Defect Detection in Texture using Statistical Approach and Principal Component Analysis

Main Article Content

Punamchand M. Mahajan
Dr. Satish R. Kolhe, Dr. Pradeep M. Patil


In this paper a system to detect and locate the faults in the fabric texture is proposed. In the proposed method, the image is divided into
four non-overlapping samples and then locally invariant to similarity features are extracted from each of the sample using statistical analysis.
The image is further divided till the statistical features extracted are at least 75% same from all the samples. Principal component analysis (PCA)
is used to receive feature vector describing each samples. Fuzzy C-means clustering (FCM) is used to classify all the samples in to two clusters,
defective and non defective that are defined using threshold value selected empirically. The validity test on the developed algorithm has been
performed with some fabric defect images. Experimental results show that the proposed method can detect fabric defect correctly.


Keywords: Industrial automation, Statistical feature extraction, principal component analysis, Histogram data interpretation, Fuzzy C-means


Download data is not yet available.

Article Details