Application Of Wavelet Analysis To Wood Defect Images Processing

With the development of wood science, using X-ray as non-destructive testing method and computer digital image processing technology for wood defect images became an important direction of wood non-destructive testing. During the wood defect images were obtained by wood X-ray non-destructive testing system, factors such as the camera system quality, the given light condition and input device increased noise to deteriorate image quality, make image fuzzy, submerge features and reduce contrast. It was different to identify wood defect.As a popular digital image processing tool, wavelet analysis had been widely used in image processing because of its good time-frequency localization characteristic and multi-scale feature. According to the characteristics of wood defect images, wavelet analysis had been used in image denoising and image edge detection to make image clearer, easy to identify.The basic theories of wavelet analysis, multi-resolution analysis, MALLAT algorithm, wavelet base selection and other related contents were analysed in this paper. The principle and realization methods of image denoising and edge detection algorithm using wavelet transform were researched, compared with traditional image denoising methods and classic edge detection methods.The wavelet threshold denoising was the key research among wavelet denoising methods for wood defect images. Several key problems of denoising wavelet, threshold and threshold function selection had been carried on analysis and discussion in detail. Two different thresholds of global threshold and local threshold, three different threshold value initialization functions of hard threshold, soft threshold and high frequency coefficients set zero were used to denoise processing for wood defect images. With Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR) as evaluation standard, wavelet threshold denoising algorithm and the traditional image denoising algorithm had been done simulation experiments to find the suitable denoising method for wood defect image. Single denoising method could not meet the need of image processing. Denoising method based on wavelet packet combined with gray-scale morphological filtering was proposed for wood defect image. The method received better PSNR and visual effect. It was better than other denoising methods.Single scale edge detection was not good for image detail position and extraction. Multi-scale characteristic of wavelet transform provided a new method for edge detection. B-spline wavelet as edge detection wavelet, application of multi-scale edge detection algorithm based on the dyadic wavelet transforms achieved edge detection for wood defect images.In addition, wavelet packet can decompose low frequency and high frequency to obtain more image information. The approximate part which was obtained by reconstruction can remove the high frequency to detect clearer, more continuous edge. Wavelet packet was used in edge detection for wood defect images and also got good results.