Fabric Defect Detection Based On Local Binary Patterns

As a simple and effective texture operator,local binary pattern (LBP) has been widely and deeply studied in recent years.It also has been widely used in many fields.In this paper,we focus on the local binary patterns in the fields of cloth defect detection.Firstly,the shortcomings of LBP and its extensions used in cloth defect detection are analysed in detail.Then,the improved schemes are presented in the paper.The main contents are as follows:(1)Based on the analysis of fabric defect detection,the LBP and its extensions are summarized from four aspects,texture classification,face recognition,image retrieval and image detection.Then,the procedure of defect detection by LBP is introduced in detail.(2)In order to reduce the dimensionality and noise influence,the operator CS-LBPV (Center-Symmetric Local Binary Patterns Variance) based on arithmetic mean filter is proposed.In comparison to the basic LBP,the proposed method can effectively reduce the noise influence by the arithmetic mean filter.Further,the local contrast is introduced to improve the performance of the scheme,which can distinguish the defective region and the normal texture region effectively.(3)For the LBP and LBPV (LBP variance),they are sensietive to noise and not effective for linear defects and the regions with small gray changes.Therefore,a detection algorithm,called MBLBPV (multi-scale block local binary patterns variance) is proposed in the paper.In order to reduce the noise influence,it uses the gray mean value of the appropriate sub-region to replace that of a single pixel.In addition,the local variance is introduced as the weight for coding to denote the difference between the defective and the normal texture.(4)A fabric defect detection algorithm based on enhanced local gradient patterns (ELGP) and morphology is proposed.Firstly,the absolute value of the difference between the global gradient mean and the local gradient mean is used as the threshold to extract image feature.In order to highlight the defective region and improve the defect detection rate,the morphological operation is introduced to reduce the noise and fill the defective holes.