[1]徐姗姗,刘应安,徐昇.针对图像区域连续化问题的立体匹配算法[J].南京师范大学学报(工程技术版),2011,11(04):047-52.
　Xu Shanshan,Liu Yingan,Xu Sheng.Continuous Problem of Image Region Stereo Matching Algorithm[J].Journal of Nanjing Normal University(Engineering and Technology),2011,11(04):047-52.
点击复制

针对图像区域连续化问题的立体匹配算法
分享到：

南京师范大学学报（工程技术版）[ISSN:1006-6977/CN:61-1281/TN]

卷:
11卷
期数:
2011年04期
页码:
047-52
栏目:
出版日期:
2011-12-31

文章信息/Info

Title:
Continuous Problem of Image Region Stereo Matching Algorithm
作者:
徐姗姗; 刘应安; 徐昇
南京林业大学信息科学与技术学院，江苏南京210037
Author(s):
Xu Shanshan; Liu Yingan; Xu Sheng
College of Information Science and Technology,Nanjing Forestry University,Nanjing 210037,China
关键词:
图割; 图像区域连续化; 能量函数最小化; 区域匹配; 窗口单位化
Keywords:
graph cut; continuity of image region; minimization of energy function; region matching; scale the window cost
分类号:
TP391.41
摘要:
传统图割算法解决双目立体匹配问题,在高精度的同时需要消耗大量时间.提出一种新的算法,将最小割求取问题转化为贪心问题,从而降低算法复杂度.由于转化后的图割在处理图像区域连续化问题时效率低下,给出了图割与区域匹配相结合的GR(Graphic Cut in Region)算法,算法不仅将图割理论运用到立体匹配问题中,且在求取初始视差时提出了用窗口单位化匹配代价算法来提高初始视差的精度.实验证明,该算法在图像区域连续化时有较好的效果,明显提高了匹配的精度,且复杂度也大大降低.
Abstract:
Traditional graph cut algorithm to solve the binocular matching problem is time-consuming while it requires great precision． This paper proposes a new algorithm by which the minimum cut graph cut problem is converted into the greedy algorithm to reduce the complexity of the problem． At the same time as the transformed graph cut has a low effect in dealing with the image continuous problem，we propose the graph cuts combined with the region matching algorithm called GR( Graphic Cut in Region) algorithm． They not only apply the Graph cut algorithm theory to the stereo matching problem，but also use a new method which computes the initial disparity cost through scaling the SAD window and twice compute the window cost to improve the accuracy of the initial disparity． The experiments show that this new algorithm in image continuous region has a better effect，and significantly improves the accuracy of matching at this time，and also that the algorithm complexity is greatly reduced．

参考文献/References:

［1］Hsieh Y C，D McKeown，F P Perlant． Performance evaluation of scene registration and stereomatching for car-tographic feature extraction［J］． IEEE Trans，1992，PAMI-14( 2) : 214-238．
［2］D Scharstein，R Szeliski． A taxonomy and evaluation of dense two-frame stereo correspondence algorithms［J］． Internation Journal of Computer Vision，2002， 47( 1) : 7-42．
［3］Sun Dongmei，Qiu Zhengding． A new non-rigid image matching algorithm using thin-plate spline［J］． Acta Electronica Sinica， 2002，30( 8) : 1 104-1 107．
［4］Boykov Y，Kolmogorov V． An experimental comparison of mincut /maxflow algorithms for energy minimization in vision［J］． IEEE Transon Pattern Analysis and Machine Intelligence，2004， 26( 9) : 1 124-1 137．
［5］王年，范益政，鲍文霞． 基于图割的图像匹配算法［J］． 电子学报， 2006， 34( 2) : 232-236． Wang Nian，Fan Yizheng，Bao Wenxia． Graph matching algorithm base on graph cut［J］． E-Journal， 2006， 34( 2) : 232-236． ( in Chinese)
［6］Alvarez L，Deriche R，S’anchez J，et al． Dense disparity map estimation respecting image derivatives: a PDE and scale-space based approach［J］． Journal of Visual Communication and Image Representation，2002，13( 1 /2) : 3-21．

备注/Memo

备注/Memo:
通讯联系人: 刘应安，博士后，教授，研究方向: 数据挖掘、概率统计． E-mail: lyastat@ yahoo． com． Cn
更新日期/Last Update: 2013-03-21