Method of extracting contour of a subject image from an original

• US 5,091,967 A
• Filed: 04/10/1989
• Issued: 02/25/1992
• Est. Priority Date: 04/08/1988
• Status: Expired due to Term

• Current Assignee: Dainippon Screen Manufacturing Company Limited (SCREEN Holdings Co., Ltd.)
• Original Assignee: Dainippon Screen Manufacturing Company Limited (SCREEN Holdings Co., Ltd.)
• Inventors: Ohsawa, Ikuo
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): COUSO, YON JUNG
• Application Number:
  US07/335,866
• Time in Patent Office: 1,051 Days
• Field of Search: 382/22, 382/23, 382/26, 382/50, 382/51, 382/55, 382/33, 358/96, 358/458, 358/464, 358/465, 358/466
• US Class Current: 382/172
• CPC Class Codes: G06T 2207/20092 Interactive image processing based on input by user
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation

Abstract

An image window (N) is intermittently moved along the contour (CT) of a subject image (I). At a respective paused position of the image window, a spacial differential of the density level is calculated for each pixel in the image window. A differential histogram is produced from the respective differentials, and a threshold level (Eth) is determined from the differential histogram and a designated percentile ratio. A contour image is then obtained by discriminating the respective differentials with the threshold level.

Claims

1. A method of generating contour image data expressing a contour of a subject image that is included in an original image, said original image being expressed by a plurality of pixels having respective gradation levels, the method comprising the steps of:

(a) displaying said original image on a display means;

(b) designating an area on said original image while observing said original image on said display means so that said contour belongs to said area;

(c) calculating respective spatial differentials of gradation levels with respect to pixels being included in said area;

(d) determining a threshold level for said respective spatial differentials, the step (d) including the substeps;

(d-1-1) generating a differential histogram which expresses a distribution of said respective spatial differentials over each of several partial areas of said area;

(d-1-2) designating criteria for dividing said differential histogram into two parts; and

(d-1-3) finding a boundary level at which said differential histogram is divided into said two parts, said threshold level being determined by said boundary level;

(e) comparing said respective spatial differentials with said threshold level, to thereby classify said pixels being included in said area into a first group of pixels having differentials larger than said threshold level and a second group of pixels having differentials smaller than said threshold level;

(f) extracting pixels belonging to said first group from said pixels being included in said area, to thereby specify extracted pixels; and

(g) generating contour image data expressing a contour image on which said extracted pixels represent said contour.

2. A method of claim 1, where the step (b) includes the steps of:

(b-1) providing an image window at a position which is so designated on said original image being displayed that a part of said contour appears in said image window; and

(b-2) repeating designation of said position while changing said position along said contour so that said image window moves along said contour, said area being defined by a trace of said image window.

3. A method of claim 2, wherein the step (c) includes the steps of:

(c-1) specifying a partial area of said original image which appears through said window;

(c-2) calculating respective spatial differentials of gradation levels with respect to pixels belonging to said partial area; and

(c-3) repeating the steps (c-1) and (c-2) every time said position is designated in the step (b-2), whereby said area is covered with a plurality of partial areas each of which is specified in step (c-1).

4. A method of claim 3, wherein the step includes the steps of:

(d-1) determining said threshold level for said respective spatial differentials obtained in the step (c-2); and

(d-2) repeating the step (d-1) every time the step (c-2) is repeated;

the step (e) includes a first repetition cycle for obtaining said first and second groups for each partial area;

the step of (f) includes a second repetition for obtaining said extracted pixels for each partial area; and

the step of (g) includes the step of;

(g-1) connecting said extracted pixels for said plurality of said partial areas with each other in an image memory, to thereby generate said contour image data in said image memory.

5. The method of claim 1, wherein:

said original image is a color image being separated into color-component images whose number is M, where M is an integer larger than one;

said method further comprises the steps of;

(h) designating an integer T which satisfies the condition 2≦

T≦

M;

(i) selecting one of said color-component images as a selected component according to degree of clarity of said contour appearing on said color-component images;

(j) conducting the steps (d) through (f) for said selected component, to obtain said extracted pixels from said selected component; and

(k) repeating the steps (i) and (j) while updating said selected component T-times within said color-component images, to obtain T-sets of said extracted pixels; and

the step (g) includes the step of;

(g-1) logically summing data associated with said T-sets of said extracted pixels with each other in an image memory, to thereby generate said contour image data in said image memory.

6. A method of claim 5, wherein the step (e) includes the steps of:

(e-1) giving a first logical level to pixels belonging to said first group, whereby said extracted pixels are provided with said first logical level; and

(e-2) giving a second logical level to pixels belonging to said second group; and

the step (g-1) includes the steps of;

(g-1-1) with respect to each selected component, defining a logical image on which pixels corresponding to said extracted pixels have said first logical level; and

(g-1-2) taking logical summation between respective logical images obtained in the step (g-1-1), to thereby obtain said contour image data.

7. The method of claim 1, wherein:

said original image is a color image which is being separated into a plurality of color-component images; and

said method further comprises the steps of;

(h) selecting one of said plurality of said color-component images having a highest clarity of said contour appearing on said plurality of color-component images; and

(i) conducting the steps (d) through (f) for said one of said plurality of said color-component images to obtain said extracted pixels thereof.

8. A method of generating contour image data expressing a contour of a subject image that is included in an original image, said original image being expressed by a plurality of pixels having respective gradation levels, the method comprising the steps of:

(a) a displaying said original image on a display means;

(b) designating an area on said original image while observing said original image on said display means so that said contour belongs to said area, the step (b) including the substeps;

(b-1) providing an image window at a position which is so designated on said original image being displayed that a part of said contour appears in said image window; and

(b-2) repeating the designation of said position while changing said position along said contour so that said image window moves along said contour, said area being defined by a trace of said image window;

(c) calculating respective spatial differentials of gradation levels with respect to pixels being included in said area, the step (c) including the substeps;

(c-1) specifying a partial area of said original image which appears through said window,

(c-2) calculating respective spatial differentials of gradation levels with respect to pixels belonging to said partial area, and

(c-3) repeating the steps (c-1) and (c-2) every time said position is designated in step (b-2), whereby said area is covered with a plurality of partial areas each of which is specified in step (c-1);

(d) determining a threshold level for said respective spatial differentials, the step (d) including the substeps;

(d-1) determining said threshold level for said respective spatial differentials obtained in step (c-2), and

(d-2) repeating the step (d-1) every time the step (c-2) is repeated, the step (d-1) including the substeps;

(d-1-1) generating a differential histogram which expresses a distribution of said respective spatial differential,

(d-1-2) designating a criteria for dividing said differential histogram into two parts, and

(d-1-3) finding a boundary level at which said differential histogram is divided into said two parts, said threshold level being determined by said boundary level;

(e) comparing said respective spatial differentials with said threshold level, to thereby classify said pixels being included in said area into a first group of pixels having differentials larger than said threshold level and a second group of pixels having differentials smaller than said threshold level, the step (e) including a first repetition cycle for obtaining said first and second groups for each partial area;

(f) extracting pixels belonging to said first group from said pixels being included in said area, to thereby specify extracted pixels, the step (f) including a second repetition for obtaining said extracted pixels for each partial area; and

(g) generating contour image data expressing a contour image on which said extracted pixels represent said contour, the step (g) including the substep;

(g-1) connecting said extracted pixels for said plurality of said partial areas with each other in an image memory, to thereby generate said contour image data in said image memory.

9. A method of claim 8, wherein said criteria is expressed by a variable percentile ratio for said differential histogram.

10. A method of claim 9, wherein the step (e) is attained through the step of:

(E-1) binarizing said respective spatial differentials with respect to said pixels included in said area, whereby a first logical level is given to pixels belonging to said first group and a second logical level is given to pixels belonging to said second group, said extracted pixels being provided with said first logical level.

11. A method of claim 10, wherein the step (g-1) includes the steps of:

(g-1-1) with respect to each partial area, defining a logical image on which pixels corresponding to said extracted pixels have said first logical level; and

(g-1-2) taking logical summation between respective logical images obtained in the step (g-1-1), to thereby obtain said contour image data.

12. A method of generating contour image data expressing a contour of a subject image which is included in an original image, said original image being expressed by a plurality of pixels having respective gradation levels, said method of comprising the steps of:

(a) displaying said original image on a display means;

(b) designating an area on said original image while observing said original image on said display means so that said contour belongs to said area,

(c) calculating respective spatial differentials of gradation levels with respect to pixels being included in said area,

(d) determining a threshold level for said respective spatial differentials,

(e) comparing said respective spatial differentials with said threshold level, to thereby classify said pixels being included in said area into a first group of pixels having differentials larger than said threshold level and a second group of pixels having differentials smaller than said threshold level, step (e) including the substeps;

(e-1) giving a first logical level to pixels belonging to said first group, whereby said extracted pixels are provided with said first logical level; and

(e-2) giving a second logical level to pixels belonging to said second group;

(f) extracting pixels belonging to said first group from said pixels being included in said area, to thereby specify extracted pixels;

(g) generating contour image data expressing a contour image on which said extracted pixels represent said contour;

said original image being a color image and being separated into color-component images whose number is M, where M is an integer larger than one, said method further including;

(h) designating an integer T which satisfies the condition 2≦

T≦

M,

(i) selecting one of said color-component images as a selected component according to degree of clarity of said contour appearing on said color component images, the step of (i) including the substeps;

(i-1) for each color-component image, calculating a statistical value which expresses a degree of statistical split in distribution of said respective spatial differentials,

(i-2) comparing respective statistical values obtained in the step (i-1) with each other, to thereby find a maximum value within said respective statistical values, and

(i-3) selecting one of said color-component images corresponding to said maximum value, to thereby specify said selected component,

(j) conducting the steps (d) through (f) for said selected component, to obtain said extracted pixels from said selected component, and

(k) repeating the steps (i) through (j) while updating said selected component T-times within said color component images, to obtain T-sets of said extracted pixels,

the step (k) includes the step of;

(k-1) repeating the steps (i-2) and (i-3) T-times, where a color component image having been selected in a previous repetition of step (i-3) is excluded from selection in following repetition of step (i-3);

the step (g) including the substeps;

(g-1) logically summing data associated with said T-sets of said extracted pixels with each other in an image memory, to thereby generate said contour image data in said image memory;

(g-1-1) with respect to each selected component, defining a logical image on which pixels corresponding to said extracted pixels have said first logical level, and

(g-1-2) taking logical summation between respective logical images obtained in the step (g-1-1), to thereby obtain said contour image.

13. A method of claim 12, wherein:

said area is covered with a plurality of partial areas each of which is designated with an image window provided on said original image being displayed; and

the steps (h) through (k) are conducted for each partial area, whereby said integer T is variable for each partial area.

14. A method of claim 13, further comprising the steps of:

(l) designating initial size of said image window; and

(m) changing size of said image window for a designated partial area.

15. A method of claim 14, wherein:

the step (m) includes the step of;

(m) changing the size of said image window according to curvature of said contour on said designated partial area.

16. A method of generating contour image data expressing a contour of a subject image that is included in an original image, said original image being expressed by a plurality of pixels having respective gradation levels, the method comprising the steps of:

(a) calculating respective spatial differentials of gradation levels with respect to pixels on said original image;

(b) comparing said respective spatial differentials with a threshold level to binarize said pixels on said original image, whereby a differential contour image expressing said contour is obtained;

(c) generating a ring image covering said differential contour image with a margin;

(d) dividing said ring image into an inner ring image and an outer ring image with said differential contour image, said inner and outer ring images being separated from each other by a gap corresponding to said differential contour image;

(e) deleting one of said inner and outer ring images, to thereby generate a remaining ring image;

(f) combining said remaining ring image with said differential contour image to generate a composite image in the shape of a ring;

(g) generating a loop image which expresses a loop extending along an edge of said composite image; and

(h) generating said contour image data in accordance with said loop image.

17. A method of claim 16, wherein the step (g) includes the step of:

(g-1) thinning said composite image by deleting one of inner and outer edge portions of said composite image, to obtain a thinned image having the shape of a ring;

(g-2) generating a first edge image which expresses inner and outer edges of said thinned image;

(g-3) generating a second edge image which expresses inner and outer edges of said ring image; and

(g-4) deleting a part of said first edge image at which said second edge image overlaps with said first edge image, to thereby obtain said loop image.

18. A method of claim 17, wherein said method further comprises the steps of:

(i) displaying said original image on a display means;

(j) designing a point at a neighborhood of said contour included in said original image being displayed;

(k) generating first and second image windows around said point on said display means, in which said second image window has a size larger than a size of said first image window and said first image window is surrounded by said second window; and

(l) moving said point along said contour so that said first and second image windows follow movement of said point, whereby first and second rings are defined by respective traces of said first and second image windows, respectively;

said pixels on said original image being subjected to the step (a) are pixels belonging to said first ring; and

said ring image which is generated in the step (c) is an image expressing said second ring.

19. A method of claim 18, wherein the step (g-1) includes the steps of:

(g-1-1) detecting said inner and outer edge portions of said composite image; and

(g-1-2) deleting one of said inner and outer edge portions which overlaps with said first image, to obtain said thinned image.

20. A method of claim 19, wherein the step (1) includes the steps of:

(1-1) pausing movement of said point at a position where a part of said contour appears in said first window;

(1-2) conducting the steps (a) and (b) for pixels within said first image window being paused at said position by the step (1-1), whereby a part of said differential contour image is obtained;

(1-3) repeating the steps (1-1) and (1-2) while moving said point along said contour;

whereby respective parts of said differential contour image are obtained; and

(1-4) connecting said respective parts of said differential contour image with each other, to thereby obtain the whole of said differential contour image.

21. A method of claim 20, further comprising the step of:

(m) displaying said differential contour image and said ring image on said display means while overlapping said differential contour image and said ring image with said original image on said display means.

22. A method of claim 21, wherein said differential contour image and said ring image are displayed on said display means in semitransparent colors.

23. A method of claim 22, wherein:

said differential contour image has pin holes;

said reverse image of said differential contour image has reversed pin holes; and

said reversed pin holes are compensated with said pin holes through the step (f).

24. A method of claim 16, wherein the step (g) includes the steps of:

(g-1) thinning said composite image by deleting one of inner and outer edge portions of said composite image, to obtain a thinned image having the shape of a ring; and

(g-2) tracing an edge of said thinned image which is not deleted in the step (g-1), to obtain said loop image.