Automatic Multilevel Thresholding for Image Segmentation Based on Block Sampling and Genetic Algorithm

Multilevel thresholding is an important technique for image compression,image analysis and pattern recognition.However,it is a hard problem to determine the number of thresholds automatically.In this paper,a new multilevel thresholding method called as automatic multilevel thresholding algorithm for image segmentation based on block sampling and genetic algorithm(AMT-BSGA) is proposed on the basis of block sampling and genetic algorithm.The proposed method can automatically determine the appropriate number of thresholds and the proper threshold values.In AMT-BSGA,an image is treated as a group of individual pixels with the gray values.First,an image is evenly divided into several blocks,and a sample is drawn from each block.Then,genetic algorithm based optimization is applied to each sample to maximize the ratio of mean and variance of the sample.Based on the optimized samples,the number of thresholds and threshold values are preliminarily determined.Finally,a deterministic method is implemented to further optimize the number of thresholds and threshold values.AMT-BSGA can work without prior knowledge on other auxiliary information,such as contextual or textual properties,and the number of thresholds.It is low in computing complexity which is almost independent from the number of thresholds and can avoid the burden of analyzing histograms.AMT-BSGA can produce effective,efficient and smoother results,which has been verified by extensive simulations on Berkeley datasets.