关键词: 图像配准, Harris角点, 随机抽样一致性算法, 预检测, 随机块选

Abstract: In order to improve the speed of image registration, a fast method based on improved RANSAC algorithm is proposed. At first, Harris corner detector is used to extract the feature points in the reference and target image. Next, based on proximity and similarity of their intensity the feature points are matched. Finally, Improved RANSAC algorithm is used to estimate transform matrix more speed and accurately. To improve the speed of computation, a method of pre-detection is used to discard those temporary models which are not Preview Models. To delete outliers a random block selecting is used to select sample, which improve precisions of algorithm. The experiment shows that this algorithm reduces the amount of computation largely and improves the speed of image registration when the precisions have not notable change．

Key words: image registration, Harris corner, RANdom SAmple Consensus （RANSAC） algorithm, pre-detection, random block selecting