研究目的
To propose a method of cloud shadow removal for multispectral images to retrieve the ground reflectance of areas shadowed by clouds, taking the physical model of cloud shadow formation into account to accurately estimate the attenuation factors.
研究成果
The proposed method of cloud shadow removal, which takes the physical model of cloud shadow formation into account, outperforms the conventional method, improving the average correlation coefficient from 0.45 to 0.75 and achieving a 10% improvement in the average spectral angle.
研究不足
The method assumes that a cloud and the shadow cast by it coexist in an input image. It may not correct shadows for which corresponding clouds are not present in the image.
1：Experimental Design and Method Selection:
The proposed method involves cloud transmittance estimation, cloud-shadow pair matching, AFSI estimation, and cloud shadow correction. It employs a spectral unmixing based approach for cloud transmittance estimation and a template matching technique for cloud-shadow pair matching.
2：Sample Selection and Data Sources:
The evaluation was performed on a subset of a Landsat 8 OLI MS image captured on 9th Sept. 2013, with a shadow-free image of the same area captured on 11th Oct. 2013 used as a reference.
3：List of Experimental Equipment and Materials:
Landsat 8 OLI MS image data.
4：Experimental Procedures and Operational Workflow:
The method detects clouds, estimates their transmittance, matches cloud-shadow pairs, estimates AFSI values, and corrects shadow pixels.
5：Data Analysis Methods:
Quantitative evaluation using correlation coefficients and spectral angle with a reference shadow-free image.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
