Forest change detection by statistical object-based method

Desclée, Baudouin; Bogaert, Pierre-Maurice; Defourny, Pierre

doi:10.1016/j.rse.2006.01.013

Cited by 436 publications

(234 citation statements)

References 44 publications

Supporting

Mentioning

210

Contrasting

Unclassified

Order By: Relevance

“…With the increasing applications of the object-based approach on widely available high-spatial resolution imagery, therefore, it would be interesting to test the object-based updating/backdating approach for change analysis using high-spatial resolution image data. The BOB approach had similar accuracies for land cover classification and change analysis with those from previous studies (Aguirre -Gutiérrez, Seijmonsbergen, & Duivenvoorden, 2012;Chen, Chen, Shi, & Yamaguchi, 2012;Desclee, Bogaert, & Defourny, 2006). But we did not conduct any manual editing for refinement, and did not use any ancillary data to aid in the change detection and classification.…”

Section: Discussionsupporting

confidence: 62%

A new approach for land cover classification and change analysis: Integrating backdating and an object-based method

Zhou

Qian

et al. 2016

Remote Sensing of Environment

153

View full text Add to dashboard Cite

Section: Discussionsupporting

confidence: 62%

A new approach for land cover classification and change analysis: Integrating backdating and an object-based method

Zhou

Qian

et al. 2016

Remote Sensing of Environment

153

View full text Add to dashboard Cite

“…Others have demonstrated the limitations of purely pixel-based image analysis approaches (14,37). Classification of image pixels often results in a ''salt and pepper'' effect because many pixels are below the minimum mapping unit (MMU) size of interest and which would inflate the number of patches, increase edge length, and have otherwise undesirable consequences on the patch metrics (38).…”

Section: Discussionmentioning

confidence: 99%

“…Remote-sensing imagery with higher spatial resolution (27) or multidate image delineation (14,41) has the ability to more precisely locate an edge in the field. Further research could focus on the detectability of different types of edges and their relationships to different processes, such as fragmentation.…”

Section: Discussionmentioning

confidence: 99%

“…Object-oriented methods for landscape classification and analyses have been found to exceed accuracies of traditional spectral classification methods that treat each pixel independently (14)(15)(16). Object-oriented image segmentation algorithms such as Fractal Net Evolution (17), available in the software eCognition (18), consider both spectral and contextual information to partition pixels into discrete image objects more representative of landscape pattern.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Characterizing Forest Fragments in Boreal, Temperate, and Tropical Ecosystems

Meddens

Hudak²,

Gould³

et al. 2008

AMBIO: A Journal of the Human Environment

View full text Add to dashboard Cite

An increased ability to analyze landscapes in a spatial manner through the use of remote sensing leads to improved capabilities for quantifying human-induced forest fragmentation. Developments of spatially explicit methods in landscape analyses are emerging. In this paper, the image delineation software program eCognition and the spatial pattern analysis program FRAG-STATS were used to quantify patterns of forest fragments on six landscapes across three different climatic regions characterized by different moisture regimes and different influences of human pressure. Our results support the idea that landscapes with higher road and population density are more fragmented; however, there are other, equally influential factors contributing to fragmentation, such as moisture regime, historic land use, and fire dynamics. Our method provided an objective means to characterize landscapes and assess patterns of forest fragments across different forested ecosystems by addressing the limitations of pixel-based classification and incorporating image objects.

show abstract

“…These studies mention LCC classification accuracies of more than eighty percent and emphasize the mutual benefits of the combination of VHR imagery and LiDAR. In addition, embedding elevation data into LCC change detection is also thought to improve the accuracy of segmentation and classification (Desclee et al, 2006). The forest cover in a landslide area can not only be expressed in areal loss, but can also be characterized by changes in AGB, by using tree specific allometric relations derived from multi-temporal LiDAR-based canopy height models (Muukkonen et al, 2005).…”

Section: Introductionmentioning

confidence: 99%

Object-based integrated landscape change analysis: synergy of multi-temporal LiDAR and very high resolution orthophotos

Kamps

Seijmonsbergen

Bouten

2016

GEOBIA 2016: Solutions and Synergies

View full text Add to dashboard Cite

ABSTRACT:Active landslides have three major effects on a landscape: 1. land cover change, 2. topographical change, and 3. above ground biomass change. Data derived from multi-temporal Light Detection and Ranging technology (LiDAR) is used in combination with multi-temporal orthophotos to quantify changes between 2006 and 2012, caused by a landslide near Doren in Austria. Data synergy is used to optimize accuracies of land cover change, and to improve results of topographical change analysis and aboveground biomass estimations. Topographical change is calculated using differencing of digital terrain models. The above ground biomass is quantified by applying a local-maximum algorithm for tree top detection, in combination with allometric equations. The land cover change classification accuracies were improved from 65% (using only LiDAR) and 76% (using only orthophotos) to 90% (using synergy) for 2006. A similar increase from respectively 64% and 75% to 91% was established for 2012. The results of the improved land cover classifications were used to optimize the topographical and above ground biomass change calculations. Fine-scale improvements of the classifications included forest edges and shadows, small open spots in the vegetation, and confusion between land cover classes. The enhanced accuracies of the land cover change analysis demonstrate the effectiveness and advantages of using synergy of LiDAR and orthophotos using OBIA. The method has great potential to be transferred to larger areas for use in monitoring, although data size calls for workflows to operate on cloud-based infrastructures that provide sufficient computational power.

show abstract

Forest change detection by statistical object-based method

Cited by 436 publications

References 44 publications

A new approach for land cover classification and change analysis: Integrating backdating and an object-based method

A new approach for land cover classification and change analysis: Integrating backdating and an object-based method

Characterizing Forest Fragments in Boreal, Temperate, and Tropical Ecosystems

Object-based integrated landscape change analysis: synergy of multi-temporal LiDAR and very high resolution orthophotos

Contact Info

Product

Resources

About