Application Research of 3D Detection System Based on Deep Learning and Point Cloud Data
In order to solve the problem of weak 2D visual detection with high flatness in high and low difference scenes,based on deep learning and pointcloud data processing platform,this research integrates the 3D point cloud format image and deep learning technology to establish a 3D detection architecture mechanism with micron precision,online detection and controllable cost.The hardware scheme of phase shift and grating projection structured light technology is used to obtain the original data of 3D point cloud.Based on the powerful CPU and GPU processing chip,the grating pattern modulated by the measured object is recoded,and the 3D point cloud data is calculated combined with the calibration parameters.The 2D deep learning model is up-graded and developed,which can label,learn,train and detect the point cloud data,and integrate 3D hardware,3D software and 3D algorithms.The experimental results show that the system is conducive to the landing of 3D defect detection system,and lays the algorithm and software foundation for intelligent 3D detection equipment.
3D detectiondeep learningpoint cloud dataintelligent software3D model
万方数据
维普
