Efficient image segmentation algorithm for defect inspection of optical films

ZHONG Qiusheng1，2 ， LI Jingrong1 ， HU Guanghua1，2
（1.School of Mechanical & Automotive Engineering， South China University of Technology，Guangzhou
510640, China;2.HuaZhuo Chemicals Co.Ltd. ,Guangzhou 511458,China）

Abstract：In order to improve the precision of inspection， it’s quite significant to provide an appropriate threshold for the segmentation of images with low gray level differences, based on the error correction theory and background subtraction algorithm， a novel image segmentation algorithm was proposed. The algorithm converts defects inspection to gross errors estimation according to the departure degree of residual error from standard deviation， and works well even in the case of considerable inevitable random errors existing in the images. Moreover， since subtraction operation and updating of background image need only to be done once， the algorithm is quite efficient and each pixel will be assigned an independent segmentation threshold automatically. The experimental results indicate that the proposed algorithm can extract the defect region rapidly and completely.
Key words：error correction; optical film; defect inspection; image segmentation