HDSGI adaptive restoration of blurred image

For the problem that the high-dimensional space geometrical informatics (HDSGI) blurred image restoration method fails to adjust the parameters automatically, a new blurred image restoration method which combines the HDSGI theory with the chaotic particle swarm optimization (CPSO) algorithm is proposed. Based on the HDSGI theory, the clear restored image can be obtained, while the parameters of the distribution curve in the above method need to be regulated manually and the restored image may result in noise with inappropriate parameters. In this paper, a no-reference quality assessment method, which can measure both noise levels and blurred degrees of images, is adopted as the fitness function of the CPSO algorithm to find the best distribution curve automatically, thus the best restored image is obtained. The subjective vision assessment and the objective quantitative assessment of images demonstrate that the proposed method is practical and effective.