基于自标定双目视觉算法的3维图像构建

为了解决传统双目视觉标定方法构建3维图像的缺点,采用自标定双目视觉算法建立了双目视觉系统,确定了空间对象点;通过自标定数学模型构造矩阵,对两相邻图像的本质矩阵求解;并用立体空间方程的投影矩阵得到4个线性方程,用外极线约束法得图像点的世界坐标值。结果表明,物体空间轮廓清晰,避免重影,不再具有双重轮廓,3维表面比较逼真,具有较强的立体感。

3-D image construction based on self-calibration binocular vision algorithm

In order to solve the defaults of the traditional binocular vision calibration method for constructing 3-D images, a binocular stereo vision system was designed based on self calibration algorithm of binocular vision, and the spatial objective points were determined. The essential matrix for two adjacent images was solved based on the matrix constructed according to calibration mathematical model. Four linear equations were obtained according to the 3-D spatial equations of projection matrix. Coordinates of image points in the world coordinate by means of external the epipolar constraint method. The results show that the outline of the objective space is clear without ghost, and the 3-D surface is more realistic, stereographic.