{"title":"Reconstruction of hyperspectral images with compressed sensing based on linear mixing model and affinity propagation clustering algorithm","authors":"Youli zou, Zhi-yun Xiao, Kuntao Ye","doi":"10.1145/3599589.3599602","DOIUrl":null,"url":null,"abstract":"The increasing spatial and spectral resolution of hyperspectral images results in a significant rise in data volume, which poses a challenge for data storage and transmission. Therefore, improving the efficiency of storage and transmission by enhancing the reconstruction performance of hyperspectral images at low sampling rates or same sampling rates conditions is a crucial topic in compressed sensing. Previous research has shown that a linear mixing model and distributed compressed sensing method outperform traditional compressed sensing reconstruction algorithms in recovering original data. However, the low estimating accuracy of both the endmembers matrix and abundance matrix due to the random selection of reference bands limits the reconstruction performance. To address this problem, we proposed a compressed sensing reconstruction algorithm based on a linear mixing model and affinity propagation clustering algorithm. Our method improves reconstruction performance by enhancing the estimating accuracy of the endmembers and abundance matrices. During the sampling stage, the affinity propagation clustering algorithm is used to group the spectral bands according to the spectral correlation of hyperspectral images, where the clustering center serving as the reference band and the other bands as non-reference bands. During the reconstruction stage, the number of endmembers from the reference band is estimated fist, and the endmembers matrix and the abundance matrix are then estimated. Finally, the endmembers matrix and estimated abundance matrix are used for reconstruction. Experimental results show that our proposed algorithm achieves higher performance in reconstructing hyperspectral images than the linear mixing model-based distributed compressed sensing method.","PeriodicalId":123753,"journal":{"name":"Proceedings of the 2023 8th International Conference on Multimedia and Image Processing","volume":"24 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the 2023 8th International Conference on Multimedia and Image Processing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3599589.3599602","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
The increasing spatial and spectral resolution of hyperspectral images results in a significant rise in data volume, which poses a challenge for data storage and transmission. Therefore, improving the efficiency of storage and transmission by enhancing the reconstruction performance of hyperspectral images at low sampling rates or same sampling rates conditions is a crucial topic in compressed sensing. Previous research has shown that a linear mixing model and distributed compressed sensing method outperform traditional compressed sensing reconstruction algorithms in recovering original data. However, the low estimating accuracy of both the endmembers matrix and abundance matrix due to the random selection of reference bands limits the reconstruction performance. To address this problem, we proposed a compressed sensing reconstruction algorithm based on a linear mixing model and affinity propagation clustering algorithm. Our method improves reconstruction performance by enhancing the estimating accuracy of the endmembers and abundance matrices. During the sampling stage, the affinity propagation clustering algorithm is used to group the spectral bands according to the spectral correlation of hyperspectral images, where the clustering center serving as the reference band and the other bands as non-reference bands. During the reconstruction stage, the number of endmembers from the reference band is estimated fist, and the endmembers matrix and the abundance matrix are then estimated. Finally, the endmembers matrix and estimated abundance matrix are used for reconstruction. Experimental results show that our proposed algorithm achieves higher performance in reconstructing hyperspectral images than the linear mixing model-based distributed compressed sensing method.