Wood Defect Image Processing Based On Wavelet Transform And Multi-structural Elements Morphology

Nondestructive testing is a practical wood-detection means, and it can detect wood internal defects accurately without performance destruction. Wood nondestructive testing prominently improves the utilization rate of wood, the production cost and the production efficiency, which has high practical significance and economic value. Taking the cost and maneuverability of experiment into account, this paper uses the X-ray nondestructive testing system as detection means. After acquiring defect images by nondestructive testing system, processing and analyzing is the most important step of wood nondestructive testing. In the process of acquiring and transferring wood images, they are always polluted by external factors, so the human eyes can not locate the defects accurately. The digital image processing can reduce the influence of external factors greatly. On the basic of the applications of wavelet transform and mathematic morphology in image processing, this paper proposes some new methods and then verifies the advantage of them by comparing with traditional methods.In the application of wavelet transform in wood image detection, the main research object is wavelet threshold shrinkage and wavelet modulus maxima. First, reduce noises by wavelet threshold de-noising method, and then segment wood defects from the image. The segmentation threshold is set by the OUST method. Besides; detect the wood image edge by the method of wavelet modulus maxima based on gauss filter. Choose appropriate filter length and, wavelet decomposition scale, and threshold to determine the edge points by looking for the modulus maximum points.When researching the application of morphology, the paper applies a combined morphological edge operator and multi-structural elements morphological operator in edge detection of wood images. The improved morphological operator owns powerful ability of anti-noise, and it increases the clarity and accuracy of edge detection. Besides; make use of different structural elements to detect the image respectively and get the edge image by weighted average. The obtained edges are more complete and more continuous. For images with more interference details and noises, set the wavelet high frequency coefficients to zeros, and then use morphological algorithm to process and analysis the low frequency reconstructed image. This method has a good effect in removing noises and interference information. At the same time, using morphology can guarantee the defect information clear and accurate.The experiment results show that the adopted methods in this paper have far more superiorities in wood defects detection. They greatly improve the accuracy and efficiency of detection, which has very important significance to practical applications in the field of wood nondestructive testing.