Interpreter-Based Evaluation of Compressed SAR Images Using JPEG and HEVC Intra Coding: Compression Can Improve Usability

Abstract:
Usability-based assessment of SAR systems is time-consuming and costly, if based on human evaluation. We investigate an assessment system which aims at finding algorithms to simulate, complement and partly replace this evaluation. The system component investigated is the coding with spatial and amplitude resolution as parameters. Images of different qualities were generated by JPEG and intra-HEVC for evaluation. SAR experts solved tasks on SAR images with different qualities. A quality metric called weighted F1-score evaluates the solved tasks. Best score is at a data rate of 0.30 bit/pel with spatial resolution of 11 cm using JPEG.