Description

Abstract

Ø We have presented a level set image segmentation method based on the idea of stopping the evolving contour according to the degree of membership of the active pixel to be inside or outside of this evolving contour.

Ø It is done with the help of the FCM (Fuzzy Class Mean) partition matrix. The LSE (Level Set Equation) is solved by using the powerful, simple, and highly parallelizable LBM which allows the method to be a good candidate for GPU implementation.

Ø The method gives promising results. Experimental results on medical and real-world images have demonstrated the good performance of the proposed method in terms of speed, effectiveness in the presence of intensity in homogeneities, accuracy, robustness against noise, and efficiency whatever the initial contour.

Ø The proposed method gives better result than the Gibou-Fedkiw method whatever the shape and the position of the initial contour.

Ø It is fast when running it one time, the segmentation will take more much time since one should run the method several times.

Ø We can also remark that the proposed method can give better results than the CV method and can detect all the contours whatever the position of initial contour, while in the CV method, the contour can be trapped in a local minimum.

Ø Furthermore, the proposed method is more than 100 times faster than the CV method and can be faster when implemented on graphics processing unit (GPU) due to the local and explicit nature of the LBM solver.