Abstract:

Targetbased image coding is important to address image transmission in real time over bandlimited 〖JP〗channels for synthetic aperture radar (SAR) payloads deployed on unmanned aerial vehicles (UAV). The algorithm proposed uses a two parameters constant false alarm rate (CFAR) detector to extract regions of interest (ROIs) and classifies the subband coefficients into target sequences and background data according to ROI mask. The target sequences are allpass filtered being nearly Gaussian distribution and encoded with fixed rate trellis coded quantization (FRTCQ) with high fidelity. The background data are encoded at a low bit rate with FRTCQ directly. Encoding experiment with SAR images shows that high target area performance can be achieved while large compression ratio is obtained, limited computation burden is added.