Automatic threshold confirmation for extracting feature points in 2-dimensional images

Abstrcat According to the different distributions on the gray surface between noise and feature points,the location of feature point was located by recursively computing deviation of gray-level within the sampled area.The threshold value was ascertained by the arithmetical average method.The limitation to confirm threshold value with tentative method and a priori information on processing image would be avoided.Experimental results of extracting feature points on different natural 2-D images proved that the threshold value determined was reasonable and effective.