The Research Of Tanning Leather Image Segmentation And Defect Detection

With the application of leather products, the requirement of leather grows day by day, but a lot of defects lie in the leather products unavoidably. Identifying these defects quickly and accurately is of great importance to improve the quality of leather products and the latter manufacture.Marking the defects by digital technique not only facilitates the automatic leather industry, but also helps to build the digital evaluation platform of the leather quality. This paper studies the classification detection of defects on the leather image based on the digital image processing technique.This paper has five aspects as follows:Firstly, we collect the tanning leather defect image by the precise equipment, and then classify the leather defects into four categories which make for the digital image processing. It overcomes the problem of low-level, low-efficiency and ambiguity in the past detecting methods.Secondly, for the real-time of defect detection, we propose a quick and simple binary of processing image in the stage of image preprocessing. After analyzing the characteristic of binary leather image, we propose a new de-noise method based on the airspace geometric area. This method derives from the idea of selecting structure element in the morphological de-noise, the threshold of area is calculated by the area in the binary leather image. The affection of noise is reduced efficiently by the method.Thirdly, we study the leather defects which are irregular in the stage of image segmentation. An improved method of color image segmentation based on the genetic algorithm is proposed for the obvious defects, and an improved method of texture image segmentation based on statistical gray-level feature is proposed for the unclear defects. The experimental results show the improved methods are correct and available.Fourthly, the classification detection and location are carried out to the leather defects. The corresponding image detection methods are proposed to the different kinds of the leather defects in this paper. A lot of experimental results show that our detection methods that can detect and locate the defects on the leather image more accurately can overcome the limitation of the traditional methods.Finally, the build of digital evaluation platform for the tanning leather is discussed. It's proved to be feasible. We prospect the research of work in the future.