Visual Behaviour Based Bio - inspired Polarization Techniques In Computer Vision and Robotics Le 2 i - UMR CNRS 5158

Polarization is the phenomenon that describes the oscillations orientation of the light (or other radiation) waves which are restricted in direction [1]. Light’s vector orientation can be weekly detected by some humans with their naked eyes, but humans need the help of polarizing optics to visulize most invisible polarization effects [2]. Many fish, cephalopods, crustaceans, insects, and other animals are capable of perceiving polarized light [3]. Most animal photoreceptors are able to differentially react to partially linearly polarized light [4], [5], [6], [7]. Fish [8] and birds [9] photoreceptors respond to polarized light patterns and hence are able to analyse linear polarization [10]. For long time, it was thought that sensing of polarization in animals is invariably related to their behaviour like navigation and orientation. Recently, it was found that polarization can be part of a highlevel visual perception, permitting a wide area of vision applications. Polarization vision can be used for most tasks of color vision like : object recognition, contrast enhancement, camouflage breaking, and signal detection and discrimination. [10]. In the following sections, the polarization based visual behaviour in the animal kingdom will be briefly discussed. Then, we will touch on some of these techniques. Finally, our future view of polarization vision in autonomous robots will be illustrated and the work concluded. Our aims are: 1) to have an overview of biologically polarization-inspired techniques in computer vision and robotics, 2) to share our ongoing research toward having a visual omnidirectional polarization (pola-catadioptric) sensor for autonomous robots to take those techniques into the omnidirectional world.