The Investigation And Improvement On Semi-blind Image Restoration Techniques

In recent years, with the development and the widely application of science technology, more and more scholars have pay attention to image processing technology. Image processing technology refers to a series of operations, such as digitization, coding, transmission, analysis and so on. Obviously, processing the degraded image is very popular in the field of aerospace, remote sensing image, biological science and so on. In this paper, based on the semi-blind degraded images, the author studied some classical algorithms which used in image restoration, several improved algorithm has also been studied emphatically. To sum up, follows work is achieved.Based on the degraded image, the author firstly illustrates the reasons and the models of the degraded image. According to the PSF function in degraded factors, this paper provides the methods of PSF function and the methods of estimating the parameters. Then this paper introduces the related contents of image noise, including its classification, model and the method of estimating parameters.Secondly, based on the PSF function is known or can be estimated in advance, several kinds of comparatively typical image restoration technology are put forward, which includes the inverse filtering, the classical wiener filtering, constrained least squares filtering and Lucy-Richardson filtering. Based on the deeply introduction of algorithm, including the diagram which specifically shows the implementation of the algorithm, and the detailed way of MATLAB simulation, the above algorithm are simulated in the MATLAB, from the simulation results and the results of MSE and PSNR, their characteristics are summarized.Finally, on the basis of the above image restoration technology, several improved technology which based on half blind image restoration image are put forward. The first technology is the combine of classical wiener filtering with spatial filtering. This paper firstly introduce spatial filtering, including the algorithm of spatial filtering and the simulation on MATLAB, then selected the inverse filtering and Lucy-Richardson filtering used in combination. Through MATLAB simulation, the expected results are achieved. However, compared with the latter, the former simulation results is not perfect, so this paper has carried on the simple analysis to it.The second recovery technology is the improved classical wiener filtering, including the use of constant instead of noise to signal ratio and secondary wiener filtering. In the former algorithm, by adding a constant, this method has more obvious improvement compared with former recovery technology. The latter algorithm of wiener filtering is the extension and improved of the former algorithm, by making the use of the first result, the noise to signal ratio can be estimated, and start to wiener filtering secondly. It also gives the diagram which specifically shows the implementation of the algorithm. In the MATLAB simulation, in order to simulate the degraded image practically, the noise not only contains the gaussian noise in the degraded image, with the comparison to the former filtering, the effect of secondary wiener is more obvious.At last, The author provide a summary for the paper, and has made the further for the development of image restoration technology.