A Fast Deblurring Algorithm For Object Region

Aiming at the blur of fast moving object in video images, this paper presents a new fast deblurring algorithm for the object region. At the base of analyzing and summarizing large numbers of existing motion deblurring algorithms, we have carried out intensive research. The central ideology of this new algorithm is to lessen the operation time by cutting down the operation space, with the precondition of keeping the deblurring quality. For smaller operation space, it only make deblurring operation on motion object region, and with this method, the noise on background can not affect the deblurring operation. Orderly, this paper expatiates the new algorithm by three steps: Firstly, it discusses the discrete model of motion blur, and gets the inverse operation model which is used for motion deblurring, with this model, we can find the disadvantage of operation to the whole space, and draw forth the solution method of motion vector; Secondly, with discussing several existing methods of fetching motion vector, it calculates the differences between the neighboring frames, and with the two continuous results, it abstracts the motion vector and gets the region of motion object, this is also an innovative method; Finally, only aiming at the object region, it uses the motion vector to accomplish the deblurring processing by making inverse operation of the discrete motion blur model. Extensive experiments show that the new algorithm can keep the deblurring quality on motion object, and because of the cutdown space range, the new algorithm has a fast processing speed, and it can reduce the disturbance of noise in background, so the new algorithm improves the time complexity.