Segmentation technique of an image

• US 20060018533A1
• Filed: 06/09/2005
• Published: 01/26/2006
• Est. Priority Date: 07/21/2004
• Status: Active Grant

• Current Assignee: SCREEN Holdings Co., Ltd.
• Original Assignee: SCREEN Holdings Co., Ltd.
• Inventors: Imamura, Atsushi
• Granted Patent: US 7,386,167 B2
• Time in Patent Office: Days
• US Class Current: 382/164
• CPC Class Codes: G06T 2207/30141 Printed circuit board [PCB]
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
• CPC Class Codes: G06V 10/56 relating to colour

Abstract

The present invention provides a technique to improve reliability of area segmentation result of an image into a plurality of image areas. A plurality of representative colors and a weight associated with each of the representative colors are determined. A weighted color difference index values associated with the representative colors are calculated according to the weights, respectively, with respect to each pixel constituting the image. The weighted color difference represents difference in color between an individual color representing color of each pixel constituting the image and each of the representative colors. Then, according to the weighted color difference index values, color of each pixel is classified into one of representative color regions associated with the representative colors, to perform segmentation of the image into the plurality of image areas.

Claims

1. A method for segmentation of an image into a plurality of image areas, the method comprising the steps of:

(a) determining a plurality of representative colors;

(b) determining a weight associated with each of the plurality of representative colors;

(c) calculating weighted color difference index values associated with the plurality of representative colors according to the weights, respectively, with respect to each pixel constituting the image, the weighted color difference representing difference in color between an individual color representing color of each pixel constituting the image and each of the plurality of representative colors; and

(d) performing segmentation of the image into the plurality of image areas according to the weighted color difference index values by classifying the individual color into one of a plurality of representative color regions associated with the plurality of representative colors.

2. A method according to claim 1 wherein the step (b) includes the steps of:

generating a histogram of number of pixels which constitute the image and whose individual colors are present on a line passing through at least two representative color regions among the plurality of representative color regions; and

adjusting the weight of at least one representative color associated with the at least two representative color regions by analyzing the histogram.

3. A method according to claim 1 further comprising the steps of

(e) determining whether result of segmentation into the image areas performed at the step (d) is appropriate;

and

(f) adjusting the weight of at least one representative color among the plurality of representative colors according to the result of segmentation when the result of segmentation performed at the step (d) is determined inappropriate.

4. A method according to claim 1 wherein the step (c) includes the steps of:

calculating an angle index value associated with an angle between an individual color vector representing the individual color and a representative color vector representing each of the plurality of individual colors in a predetermined color space;

calculating a distance index value associated with a distance between the individual color and each of the plurality of representative colors in the predetermined color space; and

weighting at least one of the distance index value and the angle index value associated with the same representative color in accordance with the weight, and calculating the weighted color difference index value for each of the plurality of representative colors based on the distance index value and the angle index value after the weighting, in relation to the individual color.

5. A method according to claim 4 wherein the step (b) includes the steps of:

generating a histogram of number of pixels which constitute the image and whose individual colors are present on a line passing through at least two representative color regions among the plurality of representative color regions; and

adjusting the weight of at least one representative color associated with the at least two representative color regions by analyzing the histogram.

6. A method according to claim 4 further comprising the steps of:

(e) determining whether result of segmentation into the image areas performed at the step (d) is appropriate; and

(f) adjusting the weight of at least one representative color among the plurality of representative colors according to the result of segmentation when the result of segmentation performed at the step (d) is determined inappropriate.

7. An apparatus for segmentation of an image into a plurality of image areas comprising:

a representative color determination unit configured to determine a plurality of representative colors;

a representative color weight determination unit configured to determine a weight associated with each of the plurality of representative colors;

a weighted color difference calculation unit configured to calculate weighted color difference index values associated with the plurality of representative colors according to the weights, respectively, with respect to each pixel constituting the image, the weighted color difference representing difference in color between an individual color representing color of each pixel constituting the image and each of the plurality of representative colors; and

an area segmentation unit configured to perform segmentation of the image into the plurality of image areas according to the weighted color difference index values by classifying the individual color into one of a plurality of representative color regions associated with the plurality of representative colors.

8. A method of inspecting a test object of which a defect appears on a test image captured by imaging the test object, the method comprising the steps of:

(a) acquiring the test image of the test object;

(b) dividing the test image into a plurality of image areas; and

(c) detecting the defect in accordance with a result of division of the test image into the plurality of image areas, wherein the step (b) includes the steps of

determining a plurality of representative colors;

determining a weight associated with each of the plurality of representative colors;

calculating weighted color difference index values associated with the plurality of representative colors according to the weights, respectively, with respect to each pixel constituting the test image, the weighted color difference representing difference in color between an individual color representing color of each pixel constituting the test image and each of the plurality of representative colors; and

performing segmentation of the test image into the plurality of image areas according to the weighted color difference index value by classifying the individual color into one of a plurality of representative color regions associated with the plurality of representative colors.

9. A apparatus of inspecting a test object of which a defect appears on a test image captured by imaging the test object, the apparatus comprising:

a test image acquisition unit configured to acquire the test image of the test object;

an image division unit configured to divide the test image into a plurality of image areas; and

a testing unit configured to detect the defect in accordance with a result of division of the test image into the plurality of image areas, wherein the image division unit includes;

a representative color determination unit configured to determine a plurality of representative colors;

a representative color weight determination unit configured to determine a weight associated with each of the plurality of representative colors;

a weighted color difference calculation unit configured to calculate weighted color difference index values associated with the plurality of representative colors according to the weights, respectively, with respect to each pixel constituting the test image, the weighted color difference representing difference in color between an individual color representing color of each pixel constituting the test image and each of the plurality of representative colors; and

an area segmentation unit configured to perform segmentation of the test image into the plurality of image areas according to the weighted color difference index values by classifying the individual color into one of a plurality of representative color regions associated with the plurality of representative colors.