Research On Signal Processing Method Of Tree Radar Based On Multi-scale Analysis

Ancient trees are often subject to rot,voids,and other defects due to external factors.These defects can cause them to collapse or even die.Early defects often occur within the trees and therefore require an effective non-destructive testing technique to detect them and prevent further decay.As an effective non?destructive detection technology,radar wave is gradually applied to the field of non-destructive testing of trees for its advantages of real lossless,high detection precision,convenient carrying and simple operation.However,due to the external disturbances and the complex structure of the trees and the shapes and positions of the decayed and empty defects,the tree radar wave images have problems such as Iow signal-to-noise ratio,inconspicuous defect features,and low visibility.In this paper,multi-scale analysis algorithms such as Wavelet,Curvelet,Contourlet,and Shearlet transform are studied.The above algorithms were used to denoise from the tree radar wave image and extract the defect edge feature,so as to increase the visibility of the tree radar wave image.By comparing the experimental results of wavelet,curvelet,contourlet,and shearlet transform on forward data,standard test specimens,actual tree specimens,and live-tree radar images,the abilities of removing noise and keeping characteristic of edges of defects have been compared to select a more suitable algorithm to denoise for tree radar images.The results show that Shearlet transform is a more suitable algorithm for tree radar wave images than wavelet,Curvelet,and Contourlet transform.