Reconstrucción de imágenes hiperespectrales superresueltas entre plataformas basada en CMIFM y evaluación espectral cuantitativa: un caso de humedales kársticos

  • role: First author第一作者
  • Affiliation:

    College of Geomatics and Geoinformation, Guilin University of Technology, Guilin 541006, China

  • Email:2120211897@glut.edu.com
  • Introduction:湿E-mail 2120211897@glut.edu.com
SUN Xidong1,  
  • role: Corresponding author通信作者
  • Affiliation:

    College of Geomatics and Geoinformation, Guilin University of Technology, Guilin 541006, China

  • Email:fbl2012@126.com
  • Introduction:湿E-mail fbl2012@126.com
FU Bolin1*,  
  • Affiliation:

    College of Geomatics and Geoinformation, Guilin University of Technology, Guilin 541006, China

LI Huajian1,  
  • Affiliation:

    Laboratory of Wetland Ecology and Environment, Northeast Institute of Geography and Agroecology, Chinese Academy of Sciences, Changchun 130102, China

JIA Mingming2,  
  • Affiliation:

    Department of Geography and Spatial Information Techniques, Ningbo University, Ningbo 315211, China

SUN Weiwei3,  
  • Affiliation:

    College of Geomatics and Geoinformation, Guilin University of Technology, Guilin 541006, China

WU Yan1,  
  • Affiliation:

    College of Geomatics and Geoinformation, Guilin University of Technology, Guilin 541006, China

SONG Yiji1

resumen

La monitorización temporal precisa de la vegetación y cuerpos de agua en humedales mediante imágenes hiperespectrales se ha convertido en una base importante para la evaluación precisa y la supervisión integral del estado de los ecosistemas de humedales kársticos. Sin embargo, la resolución espacial de las imágenes hiperespectrales satelitales (Satellite-based HSI) es baja, lo que dificulta capturar detalles espaciales complejos de la vegetación en humedales. Los métodos existentes de reconstrucción de superresolución no pueden realizar reconstrucción cruzada entre plataformas, desde satélite hasta dron, ni monitoreo temporal de escenas extensas de humedales. Por ello, este estudio propone un módulo de mapeo de características de imagen multiescala entre sensores CMIFM (Cross-Sensor Multiscale Image Feature Mapping Module). Este módulo primero alinea y transforma espacialmente las imágenes hiperespectrales capturadas por vehículos aéreos no tripulados (UAV-HSI) y las imágenes satelitales (Satellite-based HSI). Luego, basado en datos de campo obtenidos con un espectrómetro portátil ASD (Analytical spectral devices), mapea UAV-HSI y Satellite-based HSI a un espacio espectral unificado, permitiendo la integración de información espaciotemporal. Finalmente, utiliza redes de superresolución (ESRGAN y SwinIR) para lograr la reconstrucción de imágenes hiperespectrales de alta calidad a partir de Satellite-based HSI. Además, este estudio emplea redes de aprendizaje profundo (DATFuse) y métodos tradicionales de fusión (GS) como comparación para evaluar cuantitativamente la calidad espectral y espacial de las comunidades vegetales y cuerpos de agua en los resultados de reconstrucción y fusión de imágenes Sentinel-2 y OHS-02. Los resultados muestran que: (1) la red de superresolución basada en CMIFM mejora la resolución espacial de Satellite-based HSI aprendiendo las características espaciotemporales de UAV-HSI, restaurando detalles finos de la textura espacial de plantas y cuerpos de agua, superando visualmente y en métricas cuantitativas al método de fusión GS, con una precisión promedio de PSNR y SSIM para Sentinel-2 y OHS-02 de 11.06 y 0.3102 respectivamente; (2) los valores espectrales de tres comunidades vegetales típicas de humedales — Kikuyugrass (狗牙根), Hualarsha (华克拉莎) y hierba (芒草) — y cuerpos de agua en las imágenes reconstruidas presentan alta confiabilidad, con una RMSE promedio por banda y precisión R2 entre datos ASD y OHS-02 de 0.1154 y 0.7239 respectivamente; (3) los métodos CMIFM+ESRGAN y CMIFM+SwinIR muestran fuerte capacidad de generalización en la reconstrucción espaciotemporal, pudiendo reconstruir imágenes en escenarios de humedales similares sin datos UAV-HSI, con precisión promedio de PSNR y SSIM de 12.74 y 0.1897 respectivamente; (4) finalmente, el estudio verifica la viabilidad de la tecnología de superresolución basada en CMIFM en la reconstrucción de imágenes hiperespectrales complejas de humedales.

palabra clave

humedales kársticos; módulo CMIFM; reconstrucción superresuelta entre plataformas; DATFuse; imágenes hiperespectrales; evaluación cuantitativa de la calidad de reconstrucción espaciotemporal de vegetación y cuerpos de agua

References

  1. 1.
    Aiazzi B, Baronti S, Lotti F and Selva M. 2009. A comparison between global and context-adaptive pansharpening of multispectral images. IEEE Geoscience and Remote Sensing Letters, 6(2): 302-306
  2. 2.
    Alparone L, Baronti S, Aiazzi B and Garzelli A. 2016. Spatial methods for multispectral pansharpening: multiresolution analysis demystified. IEEE Transactions on Geoscience and Remote Sensing, 54(5): 2563-2576
  3. 3.
    Bendoumi M A, He M Y and Mei S H. 2014. Hyperspectral image resolution enhancement using high-resolution multispectral image based on spectral unmixing. IEEE Transactions on Geoscience and Remote Sensing, 52(10): 6574-6583
  4. 4.
    Cai J T, Fu B L, Chen T X, Geng R F, Li Y, He H C, Fan D L and Deng T F. 2020. Inversion of physical and chemical parameters of plants in Huixian Karst Wetlands based on Sentinel-2 multispectral data. Wetland Science, 18(6): 693-705
  5. 5.
    Chen C, Xiong Z W, Tian X M, Zha Z J and Wu F. 2019. Camera lens super-resolution//Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach: IEEE: 1652-1660
  6. 6.
    Daniel T J, Richendrfer J, Falta R, Murdoch L, Lin H and Darnault C J G. 2023. Hydrogeologic and geomorphic processes in a karst landscape and seasonably-cold climate: linking spatial distribution and morphometric dynamics of closed depressions to bedrock fractures in a wastewater spray irrigated agricultural and forest system located at the site of the Living Filter in Central Pennsylvania, United States. Agricultural Water Management, 279: 108170
  7. 7.
    Fu B L, Li S Z, Lao Z N, Yuan B Y, Liang Y Y, He W, Sun W W and He H C. 2023b. Multi-sensor and multi-platform retrieval of water chlorophyll a concentration in karst wetlands using transfer learning frameworks with ASD, UAV, and Planet CubeSate reflectance data. Science of the Total Environment, 901: 165963
  8. 8.
    Fu B L, Sun X D, Li Y Y, Lao Z N, Deng T F, He H C, Sun W W and Zhou G Q. 2023a. Combination of super-resolution reconstruction and SGA-Net for marsh vegetation mapping using multi-resolution multispectral and hyperspectral images. International Journal of Digital Earth, 16(1): 2724-2761
  9. 9.
    Fu B L, Zuo P P, Liu M, Lan G W, He H C, Lao Z N, Zhang Y, Fan D L and Gao E T. 2022. Classifying vegetation communities karst wetland synergistic use of image fusion and object-based machine learning algorithm with Jilin-1 and UAV multispectral images. Ecological Indicators, 140: 108989
  10. 10.
    Ghassemian H. 2016. A review of remote sensing image fusion methods. Information Fusion, 32: 75-89
  11. 11.
    Gu J J, Lu H N, Zuo W M and Dong C. 2019. Blind super-resolution with iterative kernel correction//Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach: IEEE: 1604-1613
  12. 12.
    Han Y, Ke Y H, Wang Z P, Liang D Y and Zhou D M. 2023. Classification of the Yellow River Delta wetland landscape based on ZY-1 02D hyperspectral imagery. National Remote Sensing Bulletin, 27(6): 1387-1399
  13. 13.
    Hore A and Ziou D. 2010. Image quality metrics: PSNR vs. SSIM//2010 20th International Conference on Pattern Recognition. Istanbul: IEEE: 2366-2369
  14. 14.
    Lassalle G, Ferreira M P, La Rosa L E C, Scafutto R D P M and de Souza Filho C R. 2023. Advances in multi- and hyperspectral remote sensing of mangrove species: a synthesis and study case on airborne and multisource spaceborne imagery. ISPRS Journal of Photogrammetry and Remote Sensing, 195: 298-312
  15. 15.
    Li Z, Jin Z and Li Q. 2017. Changes in land use and their effects on soil properties in Huixian Karst wetland system. Polish Journal of Environmental Studies, 26(2): 699-707
  16. 16.
    Liang J Y, Cao J Z, Sun G L, Zhang K, Van Gool L and Timofte R. 2021. Swinir: image restoration using swin transformer//Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW). Montreal: IEEE: 1833-1844
  17. 17.
    Maeda S. 2020. Unpaired image super-resolution using pseudo-supervision//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE: 291-300
  18. 18.
    Metternicht G. 2003. Vegetation indices derived from high-resolution airborne videography for precision crop management. International Journal of Remote Sensing, 24(14): 2855-2877
  19. 19.
    Peng K F, Jiang W G, Hou P, Wu Z F, Ling Z Y, Wang X Y, Niu Z G and Mao D H. 2023. Continental-scale wetland mapping: a novel algorithm for detailed wetland types classification based on time series Sentinel-1/2 images. Ecological Indicators, 148: 110113
  20. 20.
    Pu R L, Gong P and Yu Q. 2008. Comparative analysis of EO-1 ALI and Hyperion, and Landsat ETM+ data for mapping forest crown closure and leaf area index. Sensors, 8(6): 3744-3766
  21. 21.
    Soh J W, Cho S and Cho N I. 2020. Meta-transfer learning for zero-shot super-resolution//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE: 3513-3522
  22. 22.
    Sun W W, Liu W W, Wang Y M, Zhao R, Huang M Z, Wang Y, Yang G and Meng X C. 2023. Research progress and prospects of hyperspectral remote sensing for global wetland from 2010 to 2022. National Remote Sensing Bulletin, 27(6): 1281-1299
  23. 23.
    Sun W W, Ren K, Meng X C, Yang G, Xiao C C, Peng J T and Huang J F. 2022. MLR-DBPFN: a multi-scale low rank deep back projection fusion network for anti-noise hyperspectral and multispectral image fusion. IEEE Transactions on Geoscience and Remote Sensing, 60: 5522914
  24. 24.
    Tang W, He F Z, Liu Y, Duan Y S and Si T Z. 2023. DATFuse: infrared and visible image fusion via dual attention transformer. IEEE Transactions on Circuits and Systems for Video Technology, 33(7): 3159-3172
  25. 25.
    Waheed M, Haq S M, Arshad F, Bussmann R W, Ali H M and Siddiqui M H. 2023. Phyto-ecological distribution patterns and identification of alien invasive indicator species in relation to edaphic factors from semi-arid region. Ecological Indicators, 148: 110053
  26. 26.
    Wang F M, Huang J F, Tang Y L and Wang X Z. 2007. New vegetation index and its application in estimating leaf area index of rice. Rice Science, 14(3): 195-203
  27. 27.
    Wang P J, Bayram B and Sertel E. 2022. A comprehensive review on deep learning based remote sensing image super-resolution methods. Earth-Science Reviews, 232: 104110
  28. 28.
    Wang X T, Yu K, Wu S X, Gu J J, Liu Y H, Dong C, Qiao Y and Loy C C. 2018. ESRGAN: enhanced super-resolution generative adversarial networks//Proceedings of the European Conference on Computer Vision (ECCV). Munich: Springer: 63-79
  29. 29.
    Yang C H, Everitt J H, Bradford J M and Murden D. 2004. Airborne hyperspectral imagery and yield monitor data for mapping cotton yield variability. Precision Agriculture, 5(5): 445-461
  30. 30.
    Yang W L, Fu B L, Li S Z, Lao Z N, Deng T F, He W, He H C and Chen Z K. 2023. Monitoring multi-water quality of internationally important karst wetland through deep learning, multi-sensor and multi-platform remote sensing images: a case study of Guilin, China. Ecological Indicators, 154: 110755
  31. 31.
    Zhang H Y, Cisse M, Dauphin Y N and Lopez-Paz D. 2018. mixup: beyond empirical risk minimization. arXiv preprint arXiv: 1710.09412
  32. 32.
    Zhang Z, Jiang W G, Peng K F, Wu Z F, Ling Z Y and Li Z. 2023. Assessment of the impact of wetland changes on carbon storage in coastal urban agglomerations from 1990 to 2035 in support of SDG15.1. Science of the Total Environment, 877: 162824

Leer el texto completo

The above content is generated by Large Model Translation. The translated content is for reference only. We do not assume any commercial or legal responsibilty for any consequences arising from the use of our website