Image analysis – Pattern recognition – Feature extraction
Reexamination Certificate
1999-04-15
2001-10-16
Johns, Andrew W. (Department: 2621)
Image analysis
Pattern recognition
Feature extraction
Reexamination Certificate
active
06304672
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an edge detecting method and an edge detecting device and conducts the edge detecting processing which plays a basic role especially on the image processing, and can be suitably applied to such as the special effect processing in the video production of television and motion pictures and the element recognition processing from the camera image in the FA (factory automation).
Description of the Related Art
Heretofore, the edge detection has been a process to find the part where the picture element value is rapidly changing between dark and light images. And since normally sudden changes occur on the outline of the object, the contour of the object can be extracted from the pictorial image depending on the result of edge detection. Therefore, the edge detection has been widely utilized as the most basic processing for obtaining information from the image on the object existing in that image.
In the color images, since dark and light image exist in every primary color (e.g., red, green, and blue), heretofore the edge detection has been conducted on each primary color and if the edge is detected on any color, that picture element is regarded as an edge of the color image. This will be explained referring to the image shown in FIG.
1
A. This image is a round object shown in
FIG. 1B
which is placed in front of the background having a horizontal line shown in FIG.
1
C.
In the case of this image, the dark and light image of three primary colors, red, green, and blue become like
FIGS. 2A
,
2
B, and
2
C respectively. For example, in the case of red, the picture element value of the inside of area
1
shown by oblique lines is “1.0” and the picture element value of the inside of the remaining area
2
is “0”. The same will be applied to green and blue. If the edge is detected on each dark and light image, since the edge on the part where the sudden change occurs, i.e., only the place where the picture element value changes from “1.0” to “0” can be obtained as the edge, edges like
FIGS. 3A
,
3
B, and
3
C corresponding to
FIGS. 2A
,
2
B, and
2
C will be obtained. If the edge is detected in any one of the colors, red, green, or blue, its picture element is considered as an edge, and the edge which can be ultimately obtained will be like
FIG. 3D
by combining these edges.
However, since the edge obtained as described above contains a horizontal line
4
in the background and a vertical line
5
of the inside of a circle besides a contour
3
of the circle to be aimed originally, recognition of the real contour by means of the computer becomes very difficult. Like this example, the problem that unnecessary parts are detected as edges occurs in the case where there are patterns or colors and changes in lightness exist in the object and background.
Since it is common that there exist patterns and colors and changes in lightness on the object to be aimed for extraction from the image background or from within an image, it is practically impossible to extract the contour from the optional image by means of the conventional device, and in practice the contour extraction by the computer has been conducted only under limited condition, such as in the case where the background is painted with a uniform color.
Various improvements have heretofore been developed to alleviate these problems based on experience. For example, when the edge was detected in either primary color, that picture element was not only regarded as an edge but the method has been modified taking great account of lightness changes. However, since the application of dark and light image of each primary color is fixed in advance, these cannot be applied to images which did not follow the law of experience, and they were insufficient for practical use as the edge detection accuracy.
SUMMARY OF THE INVENTION
In view of the foregoing, an object of the present invention is to provide an edge detecting method and edge detecting device capable of improving the accuracy of edge detection and extracting the contour correctly in the general image.
The foregoing object and other objects of this invention have been achieved by the provision of an edge detecting method for detecting a picture element group changing rapidly as compared with its surroundings as an edge E(x,y) from within image data in which each picture element is composed of N numbers of independent dark and light data R(x,y), G(x,y), B(x,y) respectively. This is accomplished by a method comprising a coefficient calculating step SP
1
for calculating N sets of coefficients Wr(x,y), Wg(x,y), Wb(x,y) corresponding to N numbers of dark and light data R(x,y), G(x,y), B(x,y) respectively, and judging steps SP
2
, SP
3
, SP
4
for judging whether each picture element is the edge E(x,y) or not based on N numbers of dark and light data R(x,y), G(x,y), B(x,y) and N sets of coefficients Wr(x,y), Wg(x,y), Wb(x,y).
Furthermore, according to the present invention, an edge detecting device
10
for detecting a picture element group changing rapidly as compared with its surroundings from within image data in which each picture element is composed of N numbers of independent dark and light data
11
,
12
,
13
respectively is comprised of a coefficient calculating means
14
for calculating N sets of coefficients Wr, Wg, Wb corresponding to N numbers of dark and light data
11
,
12
,
13
respectively, and a judging means
16
for judging whether each picture element is the edge or not based on N numbers of dark and light data
11
,
12
,
13
and N sets of coefficients Wr, Wg, Wb.
In the case of detecting the group of picture elements which is changing rapidly as compared with its surroundings as the edge E(x,y) from within image data in which picture elements is composed of N numbers of independent dark and light data R(x,y), G(x,y), B(x,y), since N sets of coefficients Wr(x,y), Wg(x,y), Wb(x,y) corresponding to N numbers of dark and light data R(x,y), G(x,y), B(x,y) respectively will be calculated and each picture element will be judged as to whether it is the edge or not based on N numbers of dark and light data R(x,y), G(x,y), B(x,y) and N sets of coefficients Wr(x,y), Wg(x,y), Wb(x,y), the edge detection accuracy can be improved and the contour can be correctly extracted even in the general image.
The nature, principle and utility of the invention will become more apparent from the following detailed description when read in conjunction with the accompanying drawings in which like parts are designated by like reference numerals or characters.
REFERENCES:
patent: 5038223 (1991-08-01), Yamada
patent: 5267031 (1993-11-01), Katoh et al.
patent: 5341142 (1994-08-01), Reis et al.
patent: 5574800 (1996-11-01), Inoue et al.
patent: 5602942 (1997-02-01), Takemoto et al.
patent: 5708632 (1998-01-01), Totsuka et al.
patent: 5887082 (1999-03-01), Mitsunaga et al.
patent: 5892853 (1999-04-01), Hirani et al.
Mitsunaga Tomoo
Totsuka Takashi
Azarian Seyed
Bell Boyd & Lloyd LLC
Johns Andrew W.
Sony Corporation
LandOfFree
Edge detecting method and edge detecting device which... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Edge detecting method and edge detecting device which..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Edge detecting method and edge detecting device which... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2556867