Relative radiometric correction methods for remote sensing images and their applicability analysis

Relative radiometric correction of remote sensing images is a basic data preprocessing technique used to eliminate r adiometric problems in images such as non-uniformity,stripe noises,and defective lines. Over the past 30 years,numerous relative radiometric correction methods and algorithms have been developed. This situation brings up three questions.(1) What are the differences and relationships among these methods?(2)What are the characteristics of each method?(3) How is a suitable method selected? To answer these questions,the methods are first classified into three categories: calibration-based,scene-based,and c omprehensive methods. Second,the mathematical models of the three categories of methods are provided in the new classification system. Then,each method is introduced and compared with the others. Third,the applicability of the methods is comprehensively analyzed according to the following aspects: non-uniformity characteristics,geometric characteristics,sensor calibration,and the comprehensive characteristics of images. Suggestions on selecting a suitable method are provided,and an example of image correction is demonstrated. Finally,the research tendency and several existing problems of relative radiometric correction are analyzed.The computer criteria for signal and noises,the evaluation system,and the influence on subsequent absolute corrections require f urther studies.