The Study On Algorithm Of Motion Blurred Image Restoration

Motion Blurred Image Restoration (MBIR) is to reconstruct the origin image from MBIR using the Point Spread Function(PSF). MBIR is widely used in many fields.This thesis introduces the basic theories of restoring motion blurred image firstly. Traditional image restoration methods are retrospected, Their performance is illustrated via simulation. The relationship between PSF and the frequency spectrums of MBIR are revealed. An improved restoration algorithm is proposed based on Traditional Image Restoration (TIR). The main idea is to apply the Radon transform of the spectrum of MBIR to find motion direction. The blurring extent was computed by Autocorrelation Function (ACF) of MBIR along this direction. The experiments results show that the proposed algorithm can estimate the blur parameters more accurate.As the advantage of Total Variation Model (TVM) in restoring motion blurred image over the tradition algorithm, it becomes the research focus in the area of the image restoration. Total Variation model for image restoration is thoroughly studied in this thesis. An improved Adaptive Total Variational Model (ATVM) is applied to restoration of some actual Motion Blurred Images (MBI). Two threshold parameters were selected to further improve its performance. Demonstration shows that the newly proposed algorithm can get a better recovery results than previous algorithm in the aspects of a balanced image de-noise and edge protection.Ringing Effect (RE) might occur in the process of restoring some motion blurred image. This thesis pointed out that that the difference of frequency spectrum between the artificial and the actual blurred image is the main reason behind it. Margin Circulation Method (MCM)is chosen to eliminate RE. Simulation illustrated the effectiveness of this method.