CT Images Segmentation Algorithm Based on Graph Cuts
Максим Привалов, Maksim Kazantsev · 2018 International Multi-Conference on Industrial Engineering and Modern Technologies (FarEastCon) · 2018
This paper describes comparison of the segmentation algorithms based on graph cuts and proposes hybrid algorithm with efficient GPGPU parallel implementation. Analysis of the Efficient Graph-Based Image Segmentation (GBS) and Parallel Heuristic for Minimum Spanning Forests (PHMSF) algorithms implementation has shown their efficiency on different segmentation stages. We propose to process CT images using graph cuts segmentation algorithm based on parallel GPGPU Minimum Spanning Trees (MST) building using OpenCL framework with further trees merging and segment labeling. Given the comparison with sequential and GPU-based implementations. According to the comparison of the algorithms segmentation results analyzed efficiency of the parallel variant and graph cuts algorithms themselves.