Comparative Analysis of GIQE 4 and GIQE 5 for Remote Sensing Image Quality Assessment

The interpretability of satellite remote sensing imagery serves as a critical metric for evaluating the imaging quality of remote sensing satellites. Currently, the National Imagery Interpretability Rating Scale (NIIRS) is a vital tool for assessing image interpretability. However, the General Image Quality Equation (GIQE), used to calculate NIIRS levels, has undergone iterative revisions across multiple versions, resulting in ambiguities in parameter definitions and persistent discrepancies in computational methodologies, which hinder standardization. To address these challenges, this paper conducts a comprehensive analysis and synthesis of the definitions and diverse solutions for the widely adopted GIQE 4 and GIQE 5 equations. By comparing the computational results of GIQE 4 and GIQE 5 for raw remote sensing image data, it is demonstrated that GIQE 5 yields values averaging 0.336 NIIRS higher than GIQE 4, as the former evaluates the holistic impact of Modulation Transfer Function Compensation (MTFC). In contrast, GIQE 4’s calculations are directly linked to MTFC, offering advantages in assessing the maximum gain MTFC can provide to NIIRS. However, due to its segmented formulation, which risks introducing "pseudo-results," it is essential to specify MTFC processing details when applying GIQE 4 to mitigate ambiguities. Additionally, this study emphasizes the influence of signal difference to noise ratio (SDNR) on NIIRS levels. When SDNR is low, noise becomes more prominent, amplifying the impact of sharpening gain and diminishing NIIRS enhancement efficacy. Conversely, higher SDNR values yield the opposite effect. Consequently, a positive correlation between SDNR and NIIRS is anticipated.