Automatic line-shaped defect evaluation of solid-state imaging device

Toshio Asano,Nonmember Seiji Hata,Susumu Koishikawa,Nonmembers Youichi Shimizu,Seiji Hata, Nonmember,Youichi Shimizu, Nonmembers

DOI: https://doi.org/10.1002/scj.4690211008

1990-01-01

Systems and Computers in Japan

Abstract:Abstract A method to evaluate quantitatively the faint line‐shaped defect existing in the imaging signal of a solid‐state imaging device has been developed. Classifying line‐shaped defects into nondirectional ones for which the direction of the line defect cannot be specified and those of vertical direction, we investigate an evaluation algorithm for each of them and evaluate them by experiments. For nondirectional line defects, noticing that the line defects resemble straight line defects, line defects are detected within an image using Hough transformation. Next, computing the length and contrast of line defects as evaluation parameters, the degree of defect was evaluated. In the computation, efficient processing was attempted by making some additions to the processing method. As a result, detection and evaluation were achieved within a practically reasonable time. Taking into consideration vertical line defects and noticing that defect evaluations by human eye are different depending on frequency ranges of lines, defects of contrast in high‐frequency range using data interpolation and contrast in low‐frequency range using smoothing process were found and then the degree of defects by these contrasts was evaluated. Using the forementioned algorithm, a system was developed for automatic evaluation of line‐shaped defects for experiment and compared with the traditional evaluation by human eye. As a result, good agreement was obtained and the line‐shaped defects of a solid‐state imaging device were evaluated automatically.