To overcome the drawback that the information carried by an image is inconsistent with the statistic data in conventional histogram,we proposed a new form of histogram whick was achieved by compressing the probabilities of gray levels in smooth areas based on the factor of local spatial activity.At the same time,excessive enhancement of noise and ringing artifacts may occur if there are high peaks in the histogram,because the difference between two neighboring gray levels is proportional to the height of the histogram during histogram equalization.A clipped histogram method was developed,in which the dynamic ranges of the major gray levels were constrained based on the optimum-noticeable-difference(OND).Experimental results show that the proposed algorithm has better performance than the existing methods,noise over-enhancement and more robust.