MapReduce based Big Data Framework for Content Searching of Surveillance System Videos

Abstract

CBIR systems go through sets of stages starting from acquiring the new images, representing these images by extracting the image features, describing the key features and eventually computing the similarity distances to get the most relevant results responding to the query image. In this paper, ICBIR an integrated CBIR Hadoop-MapReduce based framework which is split into both offline and online phases is introduced. Visual statements are built using the extracted interest points SIFTs. Later on, these visual statements are used to estimate the similarity distances which in turn are used to create the image dataset clusters. A huge vocabulary of SIFTs describing the interest points of the image is constructed. In this paper, the authors are interested in routing protocols based on clusters that aim to reduce congestion in order to have reliable data transmission and a reduced loss rate. This is achieved by balancing the traffic load, which results into a balanced energy consumption within the network.