基于Hough谱的任意形状表示与配准  

Arbitrary Shape Representation and Registration Based on Hough Spectrum

作　　者：王彩玲[1] 赵春霞[1] 

机构地区：[1]南京理工大学计算机科学与技术学院,南京210094

出　　处：《中国图象图形学报》2010年第5期764-769,共6页Journal of Image and Graphics

基　　金：国家自然科学基金重点项目(60632050);国防基础技术研究项目(K1702020302)

摘　　要：提出了基于Hough谱的形状表示与配准方法,定义了一种新颖的加权平均Hough谱来表示任意形状,总结并证明了形状Hough谱所具有的平移、旋转、缩放和周期特性。对任意两个形状进行配准时,可在相应Hough谱中估计出形状的旋转与缩放参数,并在相应Hough密度空间中估计形状平移参数。本文方法在SIID形状集上进行了大量的实验,结果表明,加权平均Hough谱能有效地估计形状间的平移、旋转与缩放参数,对噪声、遮挡等具有一定的鲁棒性,具有较高的配准精度。A novel shape representation and registration method based on Hough spectrum is developed in this paper. The weighted-average Hough spectrum （WAHS） is defined for the representation of arbitrary shapes, and the translation, rotation, scale and periodic properties of WAHS are summarized and proved. When registration, the parameters of rotation and scale can he estimated easily from the phase and amplitude of WAHS, and the parameter of translation can be estimated in the Hough density space. Experiments show that the method is effective to estimate the parameters of translation, rotation and scale, and is immune to noise and overlapping.

关 键 词：形状表示 形状配准 加权平均Hough谱 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]