The visual world is a complex interplay of light, encompassing a wealth of information beyond what we perceive with the human eye. Conventional cameras, inspired by human vision, are limited to measuring light intensity and color, thereby constraining our ability to capture the visual environment. The intensity and color of a scene are intricate functions of the scene's geometry, material properties, and lighting. Extracting these properties from RGB images alone is ill-posed limiting today's AR/VR and industrial monitoring systems. Moreover, the intensity captured is dominated by direct reflections constraining automotive and robotics systems to line-of-sight and making them susceptible to blind corners.

This thesis aims to transcend the limitations of traditional RGB imaging by delving into the underexplored potential of two additional light dimensions: polarization and time-of-flight. By incorporating polarization imaging, valuable insights into an object's surface properties and structural characteristics can be uncovered, enabling a deeper understanding of the visual scene. By leveraging time-of-flight distribution, objects hidden from the line of sight can be revealed, expanding the scope of vision systems. By employing emerging polarization and time-of-flight sensors, this thesis develops practical systems to uncover hidden information and enhance scene understanding.

The first part of this thesis focuses on the polarization-based vision. Techniques will be presented that leverage polarization of surface reflections to recover surface geometry and separate reflectance components. These techniques lead to accurate lighting estimation, novel view synthesis, and appearance editing. The second part of this thesis explores techniques utilizing the time-of-flight distribution, also called transients, for imaging objects hidden from line-of-sight. The wide array of applications demonstrated includes real-time change detection from a hidden region of interest at a rapid rate of 10 Hz, high-accuracy 3D imaging of the hidden region at centimeter-scale resolution, detection of specular targets in the concealed scene, and estimation of the material properties of hidden objects.

This thesis advocates the development of practical computational imaging systems utilizing emerging polarization and time-of-flight sensors. By harnessing additional dimensions of light, these systems have the potential to uncover hidden information and significantly enhance our understanding of complex visual environments.