Research on distance measurement method of moving objects in video

Abstract

We propose a smart car loading camera to simulate the vehicle driving process, the purpose of which is using monocular vision to solve the collision problem that may occur when passengers open the door after the vehicle stops. First of all, Youzheng Zhang’s calibration method is used to obtain the internal and external parameters of the camera, and then the inter-frame difference method is used to lock the moving target in the video. After preprocessing the image, an iterative method is proposed to automatically obtain the threshold to complete the target object segmentation. Combined with the calibration parameters, the distance of the target object is calculated based on the proportion of similar triangles. The test results of videos taken in multiple different scenes show that the average measurement error of the algorithm is 44mm.