Effects of SAR Image Compression on Image Interpretability and Detection Performance

The present study was designed to compare the relative impact of three image compression techniques on the interpretability of synthetic aperture radar (SAR) imagery: multiresolution encoding (MRES), Joint Photographic Experts Group (JPEG), and vector quantization (VQ). These three techniques were examined at three compression ratios (8:1, 16:1, and 32:1) and two image resolutions (fine and coarse) to determine which is most suitable for SAR imagery. Six experienced image analysts provided subjective ratings of image interpretability for original and compressed SAR images via the Radar National Imagery Interpretability Rating Scale (RNIIRS). The subjective ratings were further supplemented with objective indicators of target detection performance (i.e., hits and false alarms). For each dependent variable, difference scores representing the difference in value for comparable uncompressed and compressed images were computed and analyzed. In terms of both subjective and objective measures of effectiveness, MRES was most similar to the uncompressed imagery.