Dinh Ho Tong Minh

Dr. Dinh Ho Tong Minh

INRAE - UMR TETIS
orcid

Research Focus & Metrics

Academic Career Length

16Years (2011-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 June 2019 in Remote Sensing of Environment
Verified
Journal Article
Closed access
Shaun QueganShaun Quegan
Shaun Quegan
TTThuy Le Toan
Thuy Le Toan
JCJerome Chave
Jerome Chave
JDJørgen Dall
Jørgen Dall
JEJean-François Exbrayat
Jean-François Exbrayat
MWMathew Williams
Mathew Williams

1 December 2019 in ISPRS Journal of Photogrammetry and Remote Sensing
Verified
Journal Article
Closed access
Dino IencoDino Ienco
Dino Ienco
RIRoberto Interdonato
Roberto Interdonato
RGRaffaele Gaetano
Raffaele Gaetano
Dinh Ho Tong MinhDinh Ho Tong Minh
Dinh Ho Tong Minh

3 August 2018 in Remote Sensing
Verified
Journal Article
Open access
PDF file
ENEmile Ndikumana
Emile Ndikumana
Dinh Ho Tong MinhDinh Ho Tong Minh
Dinh Ho Tong Minh
Nicolas BaghdadiNicolas Baghdadi
Nicolas Baghdadi
DCDominique Courault
Dominique Courault
LHLaure Hossard
Laure Hossard
The development and improvement of methods to map agricultural land cover are currently major challenges, especially for radar images. This is due to the speckle noise nature of radar, leading to a less intensive use of radar rather than optical images. The European Space Agency Sentinel-1 constellation, which recently became operational, is a satellite system providing global coverage of Synthetic Aperture Radar (SAR) with a 6-days revisit period at a high spatial resolution of about 20 m. These data are valuable, as they provide spatial information on agricultural crops. The aim of this paper is to provide a better understanding of the capabilities of Sentinel-1 radar images for agricultural land cover mapping through the use of deep learning techniques. The analysis is carried out on multitemporal Sentinel-1 data over an area in Camargue, France. The data set was processed in order to produce an intensity radar data stack from May 2017 to September 2017. We improved this radar time series dataset by exploiting temporal filtering to reduce noise, while retaining as much as possible the fine structures present in the images. We revealed that even with classical machine learning approaches (K nearest neighbors, random forest, and support vector machines), good performance classification could be achieved with F-measure/Accuracy greater than 86% and Kappa coefficient better than 0.82. We found that the results of the two deep recurrent neural network (RNN)-based classifiers clearly outperformed the classical approaches. Finally, our analyses of the Camargue area results show that the same performance was obtained with two different RNN-based classifiers on the Rice class, which is the most dominant crop of this region, with a F-measure metric of 96%. These results thus highlight that in the near future these RNN-based techniques will play an important role in the analysis of remote sensing time series.

27 March 2013 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
Dinh Ho Tong MinhDinh Ho Tong Minh
Dinh Ho Tong Minh
TTThuy Le Toan
Thuy Le Toan
Fabio RoccaFabio Rocca
Fabio Rocca
Stefano TebaldiniStefano Tebaldini
Stefano Tebaldini
MDMauro Mariotti d'Alessandro
Mauro Mariotti d'Alessandro
LVLudovic Villard
Ludovic Villard
The retrieval of above-ground biomass (AGB) in dense tropical forests using synthetic aperture radar (SAR) images is widely recognized as a challenging task. The first difficulty arises from the decrease of sensitivity of the backscattered intensity to biomass at high biomass values, often referred to as the backscatter saturation effect. At P-band, the decrease of sensitivity can occur at biomass values higher than about 300 t ha -1 , e.g., those of many dense tropical forests. Another limiting factor is associated with the ground effects, as they can change significantly the magnitude of returns from vegetation-ground interactions. As a consequence, terrain topography or ground moisture status can determine the variations of the observed signal that are not due to forest biomass. A solution to reduce the ground effects is to have access to layers inside the forest canopy where the backscatter from vegetation-ground interactions is not significant. The study presented in this paper is an attempt to overcome the issues outlined above based on direct 3-D imaging of the forest volume, which is possible through multibaseline SAR tomography. In this way, forest biomass can be investigated by considering not only the backscattered power at each slant range and azimuth location but also its vertical distribution. The data analyzed in this paper are from the P-band airborne dataset acquired by Office National d'Études et de Recherches Aérospatiales (ONERA) over French Guiana in 2009, in the frame of the European Space Agency campaign TropiSAR. The dataset is characterized by a favorable baseline distribution, resulting in a vertical resolution less than half the forest height, which made it possible to decompose the vertical distribution of the backscattered power into a number of layers by coherent focusing, i.e., without assuming any prior knowledge about the forest vertical structure. For each layer, the relationship between the backscattered power and forest AGB was then analyzed. As expected, it was found that the power from the bottom layer is very weakly correlated to AGB, whereas the power from a layer at about 30 m above the ground yields the best correlation and sensitivity to AGB in all polarizations, for actual AGB values ranging from 250 to 450 t ha -1 . An interpretation of this result is also provided, based on a forest growth model simulation. Finally, the relevance of tomographic technique in P-band spaceborne mission is discussed.

11 April 2019 in Remote Sensing
Verified
Journal Article
Open access
PDF file
HBHassan Bazzi
Hassan Bazzi
Nicolas BaghdadiNicolas Baghdadi
Nicolas Baghdadi
MHMarcel M. El Hajj
Marcel M. El Hajj
Mehrez ZribiMehrez Zribi
Mehrez Zribi
Dinh Ho Tong MinhDinh Ho Tong Minh
Dinh Ho Tong Minh
Belhouchette HatemBelhouchette Hatem
Belhouchette Hatem
This study proposes an effective method to map rice crops using the Sentinel-1 SAR (Synthetic Aperture Radar) time series over the Camargue region, Southern France. First, the temporal behavior of the SAR backscattering coefficient over 832 plots containing different crop types was analyzed. Through this analysis, the rice cultivation was identified using metrics derived from the Gaussian profile of the VV/VH time series (3 metrics), the variance of the VV/VH time series (one metric), and the slope of the linear regression of the VH time series (one metric). Using the derived metrics, rice plots were mapped through two different approaches: decision tree and Random Forest (RF). To validate the accuracy of each approach, the classified rice map was compared to the available national data. Similar high overall accuracy was obtained using both approaches. The overall accuracy obtained using a simple decision tree reached 96.3%, whereas an overall accuracy of 96.6% was obtained using the RF classifier. The approach, therefore, provides a simple yet precise and powerful tool to map paddy rice areas.

1 March 2016 in Remote Sensing of Environment
Verified
Journal Article
Open access
Dinh Ho Tong MinhDinh Ho Tong Minh
Dinh Ho Tong Minh
TTThuy Le Toan
Thuy Le Toan
FRFabio Rocca
Fabio Rocca
Stefano TebaldiniStefano Tebaldini
Stefano Tebaldini
LVLudovic Villard
Ludovic Villard
JCJérôme Chave
Jérôme Chave
International audienceDeveloping and improving methods to monitor forest carbon in space and time is a timely challenge, especially for tropical forests. The next European Space Agency Earth Explorer Core Mission BIOMASS will collect synthetic aperture radar (SAR) data globally from employing a multiple baseline orbit during the initial phase of its lifetime. These data will be used for tomographic SAR (TomoSAR) processing, with a vertical resolution of about 20 m, a resolution sufficient to decompose the backscatter signal into two to three layers for most closed-canopy tropical forests. A recent study, conducted in the Paracou site, French Guiana, has already shown that TomoSAR significantly improves the retrieval of forest aboveground biomass (AGB) in a high biomass forest, with an error of only 10% at 1.5-ha resolution. However, the degree to which this TomoSAR approach can be transferred from one site to another has not been assessed. We test this approach at the Nouragues site in central French Guiana (ca 100 km away from Paracou), and develop a method to retrieve the top-of-canopy height from TomoSAR. We found a high correlation between the backscatter signal and AGB in the upper canopy layer (i.e. 20–40 m), while lower layers only showed poor correlations. The relationship between AGB and TomoSAR data was found to be highly similar for forests at Nouragues and Paracou. Cross validation using training plots from Nouragues and validation plots from Paracou, and vice versa, gave an error of 16–18% of AGB using 1-ha plots. Finally, using a high-resolution LiDAR canopy model as a reference, we showed that TomoSAR has the potential to retrieve the top-of-canopy height with an error to within 2.5 m. Our analyses show that the TomoSAR-AGB retrieval method is accurate even in hilly and high-biomass forest areas and suggest that our approach may be generalizable to other study sites, having a canopy taller than 30 m. These results have strong implications for the tomographic phase of the BIOMASS spaceborne mission