摘要
提出一种基于行列双动态规划的立体匹配算法,采用能量最小化立体匹配模型,其中包含数据项和平滑项,在求解能量最小化过程中,在行方向上利用动态规划的方法给出视差图的能量最小化解,利用行动态规划的求解结果给予对应数据项一个奖励,同时在列方向上对视差图进行动态规划求解,并将其作为最终求解结果。实验结果表明,该算法能够取得较理想的效果。
A stereo matching algorithm based on dual dynamic programming with row and array is presented, which uses stereo matching model based on an energy-minimization framework, and contains two terms such as the data term and the smoothness term. In process of solving energy-minimization problem, the primitive disparity map using dynamic programming in the row direction is got, based on the results of which a reward is given to the corresponding data term in order to be used by next dynamic programming process, and the final results in the column direction is provided. Experimental results show this algorithm can get a better performance.
出处
《计算机工程》
CAS
CSCD
北大核心
2009年第8期227-229,共3页
Computer Engineering
基金
国家自然科学基金资助项目(30570475)
教育部博士基金资助项目(20050141025)
关键词
立体匹配
动态规划
自适应加权代价函数
stereo matching
dynamic programming
adaptive support-weight cost function