In this proposal we address the problem of developing a GIQE for ATR in SAR. We do employ image features such as IPR mainlobe width, IPR sidelobe peak, Integrated sidelobe ratio (ISLR) in the context of MNR (multiplicative noise ratio), system noise, grazing angle, etc., to capture the system variability, collection geometry, scene selection, and mode of operation. SET will perform statistical regression analysis to identify the linear or non-linear functional dependencies of the ATR scores (mimicking NIIRS) to each feature where prudent order selection and mean-square weight training are exercised. Finally, residual analysis will be performed which examines the correlation between the regression (prediction) error and the value of the predictor variables.
Benefit: This research effort will Determine the influence of features on SAR image quality Provide a significant progress in realizing the Performance Driven Sensing Produce a Phase II plan to verify/validate the dependencies of image quality NIIRS to some features.
Keywords: Sar Image Formation, Digital Signal Processing, Statistical Regression Analysis, Image Quality Niirs, Image Features, Atr.
