Combining random forest and support vector machines for object-based rural-land-cover classification using high spatial resolution imagery

Abstract. Land-cover classification using remote sensing imagery is an important part of environmental research because it provides baseline information for ecological vulnerability and risk assessment, disaster management, landscape conservation, local and regional planning, and so on. Rural-land-cover classification is challenging for both object-based image analysis methods and classifiers. The objective of this study is to improve the object-oriented classification accuracy of rural land cover by combining two models based on high spatial resolution imagery. We apply the C5.0 algorithm in R to combine support vector machines (SVMs) and random forest (RF) to create the model RS_C5.0. The effectiveness of the model combination is assessed by comparing the classification results with the state-of-the-art machine learning algorithm, namely extreme gradient boosting (XGBoost). The comparisons are done based on the classification results of both the study area and the case area. Results show that in the classification of the study area, RF performs slightly better than SVM, and XGBoost performs worse than RF but better than SVM. However, in the classification of the case area, SVM performs slightly better than RF and both SVM and RF perform better than XGBoost. Furthermore, RS_C5.0 obtains the highest overall accuracies and kappa coefficients in the classifications of both the study area and the case area. In terms of training time, XGBoost runs the slowest in the classifications of both the study area and the case area. SVM and RF as well as the combined model (RS_C5.0) run much faster than XGBoost classifier. To summarize, the combination of SVM and RF classifiers using C5.0 algorithm is found to be a fast and effective way to improve rural-land-cover classification.

[1]  Sabine Vanhuysse,et al.  Very High Resolution Object-Based Land Use–Land Cover Urban Classification Using Extreme Gradient Boosting , 2018, IEEE Geoscience and Remote Sensing Letters.

[2]  Sarah N. Banks,et al.  A Comparative Analysis of Object-Based and Pixel-Based Classification of RADARSAT-2 C–Band and Optical Satellite Data for Mapping Shoreline Types in the Canadian Arctic , 2015 .

[3]  René Roland Colditz,et al.  An Evaluation of Different Training Sample Allocation Schemes for Discrete and Continuous Land Cover Classification Using Decision Tree-Based Algorithms , 2015, Remote. Sens..

[4]  Gorthi R. K. Sai Subrahmanyam,et al.  Learning-Based Fuzzy Fusion of Multiple Classifiers for Object-Oriented Classification of High Resolution Images , 2017, CVIP.

[5]  Mario Chica-Olmo,et al.  An assessment of the effectiveness of a random forest classifier for land-cover classification , 2012 .

[6]  Artur Nowakowski,et al.  Remote Sensing Data Binary Classification Using Boosting with Simple Classifiers , 2015, Acta Geophysica.

[7]  Yan Li,et al.  Land cover change detection in Chinese Zhejiang Province based on object-oriented approach , 2016, Remote Sensing.

[8]  Christopher Conrad,et al.  Impact of feature selection on the accuracy and spatial uncertainty of per-field crop classification using Support Vector Machines , 2013 .

[9]  Steven E. Franklin,et al.  Multi-scale object-based image analysis and feature selection of multi-sensor earth observation imagery using random forests , 2012 .

[10]  Björn Waske,et al.  Optimization of Object-Based Image Analysis With Random Forests for Land Cover Mapping , 2013, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[11]  P. Atkinson,et al.  Random Forest classification of Mediterranean land cover using multi-seasonal imagery and multi-seasonal texture , 2012 .

[12]  Giles M. Foody,et al.  Good practices for estimating area and assessing accuracy of land change , 2014 .

[13]  Ying Xing,et al.  Automatic Classification of Remote Sensing Images Using Multiple Classifier Systems , 2015 .

[14]  Jennifer A. Miller,et al.  Contextual land-cover classification: incorporating spatial dependence in land-cover classification models using random forests and the Getis statistic , 2010 .

[15]  Narumasa Tsutsumida,et al.  Measures of spatio-temporal accuracy for time series land cover data , 2015, Int. J. Appl. Earth Obs. Geoinformation.

[16]  Yueyan Liu,et al.  Land use and land cover classification for rural residential areas in China using soft-probability cascading of multifeatures , 2017 .

[17]  Peijun Du,et al.  Random Forest and Rotation Forest for fully polarized SAR image classification using polarimetric and spatial features , 2015 .

[18]  Bardan Ghimire,et al.  An Evaluation of Bagging, Boosting, and Random Forests for Land-Cover Classification in Cape Cod, Massachusetts, USA , 2012 .

[19]  Fei Deng,et al.  Integration of orthoimagery and lidar data for object-based urban thematic mapping using random forests , 2013 .

[20]  Tianqi Chen,et al.  XGBoost: A Scalable Tree Boosting System , 2016, KDD.

[21]  H. Lange,et al.  Randomized comparison of operator radiation exposure during coronary angiography and intervention by radial or femoral approach , 2006, Catheterization and cardiovascular interventions : official journal of the Society for Cardiac Angiography & Interventions.

[22]  Mariana Belgiu,et al.  Quantitative evaluation of variations in rule-based classifications of land cover in urban neighbourhoods using WorldView-2 imagery , 2014, ISPRS journal of photogrammetry and remote sensing : official publication of the International Society for Photogrammetry and Remote Sensing.

[23]  Samia Boukir,et al.  Relevance of airborne lidar and multispectral image data for urban scene classification using Random Forests , 2011 .

[24]  Dirk Tiede,et al.  ESP: a tool to estimate scale parameter for multiresolution image segmentation of remotely sensed data , 2010, Int. J. Geogr. Inf. Sci..

[25]  Sulin Pang,et al.  C5.0 Classification Algorithm and Application on Individual Credit Evaluation of Banks , 2009 .

[26]  J. Harris,et al.  Integration of spectral, thermal, and textural features of ASTER data using Random Forests classification for lithological mapping , 2017 .

[27]  Onisimo Mutanga,et al.  Detecting Sirex noctilio grey-attacked and lightning-struck pine trees using airborne hyperspectral data, random forest and support vector machines classifiers , 2014 .

[28]  Guillermo Castilla,et al.  We Must all Pay More Attention to Rigor in Accuracy Assessment: Additional Comment to "The Improvement of Land Cover Classification by Thermal Remote Sensing". Remote Sens, 2015, 7, 8368-8390 , 2016, Remote. Sens..

[29]  Brian W. Barrett,et al.  Temporal optimisation of image acquisition for land cover classification with Random Forest and MODIS time-series , 2015, Int. J. Appl. Earth Obs. Geoinformation.

[30]  Patricia Gober,et al.  Per-pixel vs. object-based classification of urban land cover extraction using high spatial resolution imagery , 2011, Remote Sensing of Environment.

[31]  Onisimo Mutanga,et al.  Land-use/cover classification in a heterogeneous coastal landscape using RapidEye imagery: evaluating the performance of random forest and support vector machines classifiers , 2014 .

[32]  Chris Davies,et al.  Object-based image analysis of optical and radar variables for wetland evaluation , 2015 .

[33]  Le Yu,et al.  Towards automatic lithological classification from remote sensing data using support vector machines , 2010, Comput. Geosci..

[34]  André Stumpf,et al.  Object-oriented mapping of landslides using Random Forests , 2011 .

[35]  Thomas Blaschke,et al.  Object based image analysis for remote sensing , 2010 .

[36]  Wei Zhang,et al.  Target Identification from High Resolution Remote Sensing Image by Combining Multiple Classifiers , 2009, MCS.

[37]  Jan Haas,et al.  Urban growth and environmental impacts in Jing-Jin-Ji, the Yangtze, River Delta and the Pearl River Delta , 2014, Int. J. Appl. Earth Obs. Geoinformation.

[38]  J. A. Schell,et al.  Monitoring vegetation systems in the great plains with ERTS , 1973 .

[39]  Liangpei Zhang,et al.  Polarimetric-Spatial Classification of SAR Images Based on the Fusion of Multiple Classifiers , 2014, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[40]  Dirk Tiede,et al.  Village Forms Classification by Object-based Image Analysis , 2013 .

[41]  Francisco Alonso-Sarría,et al.  Modification of the random forest algorithm to avoid statistical dependence problems when classifying remote sensing imagery , 2017, Comput. Geosci..

[42]  Shihong Du,et al.  Semantic classification of urban buildings combining VHR image and GIS data: An improved random forest approach , 2015 .

[43]  Jonathan Cheung-Wai Chan,et al.  An evaluation of ensemble classifiers for mapping Natura 2000 heathland in Belgium using spaceborne angular hyperspectral (CHRIS/Proba) imagery , 2012, Int. J. Appl. Earth Obs. Geoinformation.

[44]  Peijun Du,et al.  Improving Random Forest With Ensemble of Features and Semisupervised Feature Extraction , 2015, IEEE Geoscience and Remote Sensing Letters.

[45]  A. Vetrivel,et al.  Identification of damage in buildings based on gaps in 3D point clouds from very high resolution oblique airborne images , 2015 .

[46]  J. Southworth,et al.  Analyzing Land Cover Change and Urban Growth Trajectories of the Mega-Urban Region of Dhaka Using Remotely Sensed Data and an Ensemble Classifier , 2017 .

[47]  Ujjwal Maulik,et al.  A Robust Multiple Classifier System for Pixel Classification of Remote Sensing Images , 2010, Fundam. Informaticae.

[48]  Fulgencio Cánovas-García,et al.  Optimal Combination of Classification Algorithms and Feature Ranking Methods for Object-Based Classification of Submeter Resolution Z/I-Imaging DMC Imagery , 2015, Remote. Sens..

[49]  Onisimo Mutanga,et al.  Mapping Solanum mauritianum plant invasions using WorldView-2 imagery and unsupervised random forests , 2016 .

[50]  Aniruddha Ghosh,et al.  A comparison of selected classification algorithms for mapping bamboo patches in lower Gangetic plains using very high resolution WorldView 2 imagery , 2014, Int. J. Appl. Earth Obs. Geoinformation.

[51]  Cheryl A. Palm,et al.  An operational framework for object-based land use classification of heterogeneous rural landscapes , 2017, Int. J. Appl. Earth Obs. Geoinformation.

[52]  Pawan Kumar Joshi,et al.  Random forest classification of urban landscape using Landsat archive and ancillary data: Combining seasonal maps with decision level fusion , 2014 .

[53]  Chih-Jen Lin,et al.  LIBSVM: A library for support vector machines , 2011, TIST.

[54]  Zhongwen Hu,et al.  Object-oriented land cover classification of HJ-1B CCD image through multiple classifier fusion , 2012, 2012 20th International Conference on Geoinformatics.

[55]  Mariana Belgiu,et al.  Comparing supervised and unsupervised multiresolution segmentation approaches for extracting buildings from very high resolution imagery , 2014, ISPRS journal of photogrammetry and remote sensing : official publication of the International Society for Photogrammetry and Remote Sensing.

[56]  S. K. McFeeters.  The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features , 1996 .

[57]  T. Martha,et al.  A Tool Assessing Optimal Multi-Scale Image Segmentation , 2017, Journal of the Indian Society of Remote Sensing.

[58]  Michele Dalponte,et al.  Tree Species Classification in Boreal Forests With Hyperspectral Data , 2013, IEEE Transactions on Geoscience and Remote Sensing.

[59]  P. Gong,et al.  Efficient corn and soybean mapping with temporal extendability: A multi-year experiment using Landsat imagery , 2014 .

[60]  O. Mutanga,et al.  Evaluating the impact of red-edge band from Rapideye image for classifying insect defoliation levels , 2014 .

[61]  Laurie A. Chisholm,et al.  Classification of Australian Native Forest Species Using Hyperspectral Remote Sensing and Machine-Learning Classification Algorithms , 2014, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[62]  Koreen Millard,et al.  On the Importance of Training Data Sample Selection in Random Forest Image Classification: A Case Study in Peatland Ecosystem Mapping , 2015, Remote. Sens..

[63]  Gang Chen,et al.  Identification of Forested Landslides Using LiDar Data, Object-based Image Analysis, and Machine Learning Algorithms , 2015, Remote. Sens..

[64]  K. Millard,et al.  Wetland mapping with LiDAR derivatives, SAR polarimetric decompositions, and LiDAR–SAR fusion using a random forest classifier , 2013 .

[65]  Changshan Wu,et al.  The use of single-date MODIS imagery for estimating large-scale urban impervious surface fraction with spectral mixture analysis and machine learning techniques , 2013 .

[66]  Christian Heipke,et al.  A higher order conditional random field model for simultaneous classification of land cover and land use , 2017 .

[67]  Brian A. Johnson,et al.  Scale Issues Related to the Accuracy Assessment of Land Use/Land Cover Maps Produced Using Multi-Resolution Data: Comments on "The Improvement of Land Cover Classification by Thermal Remote Sensing". Remote Sensing 2015, 7(7), 8368-8390 , 2015, Remote. Sens..

[68]  Imas Sukaesih Sitanggang,et al.  Web-based Classification Application for Forest Fire Data Using the Shiny Framework and the C5.0 Algorithm , 2016 .

[69]  Konstantinos Topouzelis,et al.  Oil spill feature selection and classification using decision tree forest on SAR image data , 2012 .

[70]  P. Gessler,et al.  The multispectral separability of Costa Rican rainforest types with support vector machines and Random Forest decision trees , 2010 .

[71]  Heather Reese,et al.  Combining airborne laser scanning data and optical satellite data for classification of alpine vegetation , 2014, Int. J. Appl. Earth Obs. Geoinformation.

[72]  M. Mahdianpari,et al.  Random forest wetland classification using ALOS-2 L-band, RADARSAT-2 C-band, and TerraSAR-X imagery , 2017 .

[73]  Jungho Im,et al.  Support vector machines in remote sensing: A review , 2011 .

[74]  Gérard Dedieu,et al.  Assessing the robustness of Random Forests to map land cover with high resolution satellite image time series over large areas , 2016 .