Research Of The Camera Calibration Methods Based On Neural Network

As one of the world’s most popular areas. computer vision is widely used inmedical, aerospace and industrial measurement. Camera calibration is one aspect ofcomputer vision which is basical for further research. For higher calibration accuracy,the paper is focused on camera calibration, including models, traditional methods,corner detecting, BP neural analysis, measure of objecting.First, a model is given, by testing traditional methoding, which is better in bothspeed and accuracy than traditional Harris methods.Second, the paper studies the neural network and analyzes the common featuresbetween neural network and camera calibration. Based on BP neural network, a cameracalibration method is estabished, and also, its network structure and idea are given.Through simulation and physical experiments, the paper shows that BP neural networkmethod is better in robust performance and precision.Finally, by studing monocular vision theory and PnP problems, the measurementprogram based on four coplanar feature points pose is chosen and a measurementexperiment based on Halcon object pose is conducted.