AUTOMATIC IMAGE SEGMENTATION SYSTEM USING MULTIPLE TEMPLATES AND A METHOD THEREFOR TO IMPROVE THE ACCURACY OF INSPECTION IMAGE SEGMENTATION

摘要

PURPOSE: An automatic image segmentation system using multiple templates and a method therefor are provided to improve the accuracy of inspection image segmentation by automatically segmenting images using NGC(Normalized Gray-scale Correction) and the multiple templates. CONSTITUTION: An image input module(10) acquires an inspection image, and transmits the acquired inspection image to a preprocessing module(20). The preprocessing module separates a meaning image from the inspection image transmitted from the image input module using a reference template module(30) and a pattern matching module(40). The reference template module generates a reference template model to be used for searching a pattern of the inspection image. The pattern matching module compares degree in which the pattern of the inspection image is identical to the reference module of the reference template module, and matches the pattern of the inspection image with the reference model of the reference template module. A storing module(50) stores a matching coefficient value and a matching point of each node of multiple templates acquired by the pattern matching module.