NYI: Physical-Based Computer Vision: Polarization Vision, Reflectance Modeling and 3-D Vision

这项研究的大部分继续发展偏振视觉的图像理解的基本原理，这是首席研究人员在博士研究生时发起的。这项研究还继续开发偏振相机，这是一种新型的传感器，可以自动检测偏振并计算偏振信息的可视化，或与偏振有关的场景属性。除了强度之外，光的偏振的额外物理维度从场景中携带了额外的信息，可以为理解图像提供一组更丰富的描述性物理约束。主要研究人员过去的工作揭示了偏振视觉技术可用于执行介电/金属材料识别、镜面反射和漫反射分量分离和分析，以及仅使用强度和颜色信息进行复杂甚至不可行的复杂图像分割。推广了朗伯定律，建立了光滑介质表面的强度漫反射模型。人们正在研究这种更精确的漫反射模型对计算机视觉中从反射中恢复三维形状的影响。正在开发利用光度值对应从双目立体中恢复三维形状的新技术。

Much of this research continues the development of the fundamentals of Polarization Vision for image understanding which the principal investigator initiated as a doctoral student. This research also continues the development of Polarization Cameras which are a new class of sensors that automatically sense polarization and compute either a visualization of polarization information, or, properties of a scene that are physically related to polarization. The additional physical dimensions of polarization of light beyond that of intensity, carry extra information from a scene that can provide a richer set of descriptive physical constraints for the understanding of images. Past work by the principal investigator has revealed Polarization Vision techniques that can be used to perform dielectric/metal material identification, specular and diffuse reflection component separation and analysis, as well as complex image segmentations that would be immensely more complicated or even infeasible using intensity and color information alone. A new intensity diffuse reflectance model has been formally developed for smooth dielectric surfaces that generalizes Lambert's Law. Implications of this more accurate diffuse reflectance model are being studied for 3-D shape recovery from reflectance in computer vision. New techniques are being developed for 3-D shape recovery from binocular stereo using correspondence of photometric values.