Frequency Domain Improvements to Texture Discrimination Algorithms

As the production speeds of factories increase, it becomes more and challenging to inspect products in real time. The goal this article is come up with a computationally efficient texture discrimination algorithm by first testing their ability localize defects then increase efficiency removing less effective parts them. Therefore, abilities most popular classification algorithms such as GLCM, LBP SDH are tested on different datasets. These tests reveal that, small windows GLCM perform better. Frequency properties textures used fine-tune parameters these algorithms. Further experiments three datasets prove that accuracy increased almost twice while decreasing processing time considerably.