Research On Restoration Algorithms For Motion-Blurred Images

The techniques of image restoration that have become one of the hot research topics have wide application fields. The key to image restoration is how to construct proper blur model. For image restoration can be regarded as an estimation procedure. Some algorithms can construct the original images if the blur images and the system parameters were known. Some digital images contain noise that not only degrades the images' quality but also affects the restoration results. We think there are mainly three problems should be resolved to acquire good restoration results. Those include getting correct blur model and parameters, denoising and selecting proper restoration algorithms.In this paper we fully research the restoration of motion-blurred images, we research many methods to estimate blur parameters and many image denoising methods and some restoration algorithms for motion-blurred images. Some algorithms have been improved. Our main work is as following:(1) Research many methods to estimate blur parameter, carefully analyze the blur function of straight line motion with constant velocity, put forward a method to estimate blur parameters by zero value points according to the blurred images' spectrum. The complete reasoning procedure is given and the following experiments demonstrate the method's effects.(2) Analyze some denoising algorithms to process Gauss Noise and Pepper and Salt Noise that are very familiar in digital images. We also research other methods that can denoise and keep the definition for an image at the same time.(3) Research five image restoration algorithms and their feature, simplify difference restoration method, analyze restoration method for non-horizontal motion blurred images.