Image Denoising Based On Multiscale Geometric Analysis

The edge and outline of image are very useful in the pattern recognition.But there are much difficulties if the signals are polluted by noise. Ridgelet transform is a new kind of non-adaptive high-dimensional functions after Wavelet transform. For image processing, Ridgelet transform is more effective than the Wavelet transform in representing linear and super-plane singularities.The main work of this dissertation includes:In this paper,we first introduce the theory of Ridgelet transform,and then propose a method for image denoising on scale factor and Ridgelet transform.. In order to indicate its advantages, use Ridgelet to remove noises in images,and compare the results with that of Wavelet. Experiments show that the performance of the proposed method is obviously superior to other methods both in vision and in SNR.In this paper,we first introduce the multiscale geometric analysis theory,and then propose a method for image denoising on scale factor and contourlet transform. Experiments show that the performance of the proposed method is obviously superior to other methods both in vision and in SNR.