The Key Technology Research For Fabric Defects Detection

In the textile industry, the problem of fabric defects detection is vital for the quality control of textiles. The traditional artificial defect detection method can no longer satisfy the demands of quality control of products in real-time automatic production, because it has the disadvantage of low detection rate, man-made influence and so on. So the fabric based on machine vision detection is the inevitable trend of textile production line to realize the real-time quality monitoring.This paper focuses on the online fabric defect detection technology related with principles and technology research. This article has three main research directions. The first one is on the research of the mature visual detection system, according to the analysis of fabric for the design of the hardware structure of visual inspection system. The second is research-based fabric defect detection algorithm, then, propose two kinds of automatic defect detection of unsupervised algorithm. The third is to integrate the hardware and software integration algorithm to build a fabric defect detection system based on machine vision. This paper’s main research work is as follows:Ⅰ. In-depth study of existing mature domestic fabric defect detection system platform, combined with the analysis of the demand of fabric defect detection of the visual system for the selection of hardware in the visual system. In real-time and high-speed textile inspection request to design a hardware structure based on distributed on-line machine vision detection system, namely using three cameras together to collect the fabric image. This paper also designs the high-power LED with automatic frequency flash source, high frequent, the continuous synchronization flash can cooperate with camera to realize the synchronization of the image acquisition.Ⅱ. On the basis of in-depth study of relevant principle and algorithm of fabric defect detection technology, this paper proposes two kinds of corresponding improved algorithm based on singular value decomposition.(1) One method is based on the improved2DFCM and SVD for defect detection. For localization of texture defects, we calculate features of each non-overlapping region of an image via the Singular Value Decomposition (SVD) and image processing techniques. In next step the algorithm uses the fuzzy c-means clustering (FCM) to classify each region into two clusters. The traditional FCM algorithm has high computational complexity, poor anti-noisy ability, while the proposed improved2DFCM algorithm with the introduction of two-dimensional histogram to get information of pixel space location to improve the anti-noise ability. At the same time, in order to improve the convergence speed of algorithm, we join inhibitory factor correction membership functions to improve the real-time performance of the algorithm.(2) The other method is based on periodic texture to the fabric defect detection-the improved method is based on the truncated singular value decomposition (TSVD). For correction of random noise, uneven light interference, we propose a method based on morphological level cap transformation to get the best contrast enhancement and illumination uniformity Image. At the same time, the number of decision method by threshold decision method for selecting method of improved normalized singular value dimension based adaptive based on truncated singular value. Through the experiment, decomposition method has better anti noise and experiment, segmentation effect than the original truncated singular value.III. The design of architecture of the software system for fabric defect detection is on the basis of the theory and algorithm. Each image processing function is seen as a weak coupling function module, and the software can be realized by the way of modularization. Finally, combine the hardware and software system which is set up with VC++and OpenCv programming techniques to realize the fabric defect detection system based on machine vision.