%0 Journal Article
%T 基于MapReduce模型的立体图像合成方法<br>Stereoscopic Image Synthesis Method Based on MapReduce Model
%A 侯春萍
%A 王李坤
%A 王宝亮
%A 成
%A 石
%A 张文彬
%J 天津大学学报（自然科学与工程技术版）
%D 2016
%R 10.11784/tdxbz201509070
%X 随着多媒体信息的爆炸性增长，海量图像的快速处理逐渐成为当前的研究热点．针对单机模式下立体合成算法存在运行效率较低的问题，提出了一种基于MapReduce模型的立体合成改进方法．首先利用分布式计算和云存储技术的优势，对立体合成算法进行并行化设计；然后通过视差自适应调整和边缘锐化，对原有合成算法进行优化，从而提高立体显示质量．实验结果表明：提出的方法能够很好地提高立体合成算法的运算速度，同时合成出的立体图像具有较好的立体显示效果．<br>With the explosive growth of multimedia information，rapid processing of massive image has gradually become the hotspot of current research．In order to solve the problem of stereoscopic synthetic algorithm’s low efficiency in standalone mode，a stereoscopic synthetic improved algorithm based on the MapReduce programming model is pro-posed in this paper．Firstly，make use of the advantage of distributed computing and cloud storage technology to design paralleled stereoscopic synthetic algorithm．Secondly，through the parallax adaptive adjustment and edge sharpening，optimize the original synthesis algorithm to improve the quality of stereo display．Experiments show that this method can effectively improve the operation speed of the synthesis algorithm，and composite image has prefer-able perception of stero display quality
%K 分布式运算
%K 立体图像合成
%K 视差调整&lt
%K br&gt
%K distributed computation
%K stereoscopic image synthesis
%K parallax adjustment
%U http://journals.tju.edu.cn/zrb/oa/darticle.aspx?type=view&id=2006120002