Traffic Image Denoising Based on Translation Invariance Bandelet with Adaptive Multi-Thresholding Method

Image denoising can effectively overcome the noise in the video surveillance images, thus the texture detail of the target area is enhanced. Combined with the character of second bandelet transform, a novel traffic image bandelet denoising scheme is proposed by adaptive multi-thresholding redundant method. Bandelet transform with redundant is able to achieve optimal approximation and represent image of sparse for utilizing image geometrical regularity. During bandeletization, thresholding estimating is introduced to reduce influence of noise on geometrical flow, and then make use of Bayes shrinkage denoising in ban de let domain. The numerical experiments indicated that the proposed method of denoising is much better than wavelet and curvelet also contourelet. Especially, Bandelet denoising can preserve much more details of the traffic images and have better visual quality.