Saliency-based Active Contour Model for Image Segmentation and Region Detection
Aditi M. Joshi, Mohammed Saquib Khan, Usman Asim, Asad Munir, Hyun Chul Song, Kwang Nam Choi · 2021 International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS) · 2021
In this paper, we propose a novel saliency-based active contour model for image segmentation and region detection (SAIR) to overcome the problems of noise and intensity inhomogeneity in image segmentation. The proposed new level set protocol evolves adaptively and eliminates the need for initialization. In the proposed energy function, we formulate an adaptive weight function that adaptively changes the intensity of the internal and external energy functions according to the image. Moreover, according to the inner and outer regions, modulating the signs in the proposed energy function influences the elimination of noise in the image. Finally, SAIR is tested on multiple images with different initial contour positions, intensity inhomogeneity, and noise to demonstrate the robustness of SAIR.