Defect Detection in Pattern Texture Analysis Based on Kernel Selection in Support Vector Machine

Background/Objective: Finding defects in real world application is assorted process. A robust and novel method is designed to select fine distinctions of features and classifying the images lead to improve the quality of products in industrial engineering. Methods/Statistical Analysis: Image feature accentuate, feature selection and classification are the different stages in pattern texture analysis. The efficiency of the overall system depends on efficiency of individual stages. Findings: Computational complexity of kernel algorithms are more intelligent than features .We analyzed and reviewed linear kernel, Quadratic Kernel, Polynomial Kernel, Sigmoid Kernel of SVM to classify the patterns effectively for classifying the defects. Improvements/Applications: Here kernel functions such as the polynomial kernel functions are yield superb performance ratios.