Evaluation of GF-4 Satellite Multispectral Image Quality Enhancement Based on Super Resolution Enhancement Method

Abstract

GF-4 satellite is the world’s first high-resolution geostationary Earth observation satellite, which has broad application prospects in disaster prevention and relief, natural resource surveys, and atmospheric environmental monitoring. In order to improve the target recognition and classification capabilities by using GF-4 satellite image, the super-resolution reconstruction method is adopted. In this chapter, the GF-4 image quality enhancement software based on contour-consistent image super-resolution enhancement method is used to enhance the spatial resolution of GF-4 satellite PMS image. Under the condition of the same spatial resolution enhancement capability, the number of suitable image frames needed for multi-frame image super-resolution enhancement is analyzed quantitatively by selecting multiple evaluation indexes. Maximum likelihood classification method is used to classify land cover using GF-4 satellite resolution enhancement image and original image, respectively. Classification map of GF-1 satellite WFV image with 16 m resolution in the same classification method is used as reference data to evaluate the classification accuracy of GF-4 satellite images. Taking Kunshan City, Jiangsu Province as an experimental area, nine high-frequency imaging GF-4 satellite PMS images are selected for experiments. The experimental results show that with the increase of the number of GF-4 satellite PMS images participating in super-resolution enhancement, the radiation performance of the image after super-resolution enhancement presents an overall improvement trend. In addition, the GF-4 satellite PMS image after super-resolution enhancement not only improves the spatial resolution, but also improves the visual effect of the image. Combined with the characteristics of GF-4 satellite image and the geographical situation of Kunshan City, the land cover type is divided into residential land, water body, and vegetation. The classification results show that the overall accuracy of the classification based on the GF-4 satellite super-resolution enhanced image is higher than that based on the single-frame GF-4 satellite image.