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.
出处
《计算机应用研究》
CSCD
北大核心
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)
