Multi-scale Correlation-based feature selection and Random forest classification for LULC mapping from the integration of SAR and optical Sentinel images

摘要

Reliable and accurate land use/land cover (LULC) map is a crucial data source for the understanding of coupledhuman–environment systems, monitoring changes, timely low-cost planning, and management of natural resources.Improvements in sensor technologies and machine learning capabilities have shifted the attention of remote sensingcommunity to data complementarity through fusion of multi-sensor data for accurate feature extraction and mapping.Amalgamation of optical and synthetic aperture radar (SAR) images has shown promising advantages in enhancingthe accuracy of extracting LULC as such method allows exploitation of information in sensors. This study investigatedthe potential of using freely available multisource Sentinel images to extract LULC maps in semi-arid environmentsthrough multi-scale geographic object-based image analysis (GEOBIA). A multi-scale classification framework thatintegrates GEOBIA, correlation-based feature selection (CFS), and random forest (RF)-supervised classification wasadopted to extract LULC from assimilation of Sentinel multi-sensor products. First, Sentinel-1 and -2 images werepre-processed. Second, optimum multi-scale segmentation levels were selected using F-score segmentation qualitymeasures. Third, 70 features of various spectral indices and derivatives and geometrical features from optical data andmultiple ratios and textural features from dual-polarization SAR images were computed, and a CFS based on wrapperapproach was used to select the most significant features at multi-scale levels. Finally, a single and multi-scale RFclassifier was used to extract LULC classes using the most relevant features extracted from Sentinel SAR and opticalimages. Results of multi-scale image segmentation optimization showed that scale parameter (SP) values of 40, 60,and 150 were optimal for extraction of LULC classes. Results of feature selection showed that 22, 24, and 27 featureswere selected at scale SP values of 40, 60, and 150, respectively. Half of the features were common among the threescales. Single RF classification yielded overall accuracy (OA) values of 92.10%, 93%, and 91% and kappa coefficientsof 0.901, 0.912, and 0.89 at scale values of 150, 60, and 40, respectively. Multiscale RF classification from scalevalues of 150 and 60 produced better LULC classification with OA 96.06% and kappa coefficient of 0.95 comparedwith other scale SP values. The integrated approach demonstrated an effective and promising method for high-qualityLULC extraction from coupling optical and SAR images. Overall, multi-sensor Sentinel images along with theadopted approach feature a remarkable potential for improving LULC extraction and can effectively be used to updategeographic information system layers for various applications.