| The restoration of motion blurred images is closely related to human daily lives, such as the recognition of high speed vehicles’ license of violations, the discrimination of the crime person’s motion blurred images, the restoration of blurred cloudy images, and so on. Therefore, the technology of motion blurred image restoration is one of the important subjects in image restoration fields. In this paper, two new fields algorithms are proposed, the first one is get the motion blurred direction and length based on Radon transform and differential autocorrelation algorithm, the second one is restored the blurred images based on R-L algorithm.To get the direction and length of motion blurred images, a new algorithm is put forward to based on Radon transform and differential autocorrelation algorithm. The process of improved algorithm is as follows, Radon transform and differential autocorrelation is used to get the PSF. The direction of the motion blurred image is fixed which is get by Radon transform, image blurred length is iterated on initial recognition of length in certain ranges to recovery the images. According to signal-to-noise ratio between the restored image and blurred images, the optimal motion blurred length can be get. The optimal motion blurred length is taken as a fixed length, image blurred direction is iterated on initial recognition of the direction in certain ranges to get the optimal motion blurred direction. The experimental results show that the direction and length get by this algorithm is more accurate than the traditional algorithm.To recovery the motion blurred images better, a adaptive R-L algorithm is proposed. For the images with salt and pepper noise, PCNN is used to remove the noise. The images’ edge and region, whose gray changes violently are smoothed using Butterworth. The adaptive R-L algorithm is used to recovery the images, this algorithm is equal to of a low pass filter which process the image’ edge and the gray changed violent areas in the process of iteration, this algorithm can suppress ringing effectively. Compared to the original algorithm, the algorithm is better, especially in suppressing ringing. |
