基于Fourier Mellin变换和曲面拟合的IC芯片图像亚像素配准算法

PDF全文下载：2016050573

许少娟1 ， 王演2， 王鲁云1， 张晶泊2

(1.大连理工大学 城市学院，辽宁 大连 116600； 2.大连海事大学 信息科学技术学院，辽宁 大连 116026）

摘要： 为实现IC芯片图像的精确配准，提出一种FourierMellin变换和曲面拟合法结合的图像亚像素配准算法。算法首先采用FourierMellin变换计算出待配准IC芯片图像的旋转角和平移量的整像素峰值位置，然后对峰值区域进行基于最小二乘的曲面拟合，通过取极值点获得亚像素级的精确峰值位置。另外，FourierMellin变换由于旋转的频谱混叠和坐标变换中的重采样插值会产生误差，新算法采用加窗和滤波的方法加以改进。实验结果表明，所提算法实现了IC芯片图像的亚像素级配准，具有较好的精度和鲁棒性。

关键词：  图像配准； FourierMellin变换； 曲面拟合； 亚像素

中图分类号： TP 391.41文献标志码： A

IC Chip Image Subpixel Registration Algorithm Based on FourierMellin Transform and Surface Fitting

XU Shaojuan1，  WANG Yan2，  WANG Luyun1，  ZHANG Jingbo2

(1． City Institute, Dalian University of Technology, Dalian 116600，China;2．School of Information Science and Technology, Dalian Maritime University, Dalian 116026，China)

Abstract: To achieve IC chip image accurate registration, a new image subpixel registration algorithm which combines FourierMellin transform and surface fitting method is proposed. The new algorithm first calculates the peak position of rotation angle and translation of the matched IC chip image via using FourierMellin transform. Then, the peak area is fitted by using least squares method, and the accurate peak position which is a subpixel value is obtained through calculate extreme points. In addition, spectrum aliasing and resampling interpolation during coordinate transformation will bring errors by applying FourierMellin transform. The proposed algorithm resolves this problem by adding window function and filtering. Simulation results show that the proposed algorithm accomplished IC chip image subpixel registration with optimal robust and accuracy.

Key words: image registration； FourierMellin transform； surface fitting； subpixel

 收稿日期：    20160205

基金项目： 国家自然科学基金项目(61501078).

作者简介： 许少娟(1981—)，女，讲师.

文章编号： 16726987（2016）05057307； DOI： 10.16351/j.16726987.2016.05.019