Data compression effects on SAR image interpretation

Application development projects focused on the use of the JPEG/DCT data compression routine to facilitate near-real-time SAR data delivery. The key requirements were to maintain image interpretability after image compression for three distinct cases: (1) points-ship detection; (2) linear features - oil slick detection; and (3) areas-ice mapping. Maximum compression ratios and corresponding mean square-errors obtained were: ships-(53.7:1, 22.9); oil slicks (52.5:1, 25.9); ice (27.9:1, 31.8). Image interpretability was high even at the largest compression ratios for ship detection, but image interpretability degraded with increasing compression ratio for oil slick detection and ice mapping. The decrease in interpretability was due to the high frequency spatial information of the oil slick and ice mapping imagery