GPU and FPGA Acceleration of Level Set Method

Abstract

The level set method is one of the most powerful image segmentation methods. Its computational complexity, however, is very high, and many approaches to reduce the computation time have been proposed. In this paper, we describe a new level set algorithm for parallel processing, and its implementation on GPU and FPGA. The computational complexity of this algorithm is higher than previous algorithms, but it is possible to achieve higher performance by parallel processing. We implemented the algorithm on GeForce GTX780Ti, and Xilinx XC7VX485T, and compared their performances.