Research On Blind Reversion Of Single Motion Blurred Image

The restoration of motion blurred images has always been a hot topic in the field of image restoration.In the process of photographing,the motion blur is caused by the relative displacement between the camera and the object,which seriously affect the subsequent processing and application.Image deblurring is to use a certain means to recover a clear image from the fuzzy image.When the fuzzy kernel is unknown,the problem belongs to the category of blind restoration,which is a serious ill-conditioned problem.With the deepening of research,the researchers put forward many excellent algorithms.But some of them have many shortcomings that are the ringing effects and the loss of detail information of the image,which are mainly due to the inaccurate result of image edge extraction and fuzzy kernel estimation.In order to solve these problems,some improved algorithms are proposed to improve the blind restoration effect of motion blurred images from two aspects of edge extraction and fuzzy kernel estimation in this thesis.Firstly,an improved algorithm of edge extraction based on Canny operator is proposed in this thesis.By studying the algorithm of fast blind deblurring,edge detection operator is used to extract the edge,and then the fuzzy kernel is estimated based on the prior knowledge of image edge.It is a method that the threshold could be set adaptively,which is to set the high and low thresholds and filter out the false edges and the isolated edge points.Experimental results show that this algorithm can filter out a large number of pseudo edges and the edge is clearer and more continuous.Secondly,a method of refinement for the initial fuzzy kernel is proposed.The algorithm is used to estimate the fuzzy kernel by edge detection,and the iterative support domain detection algorithm is used to modify and refine the fuzzy kernel to suppress some discrete values in the initially fuzzy kernel.The fuzzy kernel is made finer and more accurate.The multi-scale restoration of the image pyramid is used to solve the problem that the fuzzy kernel is too large.The experimental results show that this algorithm can effectively suppress the discrete values in the initially fuzzy kernel and get the ideal fuzzy kernel so that it can effectively reduce the ringing effect of the restored image and can keep a lot of detail information.