Research On Image Denoising And Motion Blurred Images Parameter Estimation

Image denoising is an important part of image processing, it has significant effects on further image analysis. Clear image contains a lot of information of the image, wavelet image denoising method because of its low entropy sex, multi-resolution, to wait for good correlation features a wide range of applications. To improve image quality is of great significance. Effective and accurate get clear image for subsequent has important influence on the image processing, can make the recovery of visual effect after a further improved,research on restoration of motion blur has become one of the hot issues in the image processing technology, and the restoration technology of motion blur image helps to some real work. Therefore, it has great significance for research on the restoration technology of motion blur image. The research of restoration of motion blur image focus on how to extract useful information from a motion blur image, and finally get the clear image. The main research of motion deblurring including image noise removal, fuzzy parameter estimation and image restoration algorithm, image noise removal is to processing noisy blurred image, reduce the noise interference before image restoration, and further improved the visual effects of recovery image. Fuzzy parameter estimation is mainly used to obtain motion blur point spread function (PSF), which include the direction of movement and the fuzzy scale estimation.In this paper in view of the existing wavelet denoising algorithm and some problems in the uniform motion caused by movement the restoration of blurred image problem. We obtain some algorithms by analyzing the parameters of blurred images and the restoration technology, and combining with the information of actual images. We prove the validity and feasibility of algorithms by theory and experiment:(1) We propose a restoration algorithm based on a joint deblur filter of new threshold in this paper. First, we introduce the degradation model, discuss the effects of image noise on motion blur images, learn about the basic knowledge of wavelet denoising, set up a new wavelet function, reduce the error of reconstruction signal, compare the different denoising effect of images which are noised by different ways, and compare with the existed algorithms, we conclude that the algorithm obtain higher peak signal to noise ratio. The algorithm can also restrain noise and greatly enhance the suppress noise of restoration algorithms.(2)We propose a parameter identification method for motion blur based on image morphological gradient after blocking. First, we introduce the method of estimating parameter in frequency domain and use morphological gradient of image to decrease the effect that noise has on image. Then the image is blocked, we use a low pass filter to process the blocked image after Fourier Transformation. The effect that noise has on Fourier Spectrum will be decreased. Accurate edges are extracted by using Canny edge detection to identify the blur direction based on Radon transformation. So we can calculate estimation of motion-blur extent. Experimental results show that estimations of motion-blur extent and motion-blur direction are effective and accurate and reduce the computation amounts of motion-blur parameter estimation. Last, we denoise for emulational motion blur image by using the first method then restore the motion blur image by using the second method in our paper. Compared with wiener filter, our recovered result is better.