Fast auto-focusing based on partial image characteristics

Abstract

Focusing is started with partial image characteristics and its region attained through Sampling Restraint Hough Transform. Both rough and fine focusing procedures are used to avoid partial interference caused by noise; the values of pixels in the 8-neighbouring area of each point are used in the improved gradient gray square function used as sharpness function; searching step is changed at real time in both rough and fine searching processes to enhance the focusing sensitivity; Field Programmable Gate Array is used to expedite the calculation of sharpness function. Experimental results indicate this way of focusing features fine singe peak, fast response, and good robustness; the caulcualtion of one frame image takes less than 20 ms, and auto-focusing is possible in a a range of 200μm with a uncertainty of less than 1μm.