Software speedup techniques for binary image object recognition

The use of template matching methods to locate objects in large images is very computationally expensive. The authors describe, for binary images, a coarse to fine technique which speeds up the column and row position search by a factor of 10 or more. A matching error function is used to switch between coarse and fine search modes. Image brightness differences between template and image object will affect matching accuracy and converting template and image to binary format reduces this problem. The authors also describe what they call a sparse template technique and how it is used to give up to a 64-fold speed-up or even more for larger templates. This work is of importance in vision guided assembly operations where machine vision techniques are used for locating parts