Fabric Defect Detection Adopting Combined GLCM, Gabor Wavelet Features and Random Decision Forest

Abstract

In image analysis and pattern recognition activity, one of the most salient characteristics is texture. The global region of images in spatial domain has an enhanced processing effect with the help of co-occurrence matrix and in the frequency domain for the admirable performance such as multi-scale, multi-direction local information is obtained from Gabor wavelet. The consolidation of gray-level co-occurrence matrix and Gabor wavelet is utilized to fabric image feature texture eradication. In classification phase, random decision forest (RDFs) Classifier is applied to classify the input fabric image into defective or non-defective. RDFs are a novel and outfit machine learning strategy which fuses the element choice. Nevertheless, RDFs exhibit a lot of advantages when compared with other modeling approaches within the category. The main advantages are, RDFs can handle both the continuous and discrete variables, RDFs does not overfit as a classifier, and run quick and productively when taking care of expansive datasets. In this paper the consolidation of gray-level co-occurrence matrix (GLCM) and Gabor wavelet is utilized to fabric image feature texture eradication. In classification phase, random decision forest (RDFs) classifier is applied to classify the input fabric image into defective or non-defective.