基于图像金字塔分解的快速全局运动估计  

A fast global motion estimation algorithm using image pyramid decomposition

作　　者：齐美彬[1,2] 王宝[1] 蒋建国[1,2] 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009 [2]合肥工业大学安全关键工业测控技术教育部工程研究中心,安徽合肥230009

出　　处：《合肥工业大学学报（自然科学版）》2011年第9期1335-1340,共6页Journal of Hefei University of Technology：Natural Science

基　　金：安徽省科技计划资助项目(08020303095)

摘　　要：文章提出了一种基于图像金字塔分解的快速全局运动估计算法。为了提高计算速度与准确性,采用了3层金字塔进行多分辨率计算。在金字塔第3层图像中,进行宏块预判,去除不可靠宏块,通过块匹配算法粗略地求出摄像机运动参数,并将运动参数映射到金字塔第2层。在金字塔的前2层图像中,将宏块预判与基于块的外点去除算法相结合,去除不可靠宏块,并对剩余像素点抽取特征点,最终精确地求出摄像机运动参数。在每层金字塔图像中,参数的计算采用了双迭代的最小二乘法,进一步提高了计算的准确性。实验表明,该算法在速度和准确性上有一定的提高。This paper presents a fast global motion estimation algorithm based on image pyramid decomposition.In order to improve the calculation speed and the accuracy,a three-level pyramid is utilized in the multi-resolution calculation.In the third pyramid level,the blocks are prejudged and the unreliable ones are removed.The camera motion parameters are obtained by block-matching and mapped to the second level.In the former two pyramid levels,the unreliable blocks are removed by the algorithm of block prejudgment and blocks outlier detection,and the feature pixels in the remaining pixels are selected.Finally,the camera motion parameters are accurately calculated.In each pyramid level,the least squares method of double iteration is used to calculate the parameters,and the accuracy of the calculation is further improved.Experimental results show that the algorithm is superior to original algorithms in velocity and accuracy.

关 键 词：全局运动估计 宏块预判 外点去除 特征点提取 双迭代 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]