COMPUTER VISION SYSTEM AND METHOD EMPLOYING ILLUMINATION INVARIANT NEURAL NETWORKS

摘要

Object is classified, and is measured using normalized cross correlation (NCC), to compare two images obtained under conditions of illumination unevenness. Input pattern classification distributes tentative group indication and value. The output node input pattern is distributed to has largest classification value in radial primary function network. If input pattern and image are associated with the node, referred to as node image, both there is Uniform Illumination node image to be received, be then set above user-defined threshold value with probability. If test image or node image are uneven, do not receive the node image and classification value is remained into the numerical value that classifier is distributed. If both test image and node image are uneven, NCC is measured for being set as NCC values with above-mentioned classification value.