3D Estimation of Single Image based on Homography Transformation

Aiming at the problem that it is difficult to extract three-dimensional information from a single plane image under the condition of lacking camera parameters, this paper proposes a scheme combining perspective transformation and pseudo inverse method to realize the rough extraction of three-dimensional information from a single image. On the basis of the existing geometric model of camera imaging, taking points from the two-dimensional image and calculating the perspective transformation matrix, the homography matrix is obtained, and then the basic parameters of the camera can be marked through the corresponding relationship between the matrix and the camera internal and external parameter matrix. Pseudo inverse method is the inverse process of camera imaging, that is, it can estimate the final three-dimensional space coordinates from the pixel coordinates of the captured plan through the pseudo inverse matrix, and realize the one-to-one correspondence between two-dimensional coordinates and three-dimensional coordinates. The experimental results show that the algorithm is fast and convenient. Compared with the traditional Zhang's calibration method, this calibration result error is less than 5%, and the execution time is shortened by 70%.