Environmental sustainability research is dependent on accurate land cover information. Even with the increased number of satellite systems and sensors acquiring data with improved spectral, spatial, radiometric and temporal characteristics and the new data distribution policy, most existing land cover datasets are derived from a pixel-based, single-date multi-spectral remotely sensed image with an unacceptable accuracy. One major bottleneck for accuracy improvement is how to develop an accurate and effective image classification protocol. By incorporating and utilizing multi-spectral, multi-temporal and spatial information in remote sensing images and considering the inherit spatial and sequential interdependence among neighboring pixels, we propose a new patch-based recurrent neural network (PB-RNN) system tailored for classifying multi-temporal remote sensing data. The system is designed by incorporating distinctive characteristics of multi-temporal remote sensing data. In particular, it uses multi-temporal-spectral-spatial samples and deals with pixels contaminated by clouds/shadow present in multi-temporal data series. Using a Florida Everglades ecosystem study site covering an area of 771 square kilometers, the proposed PB-RNN system has achieved a significant improvement in the classification accuracy over a pixel-based recurrent neural network (RNN) system, a pixel-based single-image neural network (NN) system, a pixel-based multi-image NN system, a patch-based single-image NN system, and a patch-based multi-image NN system. For example, the proposed system achieves 97.21% classification accuracy while the pixel-based single-image NN system achieves 64.74%. By utilizing methods like the proposed PB-RNN one, we believe that much more accurate land cover datasets can be produced over large areas.
Modeling the dynamic interactions within an ecosystem and among ecosystems is essential to the sustainability of the biosphere and remote sensing is a proven technology that can effectively map and characterize cultural and natural landscapes. However, the lack of scalable and reliable algorithms and associated implementations to extract implicit patterns in remotely sensed images has severely limited its success. In this paper, based on a key observation that multiple samples of multispectral sensing form a curve in the spectral space (named as mspectral curve), we propose a novel mathematical change detection and land cover classification framework based on statistical shape analysis for natural ecosystem monitoring using remote sensing. By bridging multitemporal analysis and statistical shape analysis, a rich set of robust estimation and optimization techniques can be utilized via tangent space representation to classify land cover and detect changes. The framework also introduces scalable clustering algorithms based on product quantization and residual vector quantization to deal with the labeling problem. In order to deal with clouds and cloud shadows, an adapted Fmask (Function of mask) method is used. Deep network classification is also incorporated for more efficient and accurate context-sensitive land cover classification. The framework provides a fundamental building block to improve semantic classification and change detection using temporal and spatial context to disambiguate spectral classes into information classes.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.