David. R. Gerwe (Boeing Phantom Works – Advanced Networks & Space Systems), Carlos E. Luna (Boeing Phantom Works – Advanced Networks & Space Systems), Brandoch Calef (Boeing Laser Technical Services)
Keywords: Adaptive Optics, Imaging
Abstract:
This paper describes and assesses a metric for quantifying the level of image visual information content and quality in terms of the National Imagery Interpretability Rating Scale (NIIRS). The Information Theoretic Image Quality Equation (ITIQUE) metric is based on the Shannon mutual information (MI) at multiple spatial scales between a pristine object and the image output from a detailed image formation chain simulation. Integrating the MI at each spatial scale and applying a calibration offset produces a prediction of NIIRS image quality indicating the level of interpretation tasks that could be supported. The model enables prediction of NIIRS quality obtainable as dependent on image collection conditions and imaging system design including both hardware and processing algorithms. The focus of this paper is on ITIQUEs applicability to Space Situational Awareness (SSA) domain imagery, degradations, and non-linear processing techniques. ITIQUE results were compared with visual assessments from a panel of human observers for a set of 480 images that spanned a 16x resolution range, encompassed many degradation types, and included linear and non-linear image enhancement processing. ITIQUE model predictions are shown to human scores to nearly within the human-to-human variability.
Date of Conference: September 11-14, 2012
Track: Adaptive Optics and Imaging