Fast Image Dilation/Erosion with Cooperation of Multiple Neighbors

This paper proposes an efficient algorithm to implement image dilation and erosion with multi-neighbor cooperation, which eliminates typical redundant operations through establishing a division function for k-neighboring nodes which operate repeatedly. This algorithm has an apparent advantage that it can optimize dilation/erosion operation of structural element in any shape, and concurrent operation to enhance performance remarkably. The method theoretically achieves higher level abstraction than Kernel Subdivision(KSD) algorithm and Reference Points Increment algorithm (RPI), and KSD and RPI included in this proposed algorithm as two specialties. Experimental results with an 11×11 round structure element showed that by taking 8 neighboring nodes as cooperation nodes, this algorithm performed 16 times faster than KSD in its worst case, as fast as KSD in its best case, and 26% faster than KSD in integrating various typical images.