Image Denoising Research Based On Multi-Wavelet Theory

To improve the quality of image and facilitate image further processing, recognition, compression and so on, it is necessary to denoise image. Multi-wavelet image denoising has been well acknowledged as an important method of image denoising. Based on the knowledge of multi-wavelets, the article has comprehensively researched the following parts: multi-wavelet selection, threshold improvement, multilevel threshold optimization and shrinkage-function betterment.For multi-wavelet selection, the paper puts forward a new combination scheme on the basis of Time-frequency Resolution, Decomposition Coefficient Ratio (namely P) & Reconstruction Error (namely E) and Noise Variance Estimate. If the multi-wavelet holds the higher time-frequency resolution, the smaller value of the parameter P and E and the less error between adding noise variance and estimation of the variance, it can separate more high-frequency part of the wavelet coefficients caused by the noise, have better capacity of decomposition and reconfiguration and simplify the calculation.Another important problem is threshold selection. The threshold selection has direct relations with the result of denoising. The paper has improved traditional Donoho threshold algorithm and optimized noise deviation-based multi-wavelet multilevel. The improved algorithm can achieve better denoising effect and further improve the signal-to-noise ratio.It is equally important to select multi-wavelet shrinkage function. Based on Multiresolution analysis wavelet threshold de-noising method which put forward by D.L.Donoho and I.M..Johnston, a new improved shrinkage function is posed。The de-noising method adopting the new threshold function gives better MSE performance and SNR gains than hard and soft thresholding methods. However, the hard-threshold is best in preserving edges but worst in de-noising, and soft -threshold is best in reducing noise but worst in preserving edges, this method incorporates the hard and soft thresholding to achieve a compromise between the two methods.Based on those above searches of traditional multi-wavelet denoising method, the improved algorithm has been used in application of phytoplankton cellular image'edge detection. The practice has verified that the feasibility and superiority of multi-wavelet theory in image denoising.