The Model of Moveably Resampled Template Matching

When register a pair of remotely sensed images with very different spatial resolution, the diversity of their shape and tone results in lower precision and success rate. The paper proposes an algorithm of moveably resampled template match to fine register them, which resamples the high-resolution images to a sub-image by a movable mask plate, and then uses template matching method to register the sub-image and low-resolution image. Experimental results show that its error is less than 0.1 pixel.