Description »

A new class of image features called "Ray features" were developped at CvLab. These features consider image characteristics at distant contour points, capturing information which is difficult to represent with standard features. This property allows Ray features to efficiently and robustly recognize deformable or irregular shapes, such as cells in microscopic imagery.

The ray features currently rely on a good estimate of the contours of a cell, which is currently done with the Canny edge algorithm. The objective of this project is to investigate how one could replace the use of Canny with a more direct way relying on the image gradient.

The second objective will be to investigate if one could consider several potential contour points to improve the prediction. For example, one could shoot several rays with different stopping criteria.