Voxel Coloring Algorithm And Its Application In Image Recognition

With the development of computer and information technique, graphic, image processing and computer vision have been applied widely in virtual reality, artificial intelligence, mechanic vision, target recognition, etc. The main work in this thesis is shown as follows:1. In this thesis, a 3D reconstruction method, voxel coloring, has been discussed and analyzed in detail. According to one of voxel coloring algorithm's shortcomings this thesis brings forward an improvement method, with reference to the theory of Shape From Silhouette. This improvement method looks on every input image as a validate of the reconstruct results, improving the result accuracy and saving the computing time.2. Based on the reconstruction results, this thesis establishes a database with every plane model's 3D information. Then 3 kinds of recognition algorithm, including zeroth order moments, gray histogram and color histogram, have been tested for this system. Form the experiment results, we can see that zeroth order moments method is not appropriate for this reconstruction and recognition system because it discards all the detail information of an image color and texture. However, considering on all pixel color, color histogram is suitable for this recognition system.3. This thesis constructs a reconstruction system with C++ Builder and OpenGL. After the consumer gives the camera calibration parameter and the photos shot by all cameras, the reconstruction system can work out the object's 3D information, then output photo realistic images of the object with OpenGL.4. Some recognition methods are tested with Matlab. The recognition results are analyzed .Finally, the thesis analyzes some problems in this system and presents some possible resolvents.