A method of image restoration technology based on parallel phase diversity algorithm

Although the phase diversity method is an effective way to detect wave-front and restore image, it is difficult to be achieve its real time application on DSP or FPGA. The main of disadvantage of method is a great computation when it is used to estimate the wave-front phase aberration and restore the degraded images. In this paper, a parallel phase diversity method is deeply researched and the expression of the evaluation function is further clarified according the theory of Nijboer-Zernike polynomial. An outdoor image restoration verification system is established. The structure of an ordinary telescope is modified, so that it can acquire two images on the focal plane and the out-of-focus plane of the imaging system. The image objects are a chessboard at a distance of 1.1km and a worker in a construction site at a distance of 4.5km outside the laboratory window. The results indicate that the restoration image has a higher resolution. No-reference assessment methods are adopted to evaluate the quality of images. The FI value of restoration image of chessboard improved 1.478 times, and the LS value improved 2.178 times. The FI value of restoration image of worker improved 4.227 times, and the LS value improved 1.623 times.