Research On Image Color Recognition For Industrial Applications

With the application of machine vision technology increasing widespread,more and more application scenarios need for color recognition,and the application scenarios are increasingly complex.Color recognition has become a hotspot in machine vision based detection because of the increasing demand to identify and detect the color of object.However,the existing color recognition algorithms for industrial application have many limitations and the robustness of them is poor,so it is difficult to achieve efficient and accurate identification in industrial environments so far.In order to overcome the shortcomings of existing machine vision based color recognition algorithms,based on the discussions of the consistency between model representation and visual perception,this dissertation proposes a fast color recognition method based on HCL color space,and analyzes the applications of the methods including color matching and color segmentation in industrial environment.In this dissertation,we firstly analyzes the shortcomings of common color spaces when used in color image analysis,and points out the shortcoming of existing distance function while applied to HCL space,then we redefines the distance metric formula by improving the distance parameter and validates that utilizing the new distance function can get more reliable,more consistent results with the human visual perception by using the measure experiment of color similarity.And the new distance function can obtain the spatial uniformity of the HCL color space so we obtain the simple feature description of color feature extraction stage in HCL color space.In the process of color matching,the template matching algorithm of geometric feature region realizes the indirect localization of the color region of interest(ROI),which reduces the computational complexity of the localization of region of interest.In the process of color segmentation,this dissertation uses the histogram description without quantization of color interval to obtain stronger expression of color features,and uses the run-length analysis based contour extraction algorithm to acquire the connection relationship between inner and outer contours while contour analysis so as to improve the efficiency of the algorithm.The method proposed in this dissertation is implemented on Windows platform using Visual studio 2008 based on Open CV 1.0 and compared with the same functional module of Vision Pro.The experimental results show that the proposed method is not only simple and efficient,has strong color discrimination ability and good robustness,but also has low sensitivity to nonlinear illumination variations,which reduces the requirements of illumination conditions and extends the application of image color recognition.