摘要
主要研究一种基于双目立体视觉原理测量图像中物体距离的算法,要实现的目标是采用双摄像头从不同视点拍摄目标物体,投影得到两幅双目图像,用图像处理的方法先提取出需要的信息,进行特征识别,然后利用同一目标在不同投影图像上的成像差异进行立体匹配,并通过视觉算法计算被测物体的距离。由于测量的距离远大于摄像机之间的基线距离,所以传统意义上的双目测距系统近似认为两个摄像机的光轴是平行,但是实际上两光轴是必然相交的,所以文章结合实际,考虑两光轴相交,以达到提高距离测量精确度的目的,系统的主要流程包括图像获取、摄像机标定、角点检测、特征识别、立体匹配、距离计算等几个方面。
In this paper,it mainly studies a method of measuring the distance of the objects in an image based on binocular stereoscopic vision principle and aims to obtain two binocular images by using dual camera to capture the objects from the dif-ferent viewpoints. This study takes the method of image processing. Firstly, extract the needed information and have the recog-nition of their features. Then, have the stereo matching according to the imaging differences of the same target in different pro- jection images and calculate the distance of the measured object by the visual algorithm. As the measured distance is much lon-ger than the baseline distance between the cameras, the binocular distance measurement system detects the parallel optic axes between the two cameras approximately from the traditional sense, however, the two optic axes are bound to intersect actually. As a result, this paper is combined with reality to take the two optic axes, intersection into consideration, thus achieving the goal of improving the accuracy of distance measurement. The main processes of the system include image acquisition, camera cali-bration, comer detection, feature recognition, stereo matching and distance calculation, etc.
出处
《信息通信》
2017年第8期61-63,共3页
Information & Communications
关键词
测距
图像处理
Distance measurement
Image processing
