Comparative Assessment of Object-based and Pixel-based Approaches for Crop Cover Classification

Amara Sattar; Syed Muhammad Irteza; Sawaid Abbas; Sami Ullah Khan

Authors

Amara Sattar Smart Sensing for Climate and Development, Center for Geographical Information System, University of the Punjab, Lahore, Pakistan
Syed Muhammad Irteza Urban Unit, Lahore
Sawaid Abbas Department of Land Surveying and Geo-Informatics, The Hong Kong Polytechnic University, Hong Kong SAR
Sami Ullah Khan Urban Unit, Lahore

Keywords:

Sentinel-1, Sentinel-2, Google Earth Engine (GEE), Random Forest (RF), Simple Non-Iterative Clustering (SNIC), Vegetation Indices, Crop classification, Object Based Image Analysis (OBIA)

Abstract

Accurate crop classification is significant for productive agricultural planning and food security. This study compares pixel-based and object-based approaches for machine learning-based classification methods to develop crop-type maps in Rahim Yar Khan, Pakistan. Using the Google Earth Engine (GEE) cloud platform, pre-processing steps were applied to Sentinel-1 and Sentinel-2 data. By integrating Sentinel 1 (VV, VH), Sentinel-2 satellite bands and different vegetation indices were computed from Sentinel-2 imagery, composite images were also produced for subsequent assessment. The main objective was to evaluate the efficiency of the approached to classify the major cotton, rice and sugarcane. Time specific images were also use to exploit to seasonality of the crops, for example, composite image of August was prioritized to distinguish cotton, while September composite image was used for rice and sugarcane classification. The study employs two approaches for object-based segmentation: the Simple Non-Iterative Clustering (SNIC) in GEE platform and Object based Image Classification (OBIA) using Multi-resolution segmentation in eCognition software. Machine learning algorithm Random Forest (RF) was applied over the composite image for both pixel-based approach and object-based approaches. The study utilized field sample data collected for classification, validation, and accuracy assessment. The ground survey data includes cotton, rice, sugarcane, orchard, and other crops. Comparative analysis was carried out to assess the performance of pixel-based and object-based approaches. RF on pixel-based approach of Sentinel-1 and Sentinel-2 imagery bands with indices composite showed superior results. RF on pixel based approached classification achieved 98% accuracy and kappa 92%, while RF on SNIC in GEE achieved 96% accuracy and kappa 95%, and OBIA in eCognition achieved accuracy of 89%. We also conclude that the tuning of segmentation in both eCognition and SNIC algorithm can improve the accuracy of object-based classification.

References

K. Ennouri, “Remote Sensing : An Advanced Technique for Crop,” vol. 2019, 2019.

K. Heupel, D. Spengler, and S. Itzerott, “ORIGINAL ARTICLE A Progressive Crop-Type Classification Using Multitemporal Remote Sensing Data and Phenological Information,” PFG – J. Photogramm. Remote Sens. Geoinf. Sci., vol. 86, no. 2, pp. 53–69, 2018, doi: 10.1007/s41064-018-0050-7.

D. Tilman, C. Balzer, J. Hill, and B. L. Befort, “Global food demand and the sustainable intensification of agriculture,” Proc. Natl. Acad. Sci., vol. 108, no. 50, pp. 20260–20264, Dec. 2011, doi: 10.1073/pnas.1116437108.

A. Shelestov, M. Lavreniuk, N. Kussul, A. Novikov, and S. Skakun, “Exploring Google Earth Engine Platform for Big Data Processing: Classification of Multi-Temporal Satellite Imagery for Crop Mapping,” Front. Earth Sci., vol. 5, Feb. 2017, doi: 10.3389/feart.2017.00017.

S. Murod, M. Ibra, A. Mustakim, and V. Du, “A crop type dataset for consistent land cover classification in Central Asia,” pp. 1–6, 2020, doi: 10.1038/s41597-020-00591-2.

A. Khaliq, L. Peroni, and M. Chiaberge, “Land Cover and Crop Classification using Multitemporal Sentinel-2 Images Based on Crops Phenological Cycle,” 2018 IEEE Work. Environ. Energy, Struct. Monit. Syst., pp. 1–5, 2018.

A. Orynbaikyzy, U. Gessner, and C. Conrad, “Crop type classification using a combination of optical and radar remote sensing data : a review,” Int. J. Remote Sens., vol. 00, no. 00, pp. 1–43, 2019, doi: 10.1080/01431161.2019.1569791.

E. Beriaux, A. Jago, C. Lucau-danila, V. Planchon, and P. Defourny, “Sentinel-1 Time Series for Crop Identification in the Framework of the Future CAP Monitoring,” pp. 1–29, 2021.

A. Orynbaikyzy, U. Gessner, B. Mack, and C. Conrad, “Crop Type Classification Using Fusion of Sentinel-1 and Sentinel-2 Data : Assessing the Impact of Feature Selection , Optical Data Availability , and Parcel Sizes on the Accuracies,” 2020.

G. Waldho, U. Lussem, and G. Bareth, “Int J Appl Earth Obs Geoinformation,” vol. 61, no. April, pp. 55–69, 2017, doi: 10.1016/j.jag.2017.04.009.

G. R. Faqe Ibrahim, A. Rasul, and H. Abdullah, “Improving Crop Classification Accuracy with Integrated Sentinel-1 and Sentinel-2 Data: a Case Study of Barley and Wheat,” J. Geovisualization Spat. Anal., vol. 7, no. 2, pp. 1–15, 2023, doi: 10.1007/s41651-023-00152-2.

L. Yang, L. R. Mansaray, and J. Huang, “Optimal Segmentation Scale Parameter , Feature Subset and Classification Algorithm for Geographic Object-Based Crop Recognition Using Multisource Satellite Imagery”, doi: 10.3390/rs11050514.

E. O. Makinde, A. T. Salami, J. B. Olaleye, and O. C. Okewusi, “Object Based and Pixel Based Classification Using Rapideye Satellite Imager of ETI-OSA, Lagos, Nigeria,” Geoinformatics FCE CTU, vol. 15, no. 2, pp. 59–70, 2016, doi: 10.14311/gi.15.2.5.

J. Brinkhoff, J. Vardanega, and A. J. Robson, “Land Cover Classification of Nine Perennial Crops Using Sentinel-1 and -2 Data,” pp. 1–26, 2020, doi: 10.3390/rs12010096.

A. Paludo, W. R. Becker, J. Richetti, L. C. De, A. Silva, and J. A. Johann, “Mapping summer soybean and corn with remote sensing on Google Earth Engine cloud computing in Parana state – Brazil,” 2020, doi: 10.1080/17538947.2020.1772893.

S. Ahmed, “Comparison of Satellite Images Classification Techniques using Landsat-8 Data for Land Cover Extraction,” Int. J. Intell. Comput. Inf. Sci., vol. 0, no. 0, pp. 1–15, 2021, doi: 10.21608/ijicis.2021.78853.1098.

Y. Jin, X. Liu, Y. Chen, X. Liang, and Y. Chen, “Land-cover mapping using Random Forest classification and incorporating NDVI time-series and texture : a case study of central Shandong incorporating NDVI time-series and texture : a case study of central Shandong,” Int. J. Remote Sens., vol. 39, no. 23, pp. 8703–8723, 2018, doi: 10.1080/01431161.2018.1490976.

V. F. Rodriguez-Galiano and M. Chica-Rivas, “Evaluation of different machine learning methods for land cover mapping of a Mediterranean area using multi-seasonal Landsat images and Digital Terrain Models,” Int. J. Digit. Earth, vol. 7, no. 6, pp. 492–509, Jul. 2014, doi: 10.1080/17538947.2012.748848.

G. De Luca, M. N. Silva, S. Cerasoli, J. Campos, S. Di Fazio, and G. Modica, “Object-Based Land Cover Classification of Cork Oak Woodlands using UAV Imagery and Orfeo ToolBox,” 2019.

Y. Wang, Z. Li, C. Zeng, G. Xia, and S. Member, “An Urban Water Extraction Method Combining Deep Learning and Google Earth Engine,” vol. 13, pp. 769–782, 2020.

M. J. Steinhausen, P. D. Wagner, B. Narasimhan, and B. Waske, “Combining Sentinel-1 and Sentinel-2 data for improved land use and land cover mapping of monsoon regions,” Int. J. Appl. Earth Obs. Geoinf., vol. 73, pp. 595–604, Dec. 2018, doi: 10.1016/j.jag.2018.08.011.

A. Moumni and A. Lahrouni, “Machine Learning-Based Classification for Crop-Type Mapping Using the Fusion of High-Resolution Satellite Imagery in a Semiarid Area,” Scientifica (Cairo)., vol. 2021, pp. 1–20, Apr. 2021, doi: 10.1155/2021/8810279.

R. Achanta and S. Sabine, “Superpixels and Polygons using Simple Non-Iterative Clustering,” no. Ic, pp. 4651–4660.

M. Mahdianpari, B. Salehi, and F. Mohammadimanesh, “The First Wetland Inventory Map of Newfoundland at a Spatial Resolution of 10 m Using Sentinel-1 and Sentinel-2 Data on the Google Earth Engine Cloud Computing Platform”, doi: 10.3390/rs11010043.

M. Vizzari, G. Lesti, S. Acharki, and M. Vizzari, “Geo-spatial Information Science Crop classification in Google Earth Engine : leveraging Sentinel-1 , Sentinel-2 , European CAP data , and object-based machine-learning approaches European CAP data , and object-based machine-learning approaches,” Geo-spatial Inf. Sci., vol. 00, no. 00, pp. 1–16, 2024, doi: 10.1080/10095020.2024.2341748.

E. Engine, “Accuracy Improvements to Pixel-Based and Object-Based LULC Classification with Auxiliary Datasets from Google Earth Engine,” 2021.

H. Gao, L. He, Z. wei He, and W. qian Bai, “Early landslide mapping with slope units division and multi-scale object-based image analysis — A case study in the Xianshui River basin of Sichuan, China,” J. Mt. Sci., vol. 19, no. 6, pp. 1618–1632, 2022, doi: 10.1007/s11629-022-7333-6.

Z. Ye et al., “A comparison between Pixel-based deep learning and Object-based image analysis ( OBIA ) for individual detection of cabbage plants based on UAV Visible-light images,” Comput. Electron. Agric., vol. 209, no. April, p. 107822, 2023, doi: 10.1016/j.compag.2023.107822.

T. Lu, M. Gao, and L. Wang, “Crop classi fi cation in high- resolution remote sensing images based on multi-scale feature fusion semantic segmentation model,” no. August, pp. 1–16, 2023, doi: 10.3389/fpls.2023.1196634.

S. Liu, Z. Qi, X. Li, and A. G. Yeh, “Integration of Convolutional Neural Networks and Object-Based Post-Classification Refinement for Land Use and Land Cover Mapping with Optical and SAR Data,” pp. 1–25, 2019, doi: 10.3390/rs11060690.

Q. Guo et al., “Urban Tree Classification Based on Object-Oriented Approach and Random Forest Algorithm Using Unmanned Aerial Vehicle (UAV) Multispectral Imagery,” Remote Sens., vol. 14, no. 16, p. 3885, Aug. 2022, doi: 10.3390/rs14163885.

H. Xue et al., “Object-Oriented Crop Classification Using Time Series Sentinel Images from Google Earth Engine,” 2023.