A COMPUTER-AIDED DIAGNOSIS SYSTEM ARCHITECTURE FOR SEMICONDUCTOR DEFECT FACTOR USING REGION-BASED IMAGE RETRIEVAL TECHNIQUES

摘要

Many efforts have been made to detect and segment semiconductor defect. The defect itself, however, does not provide any information on defect factor. In this paper, a computer-aided diagnosis system architecure for semiconductor defect factor using region-based image retrieval techniques was developed. A sufficient number of sample of 16x16-pixel, 32x32-pixel blocks were taken from 2507 semiconductor images which include various defects. Retrieving similar images, the region of interest in the query image was divided into two sizes of block and obtained feature vector. To measure similarity, region of interest of query image was compared to the corressponding block of images in database. In experiements, average 5.1 out of 10 retrieved images were relevant to the query image. The developed system can be used to assist user to diagnose semiconductor defect factor in real world.