一种新的基于OpenCV的立体视觉三维测量方法

A new Stereo Vision 3D Measurement method based on OpenCV

主要探讨一种实现双目视觉三维测量建模的新方法;即在对已获得的两幅标定图像特征提取后,采用一种融合立体匹配方法得到两两对应的像素点,进而提出一种考虑方程组所代表几何意义的束调整算法求取物体三维空间点坐标;利用异面直线公垂线中点逼近物体空间点,并进一步优化获得更精确结果,同时给出基于Opencv算法,充分发挥了OpenCV的函数库功能;通过实验、数值分析与计算,最终验证了该方法理论分析的正确性及实践可行性。

To probe into three--dimensional measurement modeling, this new method can achieve binocular vision of a three--dimensional measurement modeling. After that has been in the calibration of the two image feature extraction, an integrated stereo matching approach is used to get the two corresponding pixels, and then propose one kind of consideration equation set which represents geometric significance of the bundle adjustment algorithm to strike a three--dimensional point objects coordinates. The algorithm makes use of the mid--point of perpendicular bisector of noncoplanar lines to appoximate to the public object space point, and further optimizes the access to more accurate results. OpenCV based algorithm is given, give full play to the OpenCV library functions. Through experiments, numerical analysis and computing, the correctness of theoretical analysis and practical feasibility of the method is validated.