Image Registration Based on Zernike SIFT

摘要：

Image registration is a key factor of computer vision, including object recognition and tracking. There have been many algorithms. Due to the invariance of scale, rotation and illumination, the Scale Invariant Feature Transform has been widely used in this area. In this paper, we combined the algorithm with Zernike moment to improve the accuracy and speed of matching. The way of getting the key point is the same with the Scale Invariant Feature Transform, but the difference is the descriptor which has the invariance of scale, rotation and illumination. The Scale Invariant Feature Transform used the rotation invariance property of the main gradient orientation, and generated the descriptor with the combined gradient orientation. The Zernike moment is rotation invariance, so we used it to generate the descriptor instead of the gradient orientation. Thus, we got the descriptor in one step instead of two steps in the Scale Invariant Feature Transform. The experiment results show that the combined algorithm is faster and more accurate then the SIFT.

关键词： Image Registration Key points SIFT Rotation Invariance Zernike Moment

作者: Jia Liu Caicheng Shi Meiguo Gao

作者单位: Beijing Institute of Technology Beijing,China

会议类型: 国际会议

会议名称: 2011 3rd International Conference on Computer and Network Technology(ICCNT 2011)(2011第三届IEEE计算机与网络技术国际会议)

会议地点: 太原

会议语种:英文

页码: 543-545

在线出版日期: 2011-02-26（万方平台首次上网日期，不代表论文的发表时间）