Abstract: A new sharp method by recognizing pixel characteristics is presented in order to reduce noise sensitivity of traditional unsharp masking technique.Every pixel of the image is classified into noise,detail,small edge,big edge and huge edge point according to different high frequency component.Then de-noising operation is performed to noise pixel.The sharpening factor varies automatically based on high frequency component.Experimental results show that image noise can be reduced,image sharpness can be improved effectively without edge overshoot.