Olmo Zavala-Romero, Graduate in Scientific Computing
This exhibit is a graphics processing unit (GPU) implementation of the Active Contours without Edges image segmentation algorithm. The main goal of the project was to accelerate this state of the art algorithm by programming it using OpenCL and expand the areas where it can be used. The original algorithm is able to separate images into two regions with different average intensities and is currently being used in different computer vision areas like medical image, tracking and pattern recognition.
This project runs up to 10 times faster than a sequential version implemented in Matlab and, being programmed also using OpenGL, it allows real time visualisation of the segmentation process and real time parameter modification. The code is Open Source and is available on GitHub at firstname.lastname@example.org:olmozavala/OpenCL-Active-Contours-without-Edges.git.