Auto-focusing Method Research For Airborne Optoelectronic Platform Camera

Nowadays, high-powered airborne optoelectronic platform develops quickly at the direction of high- altitude and oblique far-sight. In order to recognize and track the long distance target, the focus length of the optic system of airborne optoelectronic platform has been as long as 1000mm, or even longer. The long focus length results in shot focus depth of optic system. The change of back-focus length caused by the variance of temperature, pressure and other environment parameters in high altitude will lead to the defocus of image system easily. The defocus will lead to a quick down of the camera resolution, so that, the quick and precise auto-focus method is quite important to ensure the high performance of the airborne optoelectronic platform. The research of image process based auto-focus method on airborne optoelectronic platform was represented in this paper.Firstly, the mathematic model of defocused optical image system was founded. Different focused images were produced by calculate the correlation between the PSF of defocused optical system and original image. The radius of blurred circle was detected from the Fourier transform power spectra of defocused images. Two kinds of depth from defocus (DFD) auto-focus methods were represented.Secondly, kinds of focus measure method and auto-focus search strategy were represented. Based on that the attenuation of power spectra mainly happened in the high frequency, two betterments over the Power Spectra Sum (PSS) focus measure function were proposed. The two betterments could improve the saturation region and sensitiveness of the focus measure curve respectively. According to that the gradient edges in normal images were more than horizontal and vertical edges, a new focus measure method (lean operator) derived from gradient edge detection was proposed. Performance test of focus measure method discussed in this paper was performed on different type of image sequences.The third, A PC based auto-focus experiment system was designed.Series of experiments were committed to have the focus measure methods compared. According to the experiment results, the lean operator and correlation operator were respectively chosen to be used in the coarse search state and fine search state of the auto-focus method. Then, Local full search (LFS) strategy and change step hill-climb search (CHS) strategy were proposed in allusion to that the airborne image equipment was prone to be interfered by the environment parameters in high altitude.At last, an embedded auto-focus system based on DM642 DSP and cyclone III FPGA was designed. The LFS and CHS algorithm were transplanted on the embedded auto-focus system. The results show that the correlation operator with the cooperation of the lean operator can locate the prime focus position accurately,the accuracy is 95% up. The total focus time of LFS and CHS search strategy are respectively 5s and 4~7s, which satisfies the need of engineering application.