基于MapReduce的海量图像处理模型的研究

针对目前图像的格式和数量都在不断增加,传统的串行处理方法无法满足海量的图像数据处理的问题,提出一种基于MapReduce并行框架的海量图像数据处理模型.模型中取消了Reduce处理函数,在Map函数处理完成后直接输出处理结果,不仅避免了Reduce函数和Re-duce任务处理所需的时间,同时减少了Map与Reduce阶段之间操作所消耗的时间.实验基于Hadoop伪分布式云平台,实现了文中提出的并行处理模型.

Research on the massive image parallel processing model based on the MapReduce

With the increasing of image formats and quantities,the traditional stand-alone serial processing method can＇t meet the problems of processing massive image data,this paper proposes a massive image data processing model based on MapReduce parallel framework.This model canceled Reduce function,and outputed directly processing result after the Map function completed,this not only avoided Reduce processing time,but also reduced the time consumed by the Map and Reduce intermediate operations.The experiment is based on the Hadoop pseudo-distributed cloud platform,and achieves the proposed parallel processing model in handling massive image files.