ABSTRACT
Big Image Data Processing (BIDP) refers to the processing of images that are huge in terms of quantity, individual dimension, and individual size with respect to memory. This chapter elaborates on methods to deal with the three above-mentioned categories of images. In these scenarios, the data can be stored using a Distributed File System. To work with this amount of data, different programing paradigms can be used such as Hadoop’s MapReduce, Matlab’s MapReduce, and “Hadoop-Matlab” integrated environment with MapReduce Programing. The authors formed a Hadoop cluster with 116 systems and processed 1.2 TB of text data for word count task. The authors have also performed image retrieval on Corel 1000, Corel 10,000, Brodatz Textures, Mirflickr and ImageNet datasets effectively with this cluster configuration. The authors have created and processed a 32768 × 32768 dimension image and a 3.14 GB image using the MapReduce paradigm. Different applications using these technologies and methods are image retrieval and object detection, which can be used in a multiresolution environment as well.
