Abstract:Accurate land use maps, describing the territory from an anthropic utilisation point of view, are useful tools for land management and planning. To produce them, the use of optical images alone remains limited. It is therefore necessary to make use of several heterogeneous sources, each carrying complementary or contradictory information due to their imperfections or their different specifications. This study compares two different approaches i.e. a pre-classification and a post-classification fusion approach for combining several sources of spatial data in the context of land use classification. The approaches are applied on authoritative land use data located in the Gers department in the southwest of France. Pre-classification fusion, while not explicitly modeling imperfections, has the best final results, reaching an overall accuracy of 97% and a macro-mean F1 score of 88%.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to distinguish three land use categories (LU2: secondary production, LU3: tertiary production, LU5: residential use) by comparing two data fusion methods - pre - classification fusion and post - classification fusion, in order to improve the accuracy of land use classification. Specifically, the research objectives are as follows: 1. **Multi - source data fusion**: Explore how to effectively combine multiple heterogeneous data sources, which may carry complementary or contradictory information because they each have defects or different specifications. The research assumes that multiple data sources can complement each other, and the machine - learning model can use these sources to infer land - use types, thus achieving better performance compared to a single source. 2. **Comparison of classification methods**: Compare the performance of the two methods of pre - classification fusion and post - classification fusion in land use classification. Pre - classification fusion combines all attributes before classification, and a machine - learning algorithm predicts land - use categories from all sources simultaneously; post - classification fusion first makes individual predictions for each source and then combines these prediction results to obtain the final classification. The research particularly focuses on the performance differences between these two methods when dealing with data source defects (such as incomplete information, low precision, etc.). 3. **Improve classification accuracy**: Through the application of the above two methods, the aim is to improve the overall accuracy rate and macro - average F1 - score of land use classification, especially for those land use categories that are difficult to identify (such as LU2 and LU3). The research results show that the pre - classification fusion method performs best in the final results, achieving an overall accuracy rate of 97% and a macro - average F1 - score of 88%. In summary, the core problem of this paper is to improve the accuracy of land use classification through data fusion techniques, especially for those land use categories that are difficult to distinguish by traditional remote sensing techniques.

Comparison of two data fusion approaches for land use classification

A Novel Image Fusion Method of Multi-Spectral and SAR Images for Land Cover Classification

A Decision Fusion Method for Land Cover Classification Using Multi-Sensor Data

Adaptively weighted decision fusion in 30 m land-cover mapping with Landsat and MODIS data

APPLICATION OF FUSION WITH SAR AND OPTICAL IMAGES IN LAND USE CLASSIFICATION BASED ON SVM

Integrating MODIS and Landsat Data for Land Cover Classification by Multilevel Decision Rule

Fusion of Google Street View, LiDAR, and Orthophoto Classifications Using Ranking Classes Based on F1 Score for Building Land-Use Type Detection

Self-supervised SAR-optical Data Fusion and Land-cover Mapping using Sentinel-1/-2 Images

A Fusion Method for Multisource Land Cover Products Based on Superpixels and Statistical Extraction for Enhancing Resolution and Improving Accuracy

Fusion of Heterogeneous Earth Observation Data for the Classification of Local Climate Zones

Urban land-use classification using machine learning classifiers: comparative evaluation and post-classification multi-feature fusion approach

Evaluation and analysis of fusion algorithms for active and passive remote sensing image.

Three-Dimensional Urban Land Cover Classification by Prior-Level Fusion of LiDAR Point Cloud and Optical Imagery

Comparison of Pixel-Based Fusion Between Multi-Spectral Data and High Spatial Resolution Data

Learning SAR-Optical Cross Modal Features for Land Cover Classification

Land use/land cover classification of fused Sentinel-1 and Sentinel-2 imageries using ensembles of Random Forests

Decision Fusion at Pixel Level of Multi-Band Data for Land Cover Classification-A Review

A Novel Classification Method Based on Texture Analysis Using High-Resolution SAR and Optical Data

Fusion de classifieurs pour la classification d'images sonar

Correlated Decision Fusion Accompanied with Quality Information on a Multi-Band Pixel Basis for Land Cover Classification

Integration of Classification Methods for Improvement of Land-Cover Map Accuracy