Diffuse Reflection Filtering Based On Reflectance Polarization Analysis

In the field of computer vision, the ability to distinguish specular reflection from diffuse reflection is very important .We can obtain the specular reflection component by separating the reflection components, and more determine the surface normal of the object and recover surface shape of the moving objects.Reflected light from the surface of the object has the definite polarization and echoes the object's shape and the reflectance properties dissimilarity. In the paper, firstly, we will analyze the light rays of common objects' surface reflection which were broadly classified into two categories: surface reflection and body reflectance. Secondly, we will analyze the surface's roughness of the glossy opaque which was researched, bound the range of the roughness and detain the reflectance framework. We analyze the mostly reflectance components of an object's surface that we are studying, as specular reflection and diffuse reflection component, detain the simplified reflectance radiant intensity formula by using polarization and two reflected models based on physical and geometrical optics. The reflectance parameters are determined by using the least-square-method, and the surface reflected model is obtained. Physically-based rendering systems describe reflection behavior using the bidirectional reflectance distribution function (BRDF). In the paper, we present a new BRDF model that is simplified for the glossy opaque surface's reflection. Finally, we obtain the specular component image by separating the reflection component based on polarization.The result of the experiment indicates the effective and robusticity of the separation method, obtains the specular reflection of object successfully from the separation of reflection, and proves the reliability of the BRDF mathematic model.