Smart Information Reconstruction via Time-Space-Spectrum Continuum for Cloud Removal in Satellite Images
19 Mar 2015-IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (IEEE)-Vol. 8, Iss: 5, pp 1898-1912
TL;DR: 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.
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 $7.4{\hbox{E}} - 04{\hbox{sr}}^{-1}$ 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.
Citations
More filters
TL;DR: Three decades of high-resolution Landsat 5 satellite imagery are used to investigate long-term trends in intense summertime near-surface phytoplankton blooms for 71 large lakes globally, revealing a worldwide exacerbation of bloom conditions.
Abstract: Freshwater blooms of phytoplankton affect public health and ecosystem services globally1,2. Harmful effects of such blooms occur when the intensity of a bloom is too high, or when toxin-producing phytoplankton species are present. Freshwater blooms result in economic losses of more than US$4 billion annually in the United States alone, primarily from harm to aquatic food production, recreation and tourism, and drinking-water supplies3. Studies that document bloom conditions in lakes have either focused only on individual or regional subsets of lakes4–6, or have been limited by a lack of long-term observations7–9. Here we use three decades of high-resolution Landsat 5 satellite imagery to investigate long-term trends in intense summertime near-surface phytoplankton blooms for 71 large lakes globally. We find that peak summertime bloom intensity has increased in most (68 per cent) of the lakes studied, revealing a global exacerbation of bloom conditions. Lakes that have experienced a significant (P < 0.1) decrease in bloom intensity are rare (8 per cent). The reason behind the increase in phytoplankton bloom intensity remains unclear, however, as temporal trends do not track consistently with temperature, precipitation, fertilizer-use trends or other previously hypothesized drivers. We do find, however, that lakes with a decrease in bloom intensity warmed less compared to other lakes, suggesting that lake warming may already be counteracting management efforts to ameliorate eutrophication10,11. Our findings support calls for water quality management efforts to better account for the interactions between climate change and local hydrological conditions12,13. Analyses show that the peak intensity of summertime phytoplankton blooms has increased in 71 large lakes globally over the past three decades, revealing a worldwide exacerbation of bloom conditions.
595 citations
TL;DR: Compared with other cloud removal methods, the results demonstrate that S-NMF-EC is visually and quantitatively effective for the removal of thick clouds, thin clouds, and shadows.
Abstract: In the imaging process of optical remote sensing platforms, clouds are an inevitable barrier to the effective observation of sensors. To recover the original information covered by the clouds and the accompanying shadows, a nonnegative matrix factorization (NMF) and error correction method (S-NMF-EC) is proposed in this paper. Firstly, a cloud-free fused reference image is obtained by a reference image and two or more low-resolution images using the spatial and temporal nonlocal filter-based data fusion model (STNLFFM). Secondly, the cloud-free fused reference image is used to remove the cloud cover of the cloud-contaminated image based on NMF. Finally, the cloud removal result is further improved by error correction. It is worth noting that cloud detection is not required by S-NMF-EC, and the cloud-free information of the cloud-contaminated image is maximally retained. Both simulated and real-data experiments were conducted to validate the proposed S-NMF-EC method. Compared with other cloud removal methods, the results demonstrate that S-NMF-EC is visually and quantitatively effective (correlation coefficients ≥ 0.99) for the removal of thick clouds, thin clouds, and shadows.
144 citations
TL;DR: Aerosol optical depth (AOD) is widely recognized as a critical indicator in understanding atmospheric physics and regional air quality because of its capability for quantifying aerosol loading in t...
Abstract: Aerosol optical depth (AOD) is widely recognized as a critical indicator in understanding atmospheric physics and regional air quality because of its capability for quantifying aerosol loading in t...
72 citations
Cites methods from "Smart Information Reconstruction vi..."
...As mentioned before, the SMIR method via the time-space-spectrum continuum is worthwhile to apply to fill in data gaps and to recover missing pixels via memory effects (Chang et al., 2015)....
[...]
TL;DR: It is illustrated how ML can accelerate the pace of improvement in environmental data exploitation and weather prediction—first, by complementing existing systems, and second, where appropriate, as an alternative to some components of the NWP processing chain from observations to forecasts.
Abstract: Artificial intelligence (AI) techniques have had significant recent successes in multiple fields. These fields and the fields of satellite remote sensing and NWP share the same fundamental ...
58 citations
TL;DR: This study develops a remote sensing-based multiscale modeling system by integrating multi-sensor satellite data merging and image reconstruction algorithms in support of feature extraction with machine learning leading to automate continuous water quality monitoring in environmentally sensitive regions.
46 citations
References
More filters
TL;DR: High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated and the performance of the support- vector network is compared to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
Abstract: The support-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data.
High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
37,861 citations
"Smart Information Reconstruction vi..." refers background in this paper
...In specific, evolutionary algorithms, including genetic algorithm [23], genetic programming [24], evolutionary programming [25], and neural networks including backpropagation (BP) [26], support vector machine [27], and deep learning [28], [29], may fill in some gaps in pattern recognition....
[...]
Book•
01 Jan 1992TL;DR: This book discusses the evolution of architecture, primitive functions, terminals, sufficiency, and closure, and the role of representation and the lens effect in genetic programming.
Abstract: Background on genetic algorithms, LISP, and genetic programming hierarchical problem-solving introduction to automatically-defined functions - the two-boxes problem problems that straddle the breakeven point for computational effort Boolean parity functions determining the architecture of the program the lawnmower problem the bumblebee problem the increasing benefits of ADFs as problems are scaled up finding an impulse response function artificial ant on the San Mateo trail obstacle-avoiding robot the minesweeper problem automatic discovery of detectors for letter recognition flushes and four-of-a-kinds in a pinochle deck introduction to biochemistry and molecular biology prediction of transmembrane domains in proteins prediction of omega loops in proteins lookahead version of the transmembrane problem evolutionary selection of the architecture of the program evolution of primitives and sufficiency evolutionary selection of terminals evolution of closure simultaneous evolution of architecture, primitive functions, terminals, sufficiency, and closure the role of representation and the lens effect Appendices: list of special symbols list of special functions list of type fonts default parameters computer implementation annotated bibliography of genetic programming electronic mailing list and public repository
13,487 citations
TL;DR: Recent work in the area of unsupervised feature learning and deep learning is reviewed, covering advances in probabilistic models, autoencoders, manifold learning, and deep networks.
Abstract: The success of machine learning algorithms generally depends on data representation, and we hypothesize that this is because different representations can entangle and hide more or less the different explanatory factors of variation behind the data. Although specific domain knowledge can be used to help design representations, learning with generic priors can also be used, and the quest for AI is motivating the design of more powerful representation-learning algorithms implementing such priors. This paper reviews recent work in the area of unsupervised feature learning and deep learning, covering advances in probabilistic models, autoencoders, manifold learning, and deep networks. This motivates longer term unanswered questions about the appropriate objectives for learning good representations, for computing representations (i.e., inference), and the geometrical connections between representation learning, density estimation, and manifold learning.
11,201 citations
"Smart Information Reconstruction vi..." refers background in this paper
...In specific, evolutionary algorithms, including genetic algorithm [23], genetic programming [24], evolutionary programming [25], and neural networks including backpropagation (BP) [26], support vector machine [27], and deep learning [28], [29], may fill in some gaps in pattern recognition....
[...]
TL;DR: A new learning algorithm called ELM is proposed for feedforward neural networks (SLFNs) which randomly chooses hidden nodes and analytically determines the output weights of SLFNs which tends to provide good generalization performance at extremely fast learning speed.
10,217 citations
"Smart Information Reconstruction vi..." refers background in this paper
...where H† is the Moore-Penrose generalized inverse of H [30], [38]....
[...]
...activation functions are infinitely differentiable with the hidden layer output matrix H remaining unchanged [30]....
[...]
...Extreme learning machine (ELM), a simple and efficient learning algorithm that tends to provide good generalization performance at extremely fast learning speed and learning accuracy as compared to conventional neural computing algorithms such as BP [30], enables us to quickly and accurately identify the complex relationships over the timespace-spectrum continuum....
[...]
...[30] to improve the learning speed and accuracy for SLFNs....
[...]
..., L)} to minimize the difference between approximations and targets [30]...
[...]
TL;DR: The Marquardt algorithm for nonlinear least squares is presented and is incorporated into the backpropagation algorithm for training feedforward neural networks and is found to be much more efficient than either of the other techniques when the network contains no more than a few hundred weights.
Abstract: The Marquardt algorithm for nonlinear least squares is presented and is incorporated into the backpropagation algorithm for training feedforward neural networks. The algorithm is tested on several function approximation problems, and is compared with a conjugate gradient algorithm and a variable learning rate algorithm. It is found that the Marquardt algorithm is much more efficient than either of the other techniques when the network contains no more than a few hundred weights. >
6,899 citations