基于ZM相特征描述符的图像配准方法  

Method of image registration based on ZM phase feature description

作　　者：郑寇全[1] 杨文静[1] 张继周[2] 雷英杰[3] 魏强[1] 

机构地区：[1]西安通信学院,西安710106 [2]西安财经学院信息与教育技术中心,西安710100 [3]空军工程大学防空反导学院,西安710051

出　　处：《计算机应用研究》2017年第1期279-282,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61272011;61309022);国家重点实验室基金资助项目(2012ADL-DW0301);陕西省自然科学青年基金资助项目(2013JQ8031)

摘　　要：针对图像特征间的视角、模糊、旋转和尺度变换以及噪声干扰等不确定性因素对图像配准精度的影响,定义了基于Zernike矩的相特征不变区域描述符,给出了Zernike矩的归一化及其相值度量方法,并提出了一种基于此的图像配准算法。通过对比实验验证,算法能够有效提取目标图像特征点并准确进行特征匹配,且在保障配准精度的前提下显著降低了算法复杂度。In view of the image characteristics of the uncertainty factors such as angle, blur, rotation, scaling transform and noise, which affected the precision of image registration, this paper defined the distinctive image descriptor based on the Zernike moment （ZM） phase information, provided the ZM normalization and its measurement method. Meanwhile, this paper established a new image registration algorithm. Simulation experiments show that the proposed algorithm can effectively extract the image feature points and match the feature accurately, and reduce the algorithm complexity significantly for a prescribed matching accuracy.

关 键 词：图像配准 ZERNIKE矩 相特征 特征描述符 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]