Centre de documentation
scientifique

A convolution neural network for forest leaf chlorophyll and carotenoid estimation using hyperspectral reflectance / Shuo Shi in International journal of applied Earth observation and geoinformation, vol 108 (April 2022)  

Public [article]  inInternational journal of applied Earth observation and geoinformation > vol 108 (April 2022) . - n° 102719 Titre : A convolution neural network for forest leaf chlorophyll and carotenoid estimation using hyperspectral reflectance Type de document : Article/Communication Auteurs : Shuo Shi, Auteur ; Lu Xu, Auteur ; Wei Gong, Auteur ; et al., Auteur Année de publication : 2022 Article en page(s) : n° 102719 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] chlorophylle [Termes IGN] classification par réseau neuronal convolutif [Termes IGN] écosystème forestier [Termes IGN] feuille (végétation) [Termes IGN] modèle de transfert radiatif [Termes IGN] processus gaussien [Termes IGN] réflectance spectrale [Termes IGN] régression Résumé : (auteur) Forest leaf chlorophyll (Cab) and carotenoid (Cxc) are key functional indicators for the state of the forest ecosystem. Current machine learning models based on hyperspectral reflectance are widely applied to estimate leaf Cab and Cxc contents at leaf scale. However, these models have certain accuracy for non-independent datasets but have poor generalization for independent datasets when they are used to estimate leaf Cab and Cxc contents. This fact limits that hyperspectral remote sensing completely replaces destructive measurements for leaf Cab and Cxc contents. Thus, the development of an estimation model with high accuracy and satisfactory generalization is necessary. Convolutional neural networks (CNNs) have certain accuracy and generalization in many domains, and have the potential to solve above-mentioned problem. Therefore, this study developed a CNN using one-dimensional hyperspectral reflectance, which aimed to improve the model's accuracy and generalization in leaf Cab and Cxc content estimation at leaf scale. The proposed CNN was developed by three steps. First, in consideration of the correlation between leaf Cab and Cxc contents in natural leaves, 2500 physical data with leaf reflectance and corresponding Cab and Cxc contents were generated by leaf radiative transfer model and multivariable gaussian distribution function. Then, the proposed CNN was built by five strategies based on the architecture of the AlexNet. Finally, five-fold cross validation was performed with 70% of the physical data to determine the best strategy to develop the proposed CNN. These were executed to ensure the proposed CNN with the maximum accuracy and generalization. In addition, the accuracy and generalization of the proposed CNN were tested using a non-independent dataset and an independent dataset, respectively. The proposed CNN was also compared with back propagation neural network (BPNN), support vector regression (SVR) and gaussian process regression (GPR). Results showed that the best CNN could be developed with one input, five convolutional, three max-pooling and three fully-connected layers. Comprehensively considering the model's accuracy and generalization, the proposed CNN was the best model for leaf Cab and Cxc content estimation compared with BPNN, SVR and GPR. This study provides a development strategy of CNN estimation model using one-dimensional hyperspectral reflectance at leaf scale. The proposed CNN could further promote the practical application of hyperspectral remote sensing in leaf Cab and Cxc content estimation. Numéro de notice : A2022-231 Affiliation des auteurs : non IGN Thématique : FORET/IMAGERIE Nature : Article DOI : 10.1016/j.jag.2022.102719 Date de publication en ligne : 16/02/2022 En ligne : https://doi.org/10.1016/j.jag.2022.102719 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100119 [article]

A new waveform decomposition method for multispectral LiDAR / Shalei Song in ISPRS Journal of photogrammetry and remote sensing, vol 149 (March 2019)  

Public [article]  inISPRS Journal of photogrammetry and remote sensing > vol 149 (March 2019) . - pp 40 - 49 Titre : A new waveform decomposition method for multispectral LiDAR Type de document : Article/Communication Auteurs : Shalei Song, Auteur ; Binhui Wang, Auteur ; Wei Gong, Auteur ; et al., Auteur Année de publication : 2019 Article en page(s) : pp 40 - 49 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Lasergrammétrie [Termes IGN] couvert végétal [Termes IGN] décomposition de Gauss [Termes IGN] données lidar [Termes IGN] données localisées 3D [Termes IGN] extraction de la végétation [Termes IGN] extraction de traits caractéristiques [Termes IGN] forme d'onde pleine [Termes IGN] transformation en ondelettes Résumé : (Auteur) Information derived from waveform decomposition of full-waveform light detection and ranging (LiDAR) data has been widely used in vegetation detection and three-dimensional urban terrain modeling to investigate and interpret the structural diversity of surface coverage. Most prevailing waveform decomposition methods involve only a single wavelength, but these methods do not apply to full-waveform multispectral LiDAR (FWMSL) systems that simultaneously acquire spectral and geometric information. In this paper, we propose a new multispectral waveform decomposition (MSWD) method in order to explore the potential advantages of the FWMSL system. Both simulated data and measured data from our FWMSL system were used to evaluate the performance of the proposed method. The coefficient of determination (R2), root mean square error (RMSE), and relative error (rRMSE) metrics suggest that the decomposition results derived from MSWD exhibit a comparable overall fitting accuracy as a single wavelength waveform decomposition (SWWD) method. We also propose a new evaluation indicator, relative neighbor distance error (RNDE), to represent the relative error in the distance between adjacent targets. The simulation results present clear superiority of MSWD over SWWD in terms of discovering weak or overlapping components and retrieving accurate waveform parameters. The experimental results demonstrated a considerable improvement in RNDE (0.0100–0.0610) over the prevailing SWWD method (0.0566–0.2833). Unlike SWWD, MSWD initializes waveform components using mutually complementary wavelengths thus delivering higher completeness and accuracy. MSWD can be extended to other FWMSL or full-waveform hyperspectral LiDAR systems with additional wavelengths. Numéro de notice : A2019-127 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1016/j.isprsjprs.2019.01.014 Date de publication en ligne : 22/01/2019 En ligne : https://doi.org/10.1016/j.isprsjprs.2019.01.014 Format de la ressource électronique : URL Article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=92438 [article]

Exemplaires(3)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
081-2019031	RAB	Revue	Centre de documentation	En réserve L003	Disponible
081-2019033	DEP-RECP	Revue	LASTIG	Dépôt en unité	Exclu du prêt
081-2019032	DEP-RECF	Revue	Nancy	Dépôt en unité	Exclu du prêt