An Approach on Analyzing Histogram and Selecting Threshold

On the basis of observing good valley of many bimodal histograms, this paper defined three guide lines about valley and finished the automatic analysis of valley. Then, the paper established the algorithm of selecting good valley threshold by simulating human?s thinking process. This paper proposed a method on analysis of histogram and selection of threshold. This technique is not sensitive to whether bimodal histogram is distinct and whether the depth of valley is deep. Also, it is not sensitive to small fluctuation of burrs. The valley picked by principle above is same to the valley picked by artificial selection. Therefore, this method could achieve good effect of segmentation for image possessing bimodal histogram. Comparing with Otsu?s method, this proposal could get better effect of segmentation. And experiments proved that this method is effective and accurate.