A Comparative Study of Fusion-Based Change Detection Methods for Multi-Band Images with Different Spectral and Spatial Resolutions

Vinicius Ferraris, Naoto Yokoya, Nicolas Dobigeon, Marie Chabert
2018 IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium  
Open Archive Toulouse Archive Ouverte OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible To cite this version: Ferraris, Vinicius and Yokoya, Naoto and Dobigeon, Nicolas and Chabert, Marie A comparative study of fusion-based change detection methods for multi-band images with different spectral and spatial resolutions. (ABSTRACT This paper deals with a fusion-based change detection (CD) framework for
more » ... and images with different spatial and spectral resolutions. The first step of the considered CD framework consists in fusing the two observed images. The resulting fused image is subsequently spatially or spectrally degraded to produce two pseudoobserved images, with the same resolutions as the two observed images. Finally, CD can be performed through a pixel-wise comparison of the pseudo-observed and observed images since they share the same resolutions. Obviously, fusion is a key step in this framework. Thus, this paper proposes to quantitatively and qualitatively compare state-of-the-art fusion methods, gathered into four main families, namely component substitution, multi-resolution analysis, unmixing and Bayesian, with respect to the performance of the whole CD framework evaluated on simulated and real images. Index Terms-Change detection, hyperspectral and multispectral imaging, fusion, heterogeneous sensors. t j LR ∈ R m λ ×m where n λ (resp., m λ ) and n (resp. m) denote the numbers of bands and pixels of the HR (resp. LR) images with m λ > n λ and n > m. These observed images can be seen as the individual results of two degradations of two (unobserved) high spatial and spectral resolution latent images X t k ∈ R m λ ×n (k = i, j) by two operators denoted by THR [·] and TLR [·], respectively. Note that the two unobserved la-
doi:10.1109/igarss.2018.8517712 dblp:conf/igarss/FerrarisYDC18 fatcat:3wrv6eda5nb5hmr2yur3nxaxqi