Abstract: The paper presents an effective technique for tackling one of the most challenging problem of photographers viz. motion blur due to camera shake. The most common cause of a blur image is camera shake or a relative movement of a handheld camera and object in a given exposure time. The movement may be very small, but still creates blurry images. There exist different conventional techniques which are used for removal of motion blur to get clear and sharp image. Most of these methods use multiple images as a input taken through a burst mode, a feature available in all modern cameras, and combine them to get a more clean image. However there are certain limitations and disadvantages of using these conventional techniques. Removing a blur from a single image is a challenging problem. The proposed work is based on a technique that uses a single input image unlike other conventional techniques for removal of blur in images post capturing it. In the proposed technique a single blur image is divided into smaller images with an assumption that each sub image is uniformly blurred. with the implication of Fourier algorithm these sub images are used to estimate motion blur parameter -Blur Length and Radon transformation is used for determining values of Blur Angle. A local parametric blur model is prepared with the help of these estimated values of motion blur parameters, VIZ blur length and blur angle. These models are then deconvolved with blurry sub images. The resultant of this algorithm is a reconstructed, original, blur free image. The proposed technique is an effective solution to remove motion blur and serves the important requirement of clear images of various fields like medical, navigation, satellite imaging, driving assistance systems.
Keywords: Camera shake; Deconvolution; Single-image deblurring; Motion blur.