The fringe distance measure: an easily calculated image distance measure with recognition results comparable to Gaussian blurring

摘要

A fast simple distance measure for calculating the degree of dissimilarity between binary images is presented. The new distance measure, called the fringe distance, gives the same degree of distortion tolerance as optimized Gaussian blurring, but can be calculated much more rapidly. Template matching using the fringe distance executes about six times as fast as Gaussian blurring template matching on an IBM AT with coprocessor, and more than three times as fast on a SUN SPARCStation I with floating point accelerator. The greater speed means more templates can be used in process time limited applications, increasing recognition accuracy.