Info · Research · Publications · Teaching · Resume
It has been shown that white light exposure during retinal microsurgeries is detrimental to patients. To address this problem, a new device has been developed which can be used to significantly reduce the amount of phototoxicity induced in the eye. This device alternates between illuminating the retina using white and limited spectrum light, while a fully automated image processing algorithm produces a synthetic white light video by colorizing non-white light images. In order to use such a device a new family of algorithms need to be developed to efficiently fuse imaging modalities in this application. While the techniques used to solve this problem are presented for this given surgical setting, other clinical settings may benefit from such concepts. Here are some of our recent publications on the topic:
This work is in collaboration with Prof. Gregory Hager and Prof. Russell Taylor .
A proposed strategy for computational shape recognition argues that the task of visually recognizing an object can be accomplished by querying the image in a sequential and adaptive way. This twenty questions approach can be described as follows: there is a fact to be verified, (e.g., is there a face in the field of view ?) and each query, which consists in evaluating a particular function of the image, is chosen to maximally reduce the expected uncertainty about this fact. In the context of computer vision, such approaches have led to two different types of search algorithms: offline and online. In the offline versions, the "where to look" next strategy is computed once and for all, anticipating all possible queries. In the online version, the strategy is computed sequentially as information is gathered: This approach is known as Active Testing. This online strategy has the main advantage of being capable of searching over much larger spaces than its more traditional offline counterpart. Over the last few years, I have investigated several aspects surrounding this approach. Here is our latest publication on the topic:
This work is in collaboration with Prof. Bruno Jedynak and Prof. Peter Frazier .
A fundamental understanding of the locomotion of living organisms is of great practical and scientific interest, and may lead to useful applications spanning fields such as robotics, medicine, and biology. One pertinent illustration is given with the nematode C. elegans. This small, 1 mm long, roundworm is widely used as a model system for biological research; its genome has been completely sequenced. For example, phenotypes of C. elegans motility are frequently used to identify genes involved in muscle function and model aspects of human Muscular Dystrophy. However, traditional methods for motility assessment have been driven largely by qualitative assays based on observation. For this reason, we are developing a series of image processing algorithms to facilitate quantitative motility analysis. Here are some of our recent publications on the topic:
This work is in collaboration with Prof. Josue Sznitman.
This work had for objective to build a classifier for images of cats, and focused on particularly difficult images. Two approaches were implemented; one was using the Chamfer distance metric for matching, and the other learned models of our training images and used the Kullback-Leibler divergence as a metric of image comparison. A report on the implementation specifications and the results of the experimentations are presented in this report.
Worked on this at the IDIAP Research Institute with Dr. Francois Fleuret
Project was done with Giancarlo Troni