Abstract
Synthetic aperture radar (SAR) is crucial for military reconnaissance and remote sensing, but image quality can be affected by various factors, impacting target detection performance. Thus, pre-evaluation of SAR image quality is essential to filter out poor-quality images, optimize resource allocation, and enhance detection accuracy and efficiency. This paper proposes a comprehensive SAR image quality evaluation method combining objective and subjective approaches. Specifically, the processes encompassing the generation of a series of disturbed SAR images on the SAR ship detection dataset (SSDD), the calculation of various objective quality indicators for those images, and the assignment of a subjective quality label to each image through subjective evaluation. Based on the dataset constructed by the above evaluation methods, the IHHO-XGBoost model was developed. This model uses an improved harris hawk optimization (IHHO) algorithm to optimize extreme gradient boosting (XGBoost) hyperparameters. The IHHO algorithm effectively alleviates the problem of getting trapped in local optima by improving the escape energy calculation strategy and integrating the average difference evolution mechanism while maintaining the diversity of the population, showing significant advantages over the traditional HHO algorithm. Comparative experiments demonstrate the model's superiority in SAR image quality evaluation. This study validates the scientificity and practicability of the proposed method, offering new tools for SAR image quality research.