Semantic segmentation of forest stands of pure species combining airborne lidar data and very high resolution multispectral imagery

Dechesne, Clément; Mallet, Clément; Bris, Arnaud Le; Gouet-Brunet, Valérie

doi:10.1016/j.isprsjprs.2017.02.011

Cited by 57 publications

(44 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…These are the necessary inputs for all methods described below. In practice, the strategy proposed in (Dechesne et al, 2017) is as followed: a supervised classification is performed on a selection of features extracted both from 3D lidar point clouds and aerial multispectral images. The training pixels are selected according to an existing forest LC geodatabase.…”

Section: General Strategymentioning

confidence: 99%

“…In (Dechesne et al, 2017), a high number of features was extracted from available lidar and optical images (∼ 100) but can be selected. They can also be weighted according to their importance, computed through the Random Forest classification process.…”

Section: Priormentioning

confidence: 99%

“…One should note that the literature remains focused on individual tree extraction and tree species classification, developing site-specific workflows with similar advantages, drawbacks and classification performance. Consequently, no operational framework embedding the automatic analysis of remote sensing data has been yet proposed in the literature for forest stand segmentation (Dechesne et al, 2017). More surprisingly, only few methods have addressed such an issue from a research perspective.…”

Section: Introductionmentioning

confidence: 99%

“…Eventually, in (Dechesne et al, 2017), forest stand segmentation is considered from semantic segmentation point of view. Forest areas are first classified according to the tree species at the pixel and tree levels using lidar and multispectral airborne images.…”

Section: Introductionmentioning

confidence: 99%

“…In this paper, we specifically focus on semantic segmentation of forest stands through the regularization/smoothing process of an existing label map of pure species, following the strategy proposed in (Dechesne et al, 2017). Therefore, we build upon the vast amount of literature dealing with tree species classification at the tree level and investigate how the combined use of airborne lidar and VHR multispectral images can provide more accurate label maps.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Semantic Segmentation of Forest Stands of Pure Species as a Global Optimization Problem

Dechesne¹,

Mallet²,

Bris³

et al. 2017

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABSTRACT:Forest stand delineation is a fundamental task for forest management purposes, that is still mainly manually performed through visual inspection of geospatial (very) high spatial resolution images. Stand detection has been barely addressed in the literature which has mainly focused, in forested environments, on individual tree extraction and tree species classification. From a methodological point of view, stand detection can be considered as a semantic segmentation problem. It offers two advantages. First, one can retrieve the dominant tree species per segment. Secondly, one can benefit from existing low-level tree species label maps from the literature as a basis for high-level object extraction. Thus, the semantic segmentation issue becomes a regularization issue in a weakly structured environment and can be formulated in an energetical framework. This papers aims at investigating which regularization strategies of the literature are the most adapted to delineate and classify forest stands of pure species. Both airborne lidar point clouds and multispectral very high spatial resolution images are integrated for that purpose. The local methods (such as filtering and probabilistic relaxation) are not adapted for such problem since the increase of the classification accuracy is below 5%. The global methods, based on an energy model, tend to be more efficient with an accuracy gain up to 15%. The segmentation results using such models have an accuracy ranging from 96% to 99%.

show abstract

Section: General Strategymentioning

confidence: 99%

Section: Priormentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Semantic Segmentation of Forest Stands of Pure Species as a Global Optimization Problem

Dechesne¹,

Mallet²,

Bris³

et al. 2017

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

Performance Analysis of DeeplabV3+ Using State-of-the-Art Encoder Architectures for Waterbody Segmentation in Remote Sensing Images

Adarsh,

Sowmya,

Sivanpillai

et al. 2023

Lecture Notes in Networks and Systems

View full text Add to dashboard Cite

Stand delineation based on laser scanning data and simulated annealing

et al. 2021

View full text Add to dashboard Cite

The use of airborne laser scanning (LS) is increasing in forestry. Scanning can be conducted from manned aircrafts or unmanned aerial vehicles (UAV). The scanning data are often used to calculate various attributes for small raster cells. These attributes can be used to segment the forest into homogeneous areas, called segments, micro-stands, or, like in this study, stands. Delineation of stands from raster data is equal to finding the most suitable stand number for each raster cell, which is a combinatorial optimization problem. This study tested the performance of the simulated annealing (SA) metaheuristic in the delineation of stands from grids of UAV-LS attributes. The objective function included three criteria: within-stand variation of the LS attributes, stand area, and stand shape. The purpose was to create delineations that consisted of homogeneous stands with a low number of small stands and a regular and roundish stand shape. The results showed that SA is capable of producing stand delineations that meet these criteria. However, the method tended to produce delineations where the stands often consisted of disconnected parts and the stand borders were jagged. These problems were mitigated by using a mode filter on the grid of stand numbers and giving unique numbers for all disconnected parts of a stand. Three LS attributes were used in the delineation. These attributes described the canopy height, the height of the bottom of the canopy and the variation of echo intensity within 1-m2 raster cells. Besides, a texture variable that described the spatial variation of canopy height in the proximity of a 1-m2 raster cell was found to be a useful variable. Stand delineations where the average stand area was about one hectare explained more than 80% of the variation in canopy height.

show abstract

Semantic segmentation of forest stands of pure species combining airborne lidar data and very high resolution multispectral imagery

Cited by 57 publications

References 40 publications

Semantic Segmentation of Forest Stands of Pure Species as a Global Optimization Problem

Semantic Segmentation of Forest Stands of Pure Species as a Global Optimization Problem

Performance Analysis of DeeplabV3+ Using State-of-the-Art Encoder Architectures for Waterbody Segmentation in Remote Sensing Images

Stand delineation based on laser scanning data and simulated annealing

Contact Info

Product

Resources

About