Short Time Cloud-Free Image Reconstruction Based on Time Series Images

Abstract

A cost-effective method for cloud removal and reconstruction of remote sensing images has been a long-standing research challenge in remote sensing data processing. In this article, we address this challenge by presenting a fast and simple method for cloud detection and cloud-free image reconstruction using Landsat-8 OLI and Sentinel-2 MSI series data. The proposed method utilizes the spectral difference between cloud pixels and transparent pixels to develop a fast and simple cloud detection algorithm for multispectral remote sensing sensors. Subsequently, the cloud-free data of the complementary image and the target image undergo histogram matching, and the cloud-free pixels of the complementary image are seamlessly integrated into the original image, leading to the generation of a cloud-free image reconstruction algorithm. Comparative analysis between the results obtained from our proposed method and the corresponding artificial results reveals an accuracy rate exceeding 90% and a high consistency in the reconstructed spatial spectrum. By addressing the need for cost-effective cloud removal and image reconstruction, our method contributes to the advancement of remote sensing data processing and applications.

Keywords

• Cloud detection
• cloud removal
• data fusion
• image reconstruction
• time series analysis