Solid Texture Analysis Of Workpiece Surface Based On Computer Vision

With the research in depth of computer cognitive science and artificial intelligence, the branch of computer vision and image processing have developed rapidly. Texture, as a broadly existent image mode, has become the emphasis topic of development. Measurement of workpiece surface quality based on texture analysis has been widely put into action in the modern automated production. Based on the tridimensional shape of the texture on the workpiece surface, the new concept of solid texture is proposed in this paper. By monocular vision, this paper deeply studied the method of defocus microscopy image 3-D surface reconstruction, which is applied to acquiring the solid texture of workpiece surface. Extracted the solid texture feature by the texture analysis approach based on statistics of texture primitive's gray pattern, and analyzed the relativity between the texture feature and the processing conditions. The main research contents are as follows:(1) Adopted the edge characteristic image registration, and the practical experiments prove this auto-matching method is effective in focus analysis by high accuracy to serial defocusing image.(2) Studied the image focus criterion using wavelet transform based on the texture analysis. Contrast to the classics focus evaluation function, the function constructed by wavelet transform based on texture analysis is more available in the monotonicity,sensitivity and resolution.(3) Studied the method computing the texture height based on Gaussian interpolation algorithm. According to the distributing feature of the focus evaluation function, Gaussian interpolation algorithm has been selected to interpolate a small number of focus measure values to obtain accurate depth estimates to 3D reconstruction.(4) Studied the 3D reconstruction based on the information of depth and gray scale: according to the gray scale distributed in the subdivided zone of sharpness, rectified the depth of every point and obtained the more accurate information of the 3D surface.(5) Studied the texture analysis approach based on statistics of texture primitive's gray pattern, analyzed the characteristic parameter of the solid texture all sidedly to classify and subdivide the texture, then analyzed the workpiece surface shape.(6) Analyzed the relativity between the solid texture feature of turning workpiece surface and the processing conditions, proved how the distribute,mean value and the variance of texture height is relate to processing time and parameter.