Matching by pixel distribution comparison: Multisource image template matching

Efficient and accurate template matching in multisource images is a difficult task when hardware platform resources are limited. Motivated by the problems encountered present methods, scheme, called Matching Pixel Distribution Comparison (MPDC), proposed to reduce use of system combining traditional algorithms neural networks. The MPDC scheme first extracts distribution information image pixels through developed extended Slice Transform (eSLT) matrix overcome non-linear intensity difference between heterogeneous images. Then similarity evaluated comparing pixel rules reflected eSLT matrix. A score table constructed based on comparison, pairs can be calculated querying this table. comparison matrices completed either experience scoring or network learning. experimental results show that, with limited resources, 64 × matched sliding 256 query 2.85 s, it only takes at least 19.84 s for training CPU. performance also better than many popular algorithms.