Automatic Cloud Removal from Multitemporal Satellite Images
- PDF / 1,918,424 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 103 Downloads / 237 Views
RESEARCH ARTICLE
Automatic Cloud Removal from Multitemporal Satellite Images S R Surya & Philomina Simon
Received: 14 October 2013 / Accepted: 27 May 2014 # Indian Society of Remote Sensing 2014
Abstract Remote sensing images are more or less influenced by clouds and cloud shadows during the data acquisition, which pose a major challenge in data processing. As a result, many researchers have come up with different methods to detect and remove the clouds and their shadows from remote sensing images. In this paper, an automatic cloud removal algorithm is proposed to generate cloud-free and cloud shadow-free images from multi temporal registered remotesensing images. An automatic cloud detection and a shadow detection algorithm is combined in this method. The quality assessment of multitemporal images based on SSIM index is used to sort the images. Information cloning is used to fill the cloud-covered areas in the satellite image. For each cloud contaminated area, the corresponding cloud free areas are selected from sorted multitemporal images to reconstruct the clouds without any visible seams. Experimental analysis is performed on Landsat-7 Enhanced Thematic Mapper Plus (ETM+) sensor images and results are obtained. Experimental results proved the better performance of proposed method. Both the thin and thick clouds can be removed efficiently using the proposed method.
Keywords Cloud Shadow detection . Cloud Detection . Discrete Poisson Solver . Information Cloning . Multitemporal images . Cloud Removal
S. R. Surya (*) : P. Simon Department of Computer Science, University of Kerala, Kariavattom, Thiruvananthapuram, India e-mail: [email protected] P. Simon e-mail: [email protected]
Introduction The data obtained by the optical satellite sensors with high spatial resolution are used for different applications such as managing, developing, protecting our population, environment, and resources. Satellite images are often obscured by clouds due to atmospheric conditions at the time the images are captured. Places under constant cloud cover, especially in humid tropical regions; it is not possible to obtain cloud free images. So those images cannot be used for further applications. In such situations, it is necessary to detect and remove clouds from satellite images. Since most of the optical satellite sensors obtain images through reflection of sunlight, satellite cannot obtain ground information if clouds exist in the sky. The recent advancement in this research area is to use the multitemporal information. Multitemporal images are images obtained by satellite for the same geographical area at varying time. Since clouds presence in the sky varies with time, clouds in the part of an image can be replaced with cloud free information from other multitemporal image. Multitemporal imaging (Liang and Chen 2001) is the acquisition of remotely sensed data from more than one instance of time. Acquiring multiple views of features through time provides an enhanced ability to accurately identify significant informa
Data Loading...
