A new CFA based image compression technique for energy-starved wireless multimedia sensor network

摘要

The advent of efficient short range radio communication coupled with advancement in miniaturization of computing devices has enabled the development of wireless multimedia sensor network (WMSN). Energy is one of the scarcest resources in such networks, especially it is scarce in transmitting multimedia data. This work presents an energy-saving image compression scheme. The nodes in WMSN are equipped with Colour Filter Array (CFA) using Bayer Pattern. Considering alternate column dropping of pixel array of the captured image is one of the viable techniques for low-overhead image comprssion, Bayer CFA lacks in features by means of which image pixel may be reconstructed using its neighbouring pixel information. So primarily we design a new CFA (NCFA) to overcome the said problems of Bayer CFA. Subsequently, the compression scheme is proposed where the alternate columns of the pixel array of each of the macroblocks of an image captured through NCFA, are dropped. At the high-end receiving node (e.g. sink), image is reconstructed from the sequentially transmitted macroblocks. Performance of the scheme is evaluated and compared through qualitative and quantitative analyses. The results show the present scheme's dominance over all the variants of the competing scheme in terms of both energy-efficiency and reconstruction quality.