Automated Synthetic Aperture Radar Image Quality Assessment

Synthetic aperture radar (SAR) systems are generally tasked to image scenes for accomplishing particular objectives. To this end, sensor parameters are set to achieve desired resolutions and signal-to-noise ratio (SNR) levels for ensuring proper image interpretability. Predictive models, such as the Radar National Image Interpretability Rating Scale (RNIIRS) and Radar Generalized Image Quality Equation (RGIQE), are extremely helpful in mapping desired imaging objectives to sensor parameter settings. However, many complications can arise while collecting phase history data and forming images that can degrade the actual image quality. Thus, it is important to have a mechanism to assess the quality of formed images to determine if the as-collected data supports the as-tasked sensor parameters. Analyzing impulse responses (IPRs) and thermal noise regions to assess SNR are two useful image quality assessments. Manual inspection of IPRs and SNR values is tedious and does not scale with large volumes of imagery, which creates a need for automated assessment. In this paper, we introduce an automated image quality assessment that computes an estimated RGIQE score. The estimated RGIQE score can be compared to the predicted score to determine if tasking parameters were achieved, as well as be used to select from follow-on image processing algorithms that are designed for different levels of exploitation based on image quality. We demonstrate the automated RGIQE assessment on a large body of commercial satellite SAR images.