Three Dimensional Surface Reconstruction Based On Laser Triangulation And Photometric Stereo

Three dimensional surface reconstruction technique is one of the most important classic problem in Computer Vision, which is the key technique in image understanding and 3D target recognition. It bridges objective entities and abstract perception, also makes automatic understanding and automatic target recognition possible. So it has a great influence on production, living, medicine, weather and remote sensing, especially in the military, it's also leading a great prospect.Computer Vision uses computer to describe and analyze the outside world, while the research contents of low-level processes in computer vision are to recover 3D object information by edge detection, motion analysis, stereo vision and shape from texture, shading, contour (e.g., height, surface orientation,3D shape, etc).Getting the height can be the basis of Computer Vision, people have presented many methods. Essentially there are two main classes of integration techniques for finding height values from gradient values:local integration techniques and global integration techniques. The locality of the computations propagates errors, i.e. this approach strongly depends on data accuracy. Therefore, if the data is noisy, it's unsuitable to use this method. Comparing with the local methods, the results of global techniques are more smoother. Nevertheless, the global algorithm is very sensitive to the abrupt changes in orientation, i.e. there are large errors at the object boundary.3D Surface reconstruction based on Photometric Stereo can directly extract representation images from real images to construct highly realistic 3D surface, which has received much attention.Various kinds of 3D surface reconstruction techniques have been introduced, and then laser triangulation and photometric stereo techniques are used to reconstruct 3D surface. In the foundation of researching various integration methods, a local integration algorithm based on minimum spanning tree is put forward. A 3D surface reconstruction technique based on iterative cycle has been researched, focusing on how to capture 3D spatial information to reconstruct the highly realistic 3D surface. The image data acquisition, gradient data acquisition, laser height acquisition, integration methods, the method of height weighted average, resituating the light angle and gradient values have been researched in this thesis. Finally, the 3D surface reconstruction technique is applied to the 3D surface texture synthesis. This algorithm is tested by some image, the experiments show that this 3D surface reconstruction technique is feasible and effective.