Compression techniques for improving underwater acoustic transmission of images and data

Efficient use of the bandwidth-limited acoustic communications link requires that considerable attention be given to both removing redundancy in data and minimizing superfluous resolution. In this paper we present techniques for compression of laser line scan and camera images, as well as format-specific data compression for quick-look sonar mapping data. For image compression JPEG and a wavelet-based technique (EPIC) are examined. JPEG is found to be less efficient than the wavelet transform but has the advantage that it is robust with respect to lost data packets. The wavelet-based transform is more efficient at high compression rates though below a certain rate both offer similar performance. The specific context for this work is the autonomous minehunting and mapping technology (AMMT) demonstration which utilizes the DARPA large diameter underwater vehicle. The ultimate goal of the project is identification and imaging of mine-like objects. The algorithms presented here were implemented in real-time and operated in the field for this program. Details of specific issues encountered during development of the system are also described.