Wood Knot Image Enhancement Recognition Based On Quantile Algorithm

The wood defects identification is an important issue of wood material science,which improve the efficiency of wood processing and also save resources.Therefore,it has a widely applications in wood plate processing and production.Wood knot is a sort of common defects in the plate,which reduce the quality and appearance of the plate.To cope with this problem,this dissertation utilized the image enhancement and recognition of digital image processing technology to classify the dead and live knot.1.Introduced adaptive histogram equalization algorithm and direction adaptive interpolation algorithm,and applied them to improve the quality of wood knot images.2.Proposed an efficient wood knot image enhancement method which is based on discrete cosine transformation(DCT),quantile dependent and histogram equalization algorithm.In the proposed method,we apply DCT on the input image to get low frequency component and then use the quantile-based sub-division on the histogram.Finally,the histogram equalization is applied to output the enhanced wood knot image.3.Compared the performance of proposed with adaptive histogram equalization and direction adaptive interpolation algorithm in objective metrics and subjective metrics.The Objective evaluation metrics including Peak Signal-to-Noise Ratio(PSNR),Structure Similarity Index Measurement(SSIM)and Feature Similarity Index Measurement(FSIM).4.Assemble a complete a wood knot image enhancement and recognition system based on Support Vector Machine(SVM),which combine the proposed method,adaptive histogram equalization,direction adaptive interpolation algorithm and SVM recognition process.The input features of SVM contains Contrast,Homo,Dissimilarity and Entropy.