In this thesis, we address the problem of reflection and illumination analysis from images. The appearance of the world is determined by the interaction be- tween light and matter. Changes in illumination conditions or surface reflectance properties can lead to significant variations in the appearance of a scene. Specular reflections also contribute to the diversity of world appearance. Understanding reflection and light has been one of the longest standing research areas in physics and human perception. The analysis of reflection and illumination from images has also become an important and interesting problem in computer vision and graphics research. In this thesis, we focus on the following three subproblems: diffuse/specular separation, reflectance estimation, and illumination estimation from images. First, we propose a method based on color analysis and multibaseline stereo that simultaneously estimates the diffuse/specular separation and the true depth of specular reflections. This method overcomes the limiting assumptions of the previous approaches, such as accurate segmentation, regionally uniform reflectance or structured lighting. Secondly, we propose an iterative method that aggregates BRDF data and estimates reflectance values in a single image by soft grouping, where pixels con- tribute to one another's estimate according to their degree of reflectance similarity. Estimation of specular reflectance is further improved by albedo-independent soft grouping of pixels based on shape continuity. Thirdly, we propose a method that integrates cues from shading, shadow and specular reflections for estimating directional illumination in a textured scene. Unlike previous works which all assume known or uniform reflectance, our method can deal with the effects of textures by capitalizing on physical consistencies that exist among the lighting cues. A technique for texture-independent estimation of light source intensities is also presented. Experiments on synthetic and real scenes were carried on to validate our approaches.
修改评论