Improved Motion Blurred Image Restoration Based On Radon Transform

Motion-blurred image restoration is an important research topic in the field of digitalimage processing. Since the process of acquiring an image is photographed to produce arelative movement between the target object and the imaging system, which becomes blurredtime. Due to the motion blur is present, the image quality is not be guaranteed to obtain. Thisis useful information to the subsequent acquisition of the image difficult. Motion blurredimage restoration is mainly composed of the following two parts: the first part is estimate thepoint spread function (PSF). The second part is restoration of the image by the point spreadfunction,which is estimated.Usually acquired motion blurred image is affected by noise. Then, image preprocessingis added before making image restoration in this thesis. That is the image wavelet imagedenoising. This thesis proposes a motion-blurred image restoration method, which is based onRadon transform. Specific research work is as follows:(1) The basic theory of motion-blurred image restoration is introduced, such as blurredimage degradation model, the point spread function and so on. The basic principles of theinverse filtering restoration, Wiener filter restoration, Lucy-Richardson restoration and leastsquares restoration are described. And the four methods of restoration are compared andanalysed. In this thesis, the Wiener filter restoration method is selected. Several commonobjective assessment of image quality parameters are introduced.(2) The basic theory of wavelet transform is introduced, and the advantages of waveletthresholding are described. Some key issues of wavelet thresholding are researched. The mainproblems are the threshold function and thresholds. And the importance of threshold isillustrated through experiments.(3) A fuzzy angle estimation method is proposed, which is based on Radon transform.And that’s proposed by studying the spectral characteristics of image motion blur. The idea ofedge detection is introduced to the fuzzy angle estimation. The accuracy of the method isverified by several experiments. The fuzzy scale is estimated by differential autocorrelation.And the error of the method is analysed. Then this thesis also analyses the anti-noiseperformance of the PSF estimation method by experiments.(4) The overall structure of the method is gived in this thesis. The superiority of thismethod is verified by simulating the motion blur simulation experiments.(5) The actual motion-blurred images are also restored in this thesis. The representativeimages includes text information is selected. And the effect of image restoration is good. Thenthe method has a good practicability.