Study On Leather Color Measuring And Matching Based On Computer Vision

With the development of the national economy and improvement of people’s living standard, leather products has been applied in people’s lives more and more widely, and also has been played an irreplaceable role. Higher quality, more efficiency and other requirements have been demanded on the leather products. Color is one of main factors influencing the quality of leather products. The eyes have been used to identify the leather surface color in the traditional manual measurement and matching method. Multiple times of color matching, proofing and color contrasts has been needed to finish once successful leather color matching. The production efficiency and the quality of leather product has been affected by this method which only by experience subjective. The development of leather industry has been also restricted by this method. Especially the high-end market demand is unable to meet. The method of color consistency detection algorithm and leather surface color extraction have been studied in this paper which makes use of the computer vision technology, combined with the color characteristics of the leather surface. An experiment environment of leather image acquisition based on this topic has been designed for the system. A set of leather color measuring and matching system based on the computer vision has been given in this paper. The main work of this paper is summarized as follows:(1)Two kinds of methods, gray histogram and gray interval frequency statistics, have been studied to judge the consistency of leather surface color. Those two methods have been used to judge a large number of leather images, the result has been used to compare with human visual recognition results. The results show that the continuous gray interval frequency statistics method could more accurately determine whether the leather surface color is consistent.(2)An experiment environment has been designed to meet the image acquisition of computer vision technology in leather color measuring and matching system. The filament lamp has been used as the light source which has the same continuous spectral distribution with the sunlight, the picture of the samples, dyes and each to be colored leather has been extracted in the same closed and with no light reflection conditions, the results influenced by light source, relative position or other factors has been reduced by the way of extracting the same position of each image color feature.(3)A set of leather color measuring and matching system based on computer vision has been developed in this paper. Whether the non-consistency region existed in leather surface or not has been determined by the consistency test results. The color value has been extracted from the consistency leather image surface, then the color space is conversed and the color scheme is calculated. The color scheme has been used to treat color proofing leather, the chromatic difference has been computed between the proofing results with the standard sample, whether the leather matching results meeting the requirements of production or not has been judged from the color difference. The color scheme of the unsatisfied production requirements has been modified by the color difference analysis.(4)The user interface of leather color measuring and matching system based on computer vision has been developed in this paper, the leather surface color consistency detection procedures, the color extraction procedure, calculation of the scheme and the color difference evaluation results have been given in this system. This system has been developed based on MFC visual application in the Visual Studio 2010 development platform and C++ language. The information of the tested images,images’ color RGB value, consistency detection results, matching proportion and color difference evaluation result has been displayed in real time.The whole set of leather measuring and matching system based on the computer vision has passed through the verify of each part, by the determination results of standard color pictures’ color value, the color value extraction part’s accuracy has also been verified. Contrasting the result of multiple images of color consistency and non-consistency, it has the same results with the vision. Based on the method in this paper, the calculation of the leather color proportion and difference evaluation results show that those have basically reached the expected goal.