Sub-pixel registration based on phase correlation and its application to electronic image stabilization
-
摘要:介绍一种基于相位相关的图像配准方法,并将该方法应用于序列图像的运动检测,通过运动补偿,实现了序列图像的稳定输出。采用相位相关配准方法对序列图像进行亚像素精度配准,运用基于最小二乘的曲面拟合法,估计当前图像相对参考图像的亚像素级的平移量。在补偿图像运动时,为保证亚像素级的补偿精度,采用平滑算法,避免在图像补偿时出现马赛克现象。最后,对样本图像进行了亚像素级位移配准和图像补偿对比实验。实验结果表明,提出的方法可以检测到0.01 pixel的运动量, 最大配准误差为0.008 pixel;采用亚像素级运动量补偿图像,最大误差<0.5。Abstract:A novel method for the sub-pixel image registration based on phase correlation was introduced, which could be used to detect the movements of video sequences exactly, and could obtain the video outputs stably. Firstly, two images translated based on the phase correlation technique were registered, and the paraboloid surface fitting method based on least-square was used to estimate displacement parameters of the images. Then, the sub-pixel motion vector between the two sequential images could be exactly detected. A smooth algorithm in transform formula of the sub-pixel motion compensation was adopted to avoid the mosaics of images and to improve the precision of image compensation. In order to verify the algorithm in experiments, some sample simulations were performed. The results show that this method can detect the shift parameters accurately and efficiently, which makes it possible to estimate image displacement with 0.01-pixel accuracy. Further more, the maximum error of registration is only 0.008-pixel, and the root of mean square error of image compensation is also less than 0.5.
-
[1] KUGLIN C D,HINES D C. The phase correlation image alignment method .Proc. Int. Conf. Cybernetics Society.San Francisco,USA,1975:163-165. [2] FOROOSH H,ZERUBIA J,BERTHOD M. Extension of phase correlation to sub-pixel registration[J].IEEE Trans. Image Processing,2002,11(3):188-200. [3] 桂志国,韩焱. 相位相关配准法及其在射线图像数字减影中的应用[J]. 仪器仪表学报 ,2004,25(4):520-522. GUI ZH G,HAN Y. Phase correlation technique for image registration and its application in dsa of radiographic image[J].Chinese J. Sci. Instrum.,2004,25(4):520-522.(in Chinese) [4] 高世海,戴文刚,田仲. 空域和频域相结合的景象匹配算法[J]. 中国图象图形学报 ,2000,5(A)(4):345-348. GAO SH H,DAI W G,TIAN ZH. Hybrid image registration algorithm of space domain and frequency domain[J].J. Image Graphics,2000,5(A)(4):345-348.(in Chinese) [5] 吴元昊,于前洋. 基于傅里叶相位差的抗噪声位移估计算法[J]. 光学 精密工程 ,2007,15(7):1137-1142. WU Y H,YU Q Y. Denoising algorithm of displacement estimation based on phase difference of Fourier transform[J].Opt. Precision Eng.,2007,15(7):1137-1142. (in Chinese) [6] 冯林,管慧娟、藤弘飞. 基于互信息的医学图像配准技术研究进展[J]. 生物医学工程学 ,2005,22(5):1078-1081. FENG L,GUAN H J,TENG H F. Advances in medical image registration based on mutual information[J].J. Biomedical Eng. Res.,2005,22(5):1078-1081.(in Chinese) [7] 周永新,罗述谦. 基于形状特征点最大互信息的医学图像配准[J]. 计算机辅助设计与图形学学报 ,2002,14(7):654-658. ZHOU Y X,LUO SH Q. Medical image registration based on mutual information of feature points[J].J. Computer-Aided Design Computer Graphice,2002,14(7):654-658.(in Chinese) [8] 李治国,安锦文,赵银铃. 基于相位相关的全局运动估计算法[J]. 计算机测量与控制 ,2008,16(4):532-534. LI ZH G,AN J W,ZHAO Y L. Global motion estimation algorithm based on phase correlation[J].Computer Measurement Control.,2008,16(4):532-534.(in Chinese) [9] 黎俊,彭启民,范植华. 亚像素级图像配准算法研究[J]. 中国图象图形学报 ,2008,13(11):2071-2075. LI J,PENG Q M,FAN ZH H. Asurvey of sub-pixel image registration methods[J].J. Image and Graphics,2008,13(11):2071-2075.(in Chinese) [10] 徐宝昌,陈哲,赵龙. 一种改进的最小二乘景象匹配算法[J]. 北京航空航天大学学报 ,2005,31(8):848-852. XU B CH,CHEN ZH,ZHAO L. Inproved least-squares acene matching algorithm[J].J. Beijing University Aeronautics and Astronautics,2005,31(8):848-852.(in Chinese)
点击查看大图
计量
- 文章访问数:4017
- HTML全文浏览量:384
- PDF下载量:2331
- 被引次数:0