A Novel Method for Feature Extraction and Automatic Recognition of Tire Defects Using Independent Component Analysis

Abstract

In order to extract the features of tire defects in X-ray images, Linear transformation is often performed on the images. However, Gabor and wavelet transformation are predefined and unchangeable, and their basic functions can’t be adapted to the characteristics of defect images. So we propose a new approach using Independent Component Analysis (ICA) and Topographic Independent Component Analysis (TICA) reconstruction algorithm to extract the features of tire defects and apply automatic recognition of tire defects. First, the basis functions and filters which adapt to the characteristics of defect images are estimated adaptively using ICA and TICA from the tire defect library. Then, the tire defect images are filtered to extract features. Finally, the samples are classified by a nonlinear support vector machine (SVM). Experimental results show that the proposed algorithm has very high recognition rates for shape defects, texture defects and other defects. Its total recognition rate can be as high as 95.76%.