Olivier LIM

Abstract: The thesis 'Real-Time unconventional adaptive cameras for multimodal acquisition' is about the design of new cameras able to acquire information otherwise impossible to recover from a conventionnal camera.
The idea is to get inspiration from the visual system of the jumping spider, able to catch mosquitos thanks to 3D perception in addition to visual information. 3D perception of the jumping spider is enabled by a specific cornea together with a special layer of the retina and associated neural processing. The cornea adds in-purpose chromatic aberrations which are acquired with some defocused retina cells. Then the jumping spider 'calculates' 3D from this defocused image and other visual informations.
The purpose of this thesis is to study new algorithms to enable multi-modal cameras in the context of simultaneous acquisition of standard images, 3D and hyperspectral modalities.
The goal is to design new systems able to adapt the optical system to the acquired scene, in real time, and provide informations to the uses.
The focus application is medical laparoscopic surgery, where the surgeon operates trough cameras placed in the body. Acquisition of 3D and hyperspectral information is crucial to help to distinguish organs and precisely locate instruments in the body.