Detection of Earth Surface Cracks Using Parallel Genetic Algorithm Based Thresholding

A parallel genetic algorithm (PGA) based scheme for detection of cracks in images is proposed. The proposed segmentation scheme is a feature based one and the optimum threshold is determined from the the feature histogram. Parallel genetic algorithm (PGA) based clustering is proposed to detect two peaks and thereafter the optimal threshold is determined by minimum mean square error (MMSE) based strategy. The proposed method has been tested successfully with various images. The performance of the proposed method is compared with that of Otsu's, Kwon's, feature less(FL) and feature based(FB) approaches and it is found that the proposed method outperformed all these methods for crack detection.