群化要因に基づくGAを用いたLCD輝度むら境界線検出

抄録

An area of uneven brightness is a typical defect which influences the screen quality of a LCD display. This paper describes a method to detect an uneven area on a LCD using a genetic algorithm based on perceptive grouping factors in Gestalt psychology. Because the contrast of an uneven area is very low, the boundary of the uneven area is often discontinuous and many noises occur to surroundings. To extract a visual continuous boundary of the uneven area in such an image, a group of points is recognized as a curve by evaluating proximity and continuity in relative positions of points. Candidates for points to compose a visual curve are extracted effectively using a genetic algorithm. The experimental results show that extracted boundaries of uneven areas on LCDs by the proposed method were similar to forms of boundaries perceived by the visual sense.