Study On Rotation And Scale-Invariant Image Matching

Image matching is a fundamental problem in image processing. It is very difficult to match correctly between two different images owing to the existence of transformation, rotation or scale changes, etc. Aiming at these problems, this thesis does some researches based on intensity matching and feature matching respectively.First, the thesis presents a novel grayscale template matching algorithm, which is invariant to rotation, scale, brightness and contrast. Different from other techniques, it does not discard the rich grayscale information. The algorithm consists of three cascaded filters, i.e., circular sampling filter, radial sampling filter and template matching filter. Each filter successively excludes pixels that have no chance of matching the template from further processing. The algorithm is faster than the original brute force algorithm while obtaining the same results.Second, a new image matching algorithm that matches geometrically similar triangles is proposed. The Harris feature points are extracted from two images respectively at the beginning, and the triangles are formed from triplets of points. By limiting the number of triangles constructed and by only matching those triangles whose ratio of the length of the triangle's intermediate side to longest side are within a defined limit, the matching phase has a great improvement on computational complexity compared to traditional triangle-matching approach.Last, the thesis studies a new and promising approach, Scale Invariance Feature Transform (SIFT) , as a local detector and descriptor, and evaluates how well available implementations of SIFT in terms of invariance.