Research On 3d Object Surface's Shape Recovery Based On A Single Image's Gray-scale

3D shape recovery technique is a classical problem in the field of the computer vision, it refers to the technique which use a single or multiple 2D gray-scale images to derive the 3D surface of the object shape information. It has played more and more noticeable role in the fields of social productive and life. This technique has widely application and potential value especially in agriculture, industry, national defense, medicine and space technology field. SFS (Shape from Shading) technology is utilized to recovery the object's surface which the brightness changes gradually related to the height of various points or surface gradient and other parameters, with simple operation, widespread application and flexible benefits, there are a bright prospect and much value. It has been a hotspot for some researcher of relative field.The dissertation details the theory of the the SFS algorithm, and conclude systematically some classical SFS techniques, that is, minimization approaches, propagation approaches, local approaches, linear approaches. According to these four algorithms on different assumptions, we analyze their applicative aspects, and compare with the convergence speed, solution uniqueness, existence of solutions of these four algorithms.Most traditional SFS algorithms need harsh assum Ptions that don't correspond the real condition and assume some unsuitable condition so that can not get precise and universality 3D construction result, on the basis of the existing SFS algorithms, we get further improving the principle of three-dimensional shape recovery from the single image, mainly the operations of the image pre-processing and the estimation of illumination parameters are both added. The algorithm uses some techniques of image pre-processing such as the image smoothing, image segmentation and the processing of the mathematic morphing to be optimized, then estimates the tilt and the slant angle of the light source, afterwards, it recoveries the 3D description, which is needed, based on lambertian illumination model. During the period of the Implementation, we also do some programs to prove three-dimensional surface shape recovery technique based on single gray-scale image by the MATLAB software. The results of the examples for Three-dimensional surface shape recovery and its precision analysis is also discussed. The experiment is proved that this algorithms leads to a good result of the recovery of 3D object's surface. It also verified that the algorithm of this dissertation is effective, precise and of utility value.