激光主动探测成像中全景拼接的算法研究

摘要

Laser active imaging combines laser technology,range-gated technology,weak target imaging technology and so on.Because of its high resolution,strong anti-interference ability and the ability to generate three -dimensional image,it is widely used in various areas of detection.But in this way,the image's field of view is usually small,which is not conducive to global observation and analysis.A panoramic stitching method based on the improved SURF algo-rithm is proposed,and it can quickly match target features of adjacent scenes.The efficiency of this approach and stitc-hing effect are tested through the experiments.The results show that this method can increase the visible range and im-prove comprehensive ability of observation and analysis,which lays a theoretical and experimental foundation for high-speed real-time large field of view observation.%激光主动探测成像是将激光技术、距离选通技术、微弱目标的成像处理等技术结合应用的一项新技术。因其分辨率高、抗干扰能力强、能够生成三维像的优点，广泛应用于各种探测领域。但这种方式获取的图像一般视场角偏小，不利于全局的观察和分析。本文描述了一种通过改进的并行 SURF 算法检测场景角点特征，能够高速实时地将相邻场景焦点特征相匹配的全景拼接方法。测试了这种方法的运行效率和拼接效果。利用改进后的 SURF 算法将小视角的图像拼接起来，增大了可视范围，提高了观察和分析的综合能力。并且将这一方法运用到了3D 成像激光雷达距离信息的拼接中，测试了效果，验证了其可行性。本文提出的快速配准拼接算法，为高速实时大视场观察奠定了理论和实验基础。