A Method of Region Segmentation in Grayscale Images Inspired by Physical Elastic Deformation
Xiaodong Zhuang, Nikos E. Mastorakis, Jieru Chi, Hanping Wang · 2017
For digital image analysis, a virtual grid method is proposed, where the pixels in image are mapped to the grid nodes. Inspired by physical elastic deformation, two kinds of interaction are defined between the adjacent nodes in the grid. First, the virtual elastic force is defined by imitating the force of a physical spring. Second, the virtual repulsive force is defined according to the grayscale difference between the two adjacent pixels. The virtual grid is simulated by computer, and a balance state can be reached given proper border condition. A novel method of image segmentation is proposed based on the position of each node in the grid at balance state. The method is compared to the watershed method, and the results show its specific advantage.