Design Pamorama Optical System Based On Computer Vision

Tank opto-electronic detection apparatus, the most important part of the fire control system, can determine the efficiency of firepower system and also has the function of analyzing battlefield information. In this paper, the design of distributed aperture panoramic optical system will remedy the defections mentioned above. The panoramic optical system has some advantages such as real-time passive detection covering 360 degree space and multi-target tracing. It can also provide high resolution ratio panoramic pictures which can be operated by computer vision. Then the functions of advance warning, tracing, aiming, etc, can be realized.In this paper, the optical design of panoramic system is described. A wide-angle lens of 94 degree with big aperture is designed. Different from multi-sensor of DAIRS (Distributed Aperture·Inf rared System), four receiving lens are arranged according to DAIRS principle to cover the information of 360 degree. A spatial image transferring system is designed which uses a CCD receiver to detect panoramic information. Some experiments are conducted to verify the feasibility of the principle. According to the design parameters an appropriate CCD is chose and a receiving objective lens of the CCD is designed. The image quality of the whole system is analyzed. In the ideal condition some important technical indexes such as the detecting and discriminate distance, blind field, image overlap scope, invalid collection distance, depth of field and whole distortion radio are estimated. Overlap pictures are separated by clock cut shutter. The reasons of the image motion of the sight system are analyzed and the forward image motion is defined to be the main image motion of the panoramic system. The compensation of the main image motion is realized by using a focal plane curtain shutter. The formula of curtain width and its moving speed is derived to ensure the reliability of the details of the detected image. The image mosaic of distance shot and close shot are implemented by computer using Harris algorithm to select feature point, Cross Correlation algorithm to match the feature point and the panoramic mosaic algorithm to mixing the cylindrical coordinates and gray scale pixel. The accuracy and feasibility of this image mosaic algorithm are verified.