Improved Dripping Algorithm Based on Modeling the Backflow and Splashing of Water Droplets

The water droplet’s dripping path is simulated by dripping algorithms and applied in many fields. In image segmentation, droplets directed by a traditional dripping algorithm may either be trapped in a dead cycle or forced traversing obstacles. To solve this problem, we first propose an improved dripping algorithm based on modeling the backflow of a water droplet. Instead of trapping, the improved algorithm guides the water droplet flow back for dripping opportunities. When blocked by both sides like falling into a low-lying “shallow pit”, we further propose to improve the dripping algorithm based on modeling the splashing of water droplets. This improvement allows water droplets to splash one pixel higher for escaping the “shallow pit”. Experiments in image segmentation of Guqin music notations validate the proposed method and demonstrate its advantages over traditional dripping algorithms.