Socks Color Vision Classification System Based On Computer Vision Research

Polyamide yarns is the main raw material of the textile processing. Dye uniformity of the silk is a significant index for chinlon physiochemical performance. In the process of the nylon yarn production, dyeing is the last procedure. Numerous low quality of silk is generated in the case of asymmetry drying. However, current distempered evaluating method can easily lead to misjudge creating a large number of low quality silk. It is not only effect product quality but also cause a financial lose for enterprise. it is an important problem in textile engineering that using Computer Vision Technology, Pattern Recognition Technology and Image Processing Technology how to design grading stocking color expert system to improve the efficiency of producing control according to the character of stocking image.Proposed using quotient image to pretreat the sample images. It is an effective illumination normalization method. Then color features of each sample are gained via adopting hue percentage histogram of HSV color space algorithm. These features are regarded as input vectors for expert system. Finally, both the interval and the result of the color level of the stockings are automatically created by means of expert system. On this condition, the Windows-based programming is completed and applied to the "Chemical fiber industry technological upgrading—technical innovation of Tex fibre" project. To sum up, the main content of this thesis as follows:1. The grading expert system of stockings'color has been developed, which integrates the image process and analysis program with the image-capturing device. The entire system is consisted of hardware section and software section. The hardware section includes stocking image acquisition device; the software section is based on object-oriented programming ideas, using Visual C++6.0 programming environment and Matlab software to complete the calculation.2. The application software has been programmed. The chief program has a friendly user interface, i.e., capturing and reading of image, pretreatment extraction of feature parameters and reasoning machine, with which stockings'image can be captured, analyzed and processed conveniently.3. The experiment has been carried out on the grading expert system developed in this thesis by selecting 3 groups stocking image with the total number of 1500. Experiments show that the error between the experiment results and the actual conditions is 1.5%. This error are far less than the permit error of production and technical indication. this method can accurately and rapidly obtain the results. It reflects the practical situation of the color of the stocking and meets the request of industrial production. The research results indicate that this method is feasible.