Ronan Kerviche

I am a Ph.D. student working in Dr. Ashok’s Intelligent Imaging and Sensing Lab. My research activity focuses on Computational Optics, a paradigm for designing joint imaging and digital processing systems which can bypass traditional architectures shortcomings. With this perspective, I have helped develop a F/1 thermal camera with Extended Depth of Field : we optimized a large set of specific aberrations to be induced by a phase mask in the optical train. Such imaging system delivers soft images independently from the distance to the object and in a way which is revertible by a linear algorithm. The whole system can thus generate sharp images over a wider range of distances.

More recently, I participated in the construction of a compressive imager prototype. This atypical camera relies on the sparsity/redundancy of natural scenes to perform fewer measurements than a standard imager. Hence, it alleviates the constraints on the required detectors count for the acquisition and the bandwidth for reading-out and transmitting the data stream. The system uses a Digital micro-Mirror Device, commonly found in video projectors, to spatially modulate the optical signal with optimized patterns. The resulting field is integrated onto a low resolution focal plane array. We developed a non-linear single-pass algorithm to reconstruct the image in real-time from these compressed measurements. This architecture can also be tailored to specific tasks, such as the detection and classification of objects, while being less sensitive to detector noise and consuming less power than traditional camera modules.

Figure 1: (left) 80 × 80 pixels image of a Siemens star acquired by a conventional camera. (right) image acquired by a compressive imager with the same 80 × 80 detectors sensor (at 4× compression).