Block matching in rotated images

A rotationally invariant block-matching algorithm for matching point features in rotated images is presented. The algorithm matches the detected point features by extracting a small image area (block) centred on each point in the first image and searching for a highly similar block in the second image. The algorithm compensates for the rotation difference between the compared blocks using the gradient orientation difference between the matched point features. Application of the proposed algorithm to estimate global rotation difference between rotated images is outlined and the results are compared to other widely used algorithms.