论文部分内容阅读
无人机自主导航是无人机发展的必然趋势,立体视觉技术作为一种优秀的环境信息测量技术能够为无人机自主导航提供关键信息.但是,导航图像存在幅度失真,现有立体匹配方法的匹配精度较低,针对Census变换舍弃了图像像素色彩信息而造成的误匹配问题,本文提出了一种Census变换和图像色彩信息相结合的联合匹配算法,并经过理论分析提出了正交积分的方法以提高算法的实时性.首先,将Census变换和图像色彩信息联合,构造初始匹配代价;然后,采用改进的自适应窗口作为代价累积窗口,并使用正交积分思想提高累积速度;最后,经过视差提精,获得最终的视差图.实验结果表明:本文算法对幅度失真图像的匹配误差比单独使用Census变换提高了40%～50%,算法的运算时间提速了3～12倍,与Census变换和图像灰度单独作为匹配代价时相比,该方法具有更高的匹配精度,对幅度失真有很强的鲁棒性,能够较好地应用于无人机自主导航场景中.
Autonomous navigation of UAV is an inevitable trend of development of UAV. As an excellent environmental information measurement technology, stereo vision technology can provide key information for autonomous navigation of UAV. However, the amplitude distortion of navigation image exists. The existing stereo matching method The matching accuracy is low. According to the Census transform, the problem of mismatch caused by the color information of image pixels is discarded. In this paper, a joint matching algorithm based on Census transform and image color information is proposed. After theoretical analysis, the orthogonal matching Method to improve the real-time performance of the algorithm.Firstly, Census transform and image color information are combined to construct the initial matching cost. Then, the improved adaptive window is used as the cost accumulation window and the orthogonal integration method is used to improve the accumulation speed. Finally, Parallax refinement to obtain the final parallax map.Experimental results show that the matching error of the proposed algorithm for amplitude-distortion images is improved by 40% -50% compared with the Census transform alone, and the algorithm operation time is increased by 3 to 12 times. Compared with the Census transform Compared with the gray value of image alone as a matching cost, this method has higher matching accuracy, Strong robustness, can be better used in autonomous navigation scene of UAV.