which determines, based on the coupling weights calculated by the coupling weight calculation circuit, as a leader cell, the cell in which a total sum of the coupling weights with the adjacent cells is in excess of a reference value;

an image segmentation cell network having decision means in which there are alternately arranged in an array state the image segmentation cells which transit over a non-excitation state, a self-excitable state and an excitation state in accordance with each pixel of the input image and coupling weight registers which hold the inter-cell coupling weights obtained by the coupling weight calculation circuit, the decision means deciding whether each cell is excitable or not based on values held in the coupling weight registers in which the cells are arranged adjacent to each other, the decision means putting, into the excitation state, the leader cell determined by the leader cell determination circuit and putting, into the excitation state, an excitable cell selected from the adjacent cells to expand an excitation region, thereby deciding the image segmentation region;

a segmentation region storage circuit which stores information of all the cells in the image segmentation region decided by the image segmentation cell network; and

an output image memory which stores the pixel value corresponding to each cell in an arbitrary image segmentation region based on contents stored in the segmentation region storage circuit, wherein;

the image segmentation cell network is provided, as the coupling weight registers, with a vertical coupling weight register which stores the coupling weight between vertical and diagonal cells and a horizontal coupling weight register which stores the coupling weight between horizontal and diagonal cells, in which the vertical coupling weight registers and the horizontal coupling weight registers are alternately arranged between the cells so that the same coupling weight can be shared between the adjacent cells.

14. An image segmentation apparatus which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region to selectively output an image of the arbitrary image segmentation region, said apparatus comprising:

an input image memory which stores pixel values of the input image;

a coupling weight calculation circuit which reads out the pixel values from the input image memory to calculate a coupling weight between each image segmentation cell corresponding to each pixel and an adjacent cell by pipeline processing;

a leader cell determination circuit which determines, based on the coupling weights calculated by the coupling weight calculation circuit, as a leader cell, the cell in which a total sum of the coupling weights with the adjacent cells is in excess of a reference value;

an image segmentation cell network having decision means in which there are alternately arranged in an array state the image segmentation cells which transit over a non-excitation state, a self-excitable state and an excitation state in accordance with each pixel of the input image and coupling weight registers which hold the inter-cell coupling weights obtained by the coupling weight calculation circuit, the decision means deciding whether each cell is excitable or not based on values held in the coupling weight registers in which the cells are arranged adjacent to each other, the decision means putting, into the excitation state, the leader cell determined by the leader cell determination circuit and putting, into the excitation state, an excitable cell selected from the adjacent cells to expand an excitation region, thereby deciding the image segmentation region;

a segmentation region storage circuit which stores information of all the cells in the image segmentation region decided by the image segmentation cell network; and

an output image memory which stores the pixel value corresponding to each cell in an arbitrary image segmentation region based on contents stored in the segmentation region storage circuit, wherein;

the cells of the image segmentation cell network carry out additions and subtractions required in the decision of the excitable state, in parallel with each other as many adders, provided individually for each cell, as the number of the adjacent cells and one subtractor.

15. An image segmentation apparatus which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region to selectively output an image of the arbitrary image segmentation region, said apparatus comprising:

an input image memory which stores pixel values of the input image;

a coupling weight calculation circuit which reads out the pixel values from the input image memory to calculate a coupling weight between each image segmentation cell corresponding to each pixel and an adjacent cell by pipeline processing;

a leader cell determination circuit which determines, based on the coupling weights calculated by the coupling weight calculation circuit, as a leader cell, the cell in which a total sum of the coupling weights with the adjacent cells is in excess of a reference value;

an image segmentation cell network having decision means in which there are alternately arranged in an array state the image segmentation cells which transit over a non-excitation state, a self-excitable state and an excitation state in accordance with each pixel of the input image and coupling weight registers which hold the inter-cell coupling weights obtained by the coupling weight calculation circuit, the decision means deciding whether each cell is excitable or not based on values held in the coupling weight registers in which the cells are arranged adjacent to each other, the decision means putting, into the excitation state, the leader cell determined by the leader cell determination circuit and putting, into the excitation state, an excitable cell selected from the adjacent cells to expand an excitation region, thereby deciding the image segmentation region;

a segmentation region storage circuit which stores information of all the cells in the image segmentation region decided by the image segmentation cell network; and

an output image memory which stores the pixel value corresponding to each cell in an arbitrary image segmentation region based on contents stored in the segmentation region storage circuit, wherein;

the image segmentation cell network comprises;

self-excitation means for putting into the excitation state the leader cell determined by the leader cell determination circuit;

excitable cell detection means for detecting the excitable cell from the adjacent cells based on the coupling weights between the cell in the excitation state and the cells adjacent to the cell in the excitation state;

excitation means for putting into the excitation state the cells detected by the excitable cell detection means; and

inhibition means for putting into the inhibition state the cells in the excitation state in a case where no cell is detected by the excitable cell detection means;

wherein the image segmentation of one region is completed by performing repeated processing of the excitation means concurrently on the cells detected by the excitable cell detection means until no cell is detected anymore by the excitable cell detection means, so that the image segmentation of all the regions is completed by sequentially performing the processing of the respective self-excitation, excitable cell detection, excitation and inhibition means on the cells and leader cells in the non-excitation state and not in the inhibition state.

16. An image processing method including an image segmentation which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region, said method comprising:

a preparation step including,

an initialization step of putting, into a non-excitation state, a cell which is an individual image segmentation unit corresponding to pixels of the input image,

a taking step of pixel values of the pixels corresponding to the cell, and calculating each coupling weight between a plurality of adjacent cells, and

a determination step of determining leader cells based on each calculation result in the taking step;

a self-excitable cell detection step of selecting one of the leader cells determined by the determination step to detect the one leader cell as a self-excitable cell;

a self-excitation step of putting, into an excitation state, the self-excitable cell detected in the self-excitable cell detection step;

an excitable cell detection step of detecting an excitable cell from adjacent cells based on said coupling weights between cells in the excitation state including the leader cells and the adjacent cells;

an excitation step of putting, into an excitation state, the excitable cell detected in the excitable cell detection step; and

an inhibition step of putting, into an inhibition state, a cell in the excitation state if no cell is detected in the excitable cell detection step, wherein;

image segmentation of one region is completed by repeating the excitation step until no cell is detected any more in the excitable cell detection step; and

the image segmentation of all the regions is completed by repeating the respective steps until no leader cell in the non-excitation state is detected any more in the self-excitable cell detection step,

wherein I_i is a pixel value of the cell i (i indicates a cell number) which is an individual image segmentation unit corresponding to the pixels of the input image, x_i is a variable which indicates whether the cell i is in the excitation or non-excitation stat, p_i is a variable which indicates whether the self-excitation is permitted or not, W_ik is the coupling weight between the adjacent cells i and k, φ

_p is a threshold value which decides whether the self-excitation is permitted or not, φ

_z is a threshold value which decides whether the self-excitation is permitted or not, z_i is a variable which indicates whether the state of cell i is changed or not, and z is a variable of a global suppressor which decides whether are cells having changed to the excitation state based on a logical sum of the values z_i'"'"'s of all the cells;

in the preparation step, the variable x_i of the image segmentation cell i is set to 0 (x_i=0;

non-excitation), the pixel values I_i corresponding to the cells i are taken in to calculate the coupling weight W_ik between the adjacent plural cells i and k, so that if a total sum of calculation results is larger than the threshold value φ

_p, p_i=1 (self-excitable state) is set and if the total sum is equal to or smaller than the threshold value φ

_p, p_i=0 (non-self-excitable state) is set for initialization, and z=0 where the variable z is the global suppressor is set for initialization;

in each self-excitable cell detection step, at least one non-excited leader cell is selected from among the leader cells of p_i=1 determined in the determination step to detect the selected cells as the self-excitable cells;

in each self-excitation step, the variables x_i and z_i of the self-excitable cells i detected in the self-excitable cell detection step are set to 1 (x_i=1;

self-excitation, z_i=1;

changed state), respectively;

in the excitable cell detection step, if a total sum of the coupling weights W_ik between the excited cells k (x_k=1) adjacent to the cells i in the non-excitation state is larger than the threshold value φ

_z, the cells i are detected to be the excitable cells;

in each excitation step, the variables x_i and z_i of all the cells i detected in the excitable cell detection step are set to 1 (x_i=1;

excitation state, z_i=1;

changed state), respectively, and the variable z_i of the cells i already in the excitation state (x_i=1 and z_i=1) is set to 0 (z_i=0;

unchanged state); and

in each inhibition step, if no cell is detected in the excitable cell detection step, the cells of x_i=1 (excitation state) are set to x_i=0 (non-excitation state) and z_i=0 (unchanged state), and if p_i=1, the cells are set to p_i=0 (inhibition state).

17. An image processing apparatus integrated circuit apparatus for processing an image including an image segmentation which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region to selectively output an image of the arbitrary image segmentation region, said apparatus comprising:

an input image memory which stores pixel values of the input image;

a coupling weight calculation circuit which reads out the pixel values from the input image memory to calculate a coupling weight between each image segmentation cell corresponding to each pixel and an adjacent cell;

a leader cell determination circuit which determines, based on the coupling weights calculated by the coupling weight calculation circuit, as a leader cell, the cell in which a total sum of the coupling weights with the adjacent cells is in excess of a reference value;

an image segmentation cell network having decision means in which there are alternately arranged in an array state the image segmentation cells which transit over a non-excitation state, a self-excitable state and an excitation state in accordance with each pixel of the input image and coupling weight registers which hold the inter-cell coupling weights obtained by the coupling weight calculation circuit, the decision means deciding whether each cell is excitable or not based on values held in the coupling weight registers in which the cells are arranged adjacent to each other, the decision means putting, into the excitation state, the leader cell determined by the leader cell determination circuit and putting, into the excitation state, an excitable cell selected from the adjacent cells to expand an excitation region, thereby deciding the image segmentation region;

a segmentation region storage circuit which stores information of all the cells in the image segmentation region decided by the image segmentation cell network; and

an output image memory which stores the pixel value corresponding to each cell in an arbitrary image segmentation region based on contents stored in the segmentation region storage circuit,

wherein the coupling weight calculation circuit takes in the pixel values corresponding to the cells from the input image memory, to calculate the coupling weights between the adjacent cells in a column direction by parallel processing and to calculate the coupling weights between the adjacent cells in a row direction by pipeline processing.

18. An image segmentation apparatus which pinpoints one of regions from an input image belonging to a same category and identifies the one region as an image segmentation region to selectively output an image of the arbitrary image segmentation region, said apparatus comprising:

an input image memory which stores pixel values of the input image;

a coupling weight calculation circuit which reads out the pixel values from the input image memory to calculate a coupling weight between each image segmentation cell corresponding to each pixel and an adjacent cell;

a leader cell determination circuit which determines, based on the coupling weights calculated by the coupling weight calculation circuit, as a leader cell, the cell in which a total sum of the coupling weights with the adjacent cells is in excess of a reference value;

an image segmentation cell network having decision means in which there are alternately arranged in an array state the image segmentation cells which transit over a non-excitation state, a self-excitable state and an excitation state in accordance with each pixel of the input image and coupling weight registers which hold the inter-cell coupling weights obtained by the coupling weight calculation circuit, the decision means deciding whether each cell is excitable or not based on values held in the coupling weight registers in which the cells are arranged adjacent to each other, the decision means putting, into the excitation state, the leader cell determined by the leader cell determination circuit and putting, into the excitation state, an excitable cell selected from the adjacent cells to expand an excitation region, thereby deciding the image segmentation region;

a segmentation region storage circuit which stores information of all the cells in the image segmentation region decided by the image segmentation cell network; and

an output image memory which stores the pixel value corresponding to each cell in an arbitrary image segmentation region based on contents stored in the segmentation region storage circuit,

wherein the coupling weight calculation circuit takes in the pixel values corresponding to the cells from the input image memory, to calculate the coupling weights between the adjacent cells by parallel processing and to calculate the coupling weights between the adjacent cells in row direction by pipeline processing.

19. The image segmentation apparatus according to claim 18, wherein:

the coupling weight calculation circuit is provided with an encoder which reduces a number of bits of a calculation result of the coupling weights.

20. The image segmentation apparatus according to claim 18, wherein:

each row of the image segmentation cell network is provided with a bus which transfers data to all the cells and the coupling weight registers.

21. The image segmentation apparatus according to claim 18, wherein:

the coupling weight register reduces a number of bits of the coupling weight data from the coupling weight calculation circuit to a permitted number of bits and stores the reduced data.

22. The image segmentation apparatus according to claim 18, wherein:

the cells of the image segmentation cell network carry out additions and subtractions required in decision of the excitable state, serially at k number of adder/subtractors (k is an integer number smaller than the number of the adjacent cells) and at the corresponding registers.

23. The image segmentation apparatus according to claim 18, wherein:

a brightness value is used as the pixel value if the input image is a grayscale image.

24. The image segmentation apparatus according to claim 18, wherein:

color information is used as the pixel value if the input image is a color image.