AUTOMATED TEXTURE DEFECT DETECTION USING THE NON-EXTENSIVE ENTROPY WITH GAUSSIAN GAIN

Abstract

Inefficient processes in industries can cost a lot of time, money and customer satisfaction. Quality assurance is a very important aspect for industries. As a result, to improve their processes and become more efficient at the global level, they have started to automate their certain tasks for which humans were generally considered to perform them. And one such popular task is Industrial Inspection . As a result, intelligent visual inspection systems are developed to ensure high quality of products in production lines. Advancement in image processing and computer vision techniques has led to the development of Automated Visual Inspection Systems. These systems are required in industries to inspect the manufactured products, so as to identify any discrepancies in them. Thus, they play an important role in industries by ensuring that only good quality products enter into the market because selling defective products in market can lead to large losses for goods manufacturing industry. One such industry is Textile Industry where these systems are of great importance. They identify defects occurring in textile cloth. Several algorithms have been developed based on different approaches so as to identify defects in texture patterns. Presence of complex defects is a major hurdle for many such algorithms and that is why new ideas keep on emerging to develop more efficient algorithms which can identify such type of defects in texture patterns. On the same lines, this major project report presents a new algorithm based on a new approach to identify complex defects in texture in a more efficient and accurate manner.