Low-Power JPEG2000 Implementation on DSP-based Camera Node in Wireless Multimedia Sensor Networks

Recently many of the potential WSN applications such as target tracking, process control and source localization have fostered the development of Wireless Multimedia Sensor Networks (WMSNs). The amount of image data gathered by camera nodes in WMSNs is very large. They need to be compressed before transmission. JPEG2000 is the state-of-the-art image compression standard which has many advantages, such as high compression efficiency and high image fidelity. However, its high computational complexity makes it a big problem when using it in WMSNs. To satisfy the low-power and high image quality requirements of WMSNs, the paper studies the low-power JPEG2000 realization on DSP-based camera node. Fixed-point LS9/7 wavelet is used instead of CDF9/7 wavelet to decrease the complexity of computation. Embedded Block Coding is improved for high efficient DSP parallel processing. Experiments results show that the improved JPEG2000 leads to a great reduction in implementation time, and hence saves energy consumption of camera nodes.