Abstract

Controlled light conditions improve considerably theperformance of most computer vision algorithms. Dynamiclight conditions create varying spatial changes in color andintensity across the scene. These condition, caused by amoving shadow for example, force developers to create al-gorithms which are robust to such variations. We suggest acomputational camera which produces images that are notinfluenced by environmental variations in light conditions.The key insight is that many years ago, similar difficultieswere already solved in radio communication; As a result each channel is immune to interference from other radio channels. Amplitude Modulated (AM) video camera sepa-rates the influence of a modulated light from other unknownlight sources in the scene; Causing the AM video cameraframe to appear the same independent of the light con-ditions in which it was taken. We built a prototype of theAM video camera by using off the shelf hardware and testedit. AM video camera was used to demonstrate color con-stancy, shadow removal and contrast enhancement in realtime. We show theoretically and empirically that: 1. theproposed system can produce images with similar noise lev-els as a standard camera. 2. The images created by suchcamera are almost completely immune to temporal, spatialand spectral changes in the background light.