Smart Information Reconstruction via Time-Space-Spectrum Continuum for Cloud Removal in Satellite Images

Abstract

Cloud contamination is a big obstacle when processing satellite images retrieved from visible and infrared spectral ranges for application. Although computational techniques including interpolation and substitution have been applied to recover missing information caused by cloud contamination, these algorithms are subject to many limitations. In this paper, a novel smart information reconstruction (SMIR) method is proposed, in order to reconstruct cloud contaminated pixel values from the time-space-spectrum continuum with the aid of a machine learning tool, namely extreme learning machine (ELM). For the purpose of demonstration, the performance of SMIR is evaluated by reconstructing the missing remote sensing reflectance values derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) on board the Terra satellite over Lake Nicaragua, where is a very cloudy area year round. For comparison, the traditional backpropagation neural network algorithms will also be implemented to reconstruct the missing values. Experimental results show that the ELM outperforms the BP algorithms by an enhanced machine learning capacity with simulated memory effect embedded in MODIS due to linking the complex time-space-spectrum continuum between cloud-free and cloudy pixels. The ELM-based SMIR practice presents a correlation coefficient of 0.88 with root mean squared error of 7.4E − 04sr −1 between simulated and observed reflectance values. Finding suggests that the SMIR method is effective to reconstruct all the missing information providing visually logical and quantitatively assured images for further image processing and interpretation in environmental applications.Index Terms-Artificial neural network, cloud removal, computational intelligence, extreme learning machine, machine learning, satellite images.

Spectral Information Adaptation and Synthesis Scheme for Merging Cross-Mission Ocean Color Reflectance Observations From MODIS and VIIRS

“…To restore that missed spectral information, information reconstruction methods such as SMart Information Reconstruction (SMIR) [17], Neighborhood Similar Pixel Interpolator (NSPI) [33], Geostatistical Neighborhood Similar Pixel Interpolator (GNSPI) [34], and Weighted Linear Regression (WLR) integrated with a regularization method [35] can be applied to further recover the missing information toward a full clear coverage of study regions.…”

“…The two general seasons are the dry season from November to April and the wet season from May to October. Statistics show that dense cloud cover is mainly observed during the wet season, with an annual average cloud cover of nearly 70% [17]. As the lake has been considered as a future drinking water source by the Nicaragua government and several other Central American countries, monitoring biophysical parameters to characterize water quality conditions and pollution levels in this lake on a near-real-time basis is thus critical.…”

Quantification of relative contribution of Antarctic ozone depletion to increased austral extratropical precipitation during 1979–2013

“…This advantage allows the ELM to respond to the given inputs directly and quickly find a solution for the designated problem without huge intervention, enabling better universal generalization performance with less training error. Due to its fast learning speed and fair accuracy, the ELM has been widely used for solving complex generalization problems and classification purposes [ Huang et al ., ; Bai et al ., ; Chang et al ., ].…”

Foreword to the Special Issue on Information Extraction From High-Spatial-Resolution Optical Remotely Sensed Imagery

“…In [12], authors investigated how different atmospheric correction methods affect the discriminative ability of spectral features for urban tree species. Based on the time-space-spectrum continuum constructed from historical time series, [13] reconstructed missing pixels contaminated by clouds with machine learning tools. Besides optical imagery, other data sources are often helpful in the applications because of their complementary properties, but sometimes different sources do not cooperate well.…”