Analysis and optimization method of influencing factors for three-dimensional target mixed pixels under zero meteorological view distance

Abstract

Mixed pixels widely exist in the obtained hyper-spectral data,which directly affects the accuracy of optical remote sensing imaging simulation. Aiming at the three-dimensional target under zero meteorological view distance, this paper comprehensively considers the effects of directional reflection characteristics, geometric shadows and background radiation on the generation of mixed pixels.It proposes a method of modeling mixed pixels by taking each surface of a three-dimensional target as an end element component and using the weight coefficient of the distance between the end element groups to the view center. Through designing experiments to simulate different influence conditions such as directional reflection, geometric shadow and background radiation, this paper verifies the influence of the radiation value of the mixed pixels of the three-dimensional target produced by different mixing conditions in the single-pixel field of view. The experimental results show that directional reflection characteristics, geometric occlusion shadows and background radiation have significant effects on the generation of mixed pixels for the three-dimensional targets. The improved mixed pixel modeling method proposed in this paper considering the influence of multiple factors has an average error of less than 10% in modeling accuracy of three-dimensional targets, which is significantly better than traditional linear modeling methods. It can be seen that the research results of this paper have good reference significance for the research of the three-dimensional target mixed pixel radiation model.