Wood Defecet Detection Analysis Edge Recognition Based On Quaternion Matrix Singular Valve Decomposition

With the rapid development of society, the use of wood is increasing significantly, and the wood processing and the production is growing rapidly. In China, the technology of the wood defect detection is still very antiquate, human identification process is susceptible to the subjective and objective factors, such as senses> emotionn fatigue and so on. The traditional physical methods of testing need higher cost, and stricter testing situation around. The use of vision-based machine automatic detection is a traditional physical method that reduces the impact of human factors subjectively and lowers the cost. Therefore, the wood defects based on the processing of visual images can detect the wood defect quickly and accurately, which is the subject of this study.Compared with the gray image, Color image not only contains rich information content and structural image information, but meets the visual senses of people as well. Although the color defect detecton in wood defect processing has been widely used, the components of color image space are usually separated, then each component is processed respectively, at last, they compound the required wood color defect images, thus the related information of the original images will cause loss. In this article, the estalishiment of quaternion model using color image can regard wood defect images as an organic whole, which is a good solution to the problem of missing of related image information. In this paper, we choose three kinds of defects to research and analysis, that are worm holes, dead knot, live knot. We raise a new method to test wood defect based on quaternion matrix singular value decomposition. In this study, we will research and analyze the following several aspects.We establish a quaternion wood model of the original image. Quaternion method and this paper quaternion matrix representation of complex matrices mentioned in the new complex, said the improved method, based on the RGB image is converted to wood quaternion two different models, namely the quaternion matrix, and maintaining the integrity of the information related to the image.We discuss, analyze and prove the theorem of quaternion matrix singular value decomposition. In these two quaternion representation methods, the friendly nature of the vector of quaternion matrix is used smartly, it not only proves and analyzes the decomposition theorem quaternion matrix singular values, but get quaternion algorithms and formulas in MATLAB by Singular value decomposition algorithm, resulting in quaternion singular value matrix and two different left and right eigenvalues.The use of complex quaternion representation method can be carried out on wood characteristics image, and analyze the structure of the original image to get de-noising wood image based on quaternion.We propose a new improved version of the quaternion representation method of the complex, and principal components analysis of timber. Analysis of feature images singular value of timber, which uses linear weighting and nonlinear weighting methods to get varying degrees of wood defect detection.The use of mathematical morphological processing. We apply the morphological processing and Canny edge extraction to the results of defect detection based on improved complex quaternion representation.