Three-Dimensional Computer Vision Based On Colored And Structured Light

This paper investigates a method of three-dimensional (3D) vision measurement based on colored structured light. In application of three-dimensional structured light vision measurement, the establishment of reasonable mathematical model which reflects two-dimensional picture coordinates and the corresponding relational three-dimensional space coordinates and using effective method to calibrate the model parameter is the main contents of the project. In view of respective shortcoming using the linear model and the non-linear model, we use a new method based on neural networks, using the good non-linearity mapping ability as well as the study of the network, and training the network model with high accuracy from sample data. This is a novel structured light three dimensional vision method, compared with the conventional method. It does not need to consider each kind of factor that may affect the precise of the system, thus it can effectively make up the insufficiency of conventional modeling methods, and guarantee that the system has a high measuring accuracy.