Research On Restoration Algorithms Of Motion Blurred Images And Foggy Images

In the process of imaging, images may produce different levels of degradation because of various factors. However, People need to get high quality images in many applications. So, how to effectively recover the original image, or recover out the image that we expected, become an important research topic in image processing. Motion blurred image is caused by the relative motion between the camera and the object. The foggy image is blurred and the contrast is relatively low, because the air contains many suspended particles that have light scatting effect in the foggy weather conditions. The two phenomena are often encountered in daily life, so, this paper mainly researches on the recovery of motion blurred images and foggy images.The main work of this paper is listed as follows:(1)Research on the image denoising algorithms based on wavelet transform and curvelet transform. Compare the threshold denoising algorithm based on curvelet transform with the threshold denoising algorithm based on wavelet transform and the wiener filtering denoising algorithm based on wavelet domain. The experiment results show that the threshold denoising algorithm based on curvelet transform is the best. Owing to the defect of curvelet transform lack of translation invariance, this paper proposes a cycle-spinning threshold denoising algorithm based on curvelet transform. The experimental results show that the algorithm can better improve the quality of degraded images.(2) Research on the restoration of motion blurred images. The restoration process of motion blurred images is:Firstly, estimate the parameters of the point spread function of the motion blurred image. Then, adapt the classical image restoration methods to restore according to the estimated parameters of the point spread function. In the case of noise, the restoration results of wiener filter are the best, but it still can’t effectively suppress noise. To this, this paper proposes to adapt the cycle-spinning threshold denoising algorithm based on curvelet transform for further denoising after using the classical image restoration method for recovery. The ability of wiener filter to suppress noise is better than the L-R algorithm, but, the ability of the L-R algorithm to preserve the details is far better than the wiener filter. Considering this paper will adapt an algorithm for further denoising after using the classical image restoration method for recovery, so, this paper adopts the L-R algorithm for the initial recovery, and then adapts the cycle-spinning threshold denoising algorithm based on curvelet transform for further denoising.(3) Research on the image haze removal algorithm based on the dark channel prior. The effect of the algorithm is good, but the running time is long, and most of the time spend on the optimization of the transmission rate. According to this characteristic, this paper adopts a fast method to optimize the transmission rate, it reduces the optimization time greatly. Thereby, it improves the running time of the whole algorithm.