Detection of fallen trees in ALS point clouds using a Normalized Cut approach trained by simulation

Polewski, Przemyslaw; Yao, Wei; Heurich, Marco; Krzystek, Peter; Stilla, Uwe

doi:10.1016/j.isprsjprs.2015.01.010

Cited by 83 publications

(67 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Polewski et al used full-waveform LiDAR point clouds to merge short segments into whole fallen trees using Normalized Cut algorithm in a forest ecosystem with substantial canopy coverage, thereby achieving enhanced extraction. Their reported accuracy reached 80% [20]. This study also demonstrated that trunk diameter was the primary factor affecting extraction resolution.…”

Section: Introductionmentioning

confidence: 59%

A Novel Approach for Coarse-to-Fine Windthrown Tree Extraction Based on Unmanned Aerial Vehicle Images

2017

View full text Add to dashboard Cite

Surveys of windthrown trees, resulting from hurricanes and other types of natural disasters, are an important component of agricultural insurance, forestry statistics, and ecological monitoring. Aerial images are commonly used to determine the total area or number of downed trees, but conventional methods suffer from two primary issues: misclassification of windthrown trees due to the interference from other objects or artifacts, and poor extraction resolution when trunk diameters are small. The objective of this study is to develop a coarse-to-fine extraction technique for individual windthrown trees that reduces the effects of these common flaws. The developed method was tested using UAV imagery collected over rubber plantations on Hainan Island after the Nesat typhoon in China on 19 October 2011. First, a coarse extraction of the affected area was performed by analyzing the image spectrum and textural characteristics. A thinning algorithm was then used to simplify downed trees into skeletal structures. Finally, fine extraction of individual trees was achieved using a line detection algorithm. The completeness of windthrown trees in the study area was 75.7% and the correctness was 92.5%. While similar values have been reported in other studies, they often include constraints, such as tree height. This technique is proposed to be a more feasible extraction algorithm as it is capable of achieving low commission errors across a broad range of tree heights and sizes. As such, it is a viable option for extraction of windthrown trees with a small trunk diameter.

show abstract

Section: Introductionmentioning

confidence: 59%

A Novel Approach for Coarse-to-Fine Windthrown Tree Extraction Based on Unmanned Aerial Vehicle Images

2017

View full text Add to dashboard Cite

show abstract

“…We adapt the method of Polewski et al (2015) which is originally designed for fallen tree segmentation, to detect the standing stems of single trees from unstructured high density ALS point clouds. The main goal is to detect linear structures in the ALS 3D point clouds which are likely to represent single tree stems.…”

Section: Methodsmentioning

confidence: 99%

“…Point feature histograms (PFH): a local 3D shape descriptor of the neighborhood around the target point, based on the angular relationships between adjacent surface normals. It is useful for distinguishing between different types of surface classes based on their shape (plane, cylindrical, spherical, etc) (Rusu et al, 2008;Polewski et al, 2015).…”

Section: Point Levelmentioning

confidence: 99%

See 1 more Smart Citation

Detection of Single Tree Stems in Forested Areas From High Density Als Point Clouds Using 3d Shape Descriptors

Amiri¹,

Polewski²,

Yao³

et al. 2017

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

View full text Add to dashboard Cite

ABSTRACT:Airborne Laser Scanning (ALS) is a widespread method for forest mapping and management purposes. While common ALS techniques provide valuable information about the forest canopy and intermediate layers, the point density near the ground may be poor due to dense overstory conditions. The current study highlights a new method for detecting stems of single trees in 3D point clouds obtained from high density ALS with a density of 300 points/m 2 . Compared to standard ALS data, due to lower flight height (150-200 m) this elevated point density leads to more laser reflections from tree stems. In this work, we propose a three-tiered method which works on the point, segment and object levels. First, for each point we calculate the likelihood that it belongs to a tree stem, derived from the radiometric and geometric features of its neighboring points. In the next step, we construct short stem segments based on high-probability stem points, and classify the segments by considering the distribution of points around them as well as their spatial orientation, which encodes the prior knowledge that trees are mainly vertically aligned due to gravity. Finally, we apply hierarchical clustering on the positively classified segments to obtain point sets corresponding to single stems, and perform 1-based orthogonal distance regression to robustly fit lines through each stem point set. The 1-based method is less sensitive to outliers compared to the least square approaches. From the fitted lines, the planimetric tree positions can then be derived. Experiments were performed on two plots from the Hochficht forest in Oberösterreich region located in Austria. We marked a total of 196 reference stems in the point clouds of both plots by visual interpretation. The evaluation of the automatically detected stems showed a classification precision of 0.86 and 0.85, respectively for Plot 1 and 2, with recall values of 0.7 and 0.67.

show abstract

“…Such digital terrain models (DTMs) or digital surface models (DSMs) can be useful in monitoring FH indicators such as changes in tree height resulting from damage or deviations in growth due to stress [93][94][95]. Recent studies show that it is even possible to detect objects located on the ground surface such as coarse woody debris [96,97]. Coarse woody debris is an important indicator of FH, because it provides habitat to a multitude of endangered plant and animal species and plays an important role in the forest carbon cycle [98,99].…”

Section: Light Detection and Ranging (Lidar)mentioning

confidence: 99%

Understanding Forest Health with Remote Sensing-Part II—A Review of Approaches and Data Models

et al. 2017

Self Cite

View full text Add to dashboard Cite

Stress in forest ecosystems (FES) occurs as a result of land-use intensification, disturbances, resource limitations or unsustainable management, causing changes in forest health (FH) at various scales from the local to the global scale. Reactions to such stress depend on the phylogeny of forest species or communities and the characteristics of their impacting drivers and processes. There are many approaches to monitor indicators of FH using in-situ forest inventory and experimental studies, but they are generally limited to sample points or small areas, as well as being time-and labour-intensive. Long-term monitoring based on forest inventories provides valuable information about changes and trends of FH. However, abrupt short-term changes cannot sufficiently be assessed through in-situ forest inventories as they usually have repetition periods of multiple years. Furthermore, numerous FH indicators monitored in in-situ surveys are based on expert judgement. Remote sensing (RS) technologies offer means to monitor FH indicators in an effective, repetitive and comparative way. This paper reviews techniques that are currently used for monitoring, including close-range RS, airborne and satellite approaches. The implementation of optical, RADAR and LiDAR RS-techniques to assess spectral traits/spectral trait variations (ST/STV) is described in detail. We found that ST/STV can be used to record indicators of FH based on RS. Therefore, the ST/STV approach provides a framework to develop a standardized monitoring concept for FH indicators using RS techniques that is applicable to future monitoring programs. It is only through linking in-situ and RS approaches that we will be able to improve our understanding of the relationship between stressors, and the associated spectral responses in order to develop robust FH indicators.

show abstract

Detection of fallen trees in ALS point clouds using a Normalized Cut approach trained by simulation

Cited by 83 publications

References 32 publications

A Novel Approach for Coarse-to-Fine Windthrown Tree Extraction Based on Unmanned Aerial Vehicle Images

A Novel Approach for Coarse-to-Fine Windthrown Tree Extraction Based on Unmanned Aerial Vehicle Images

Detection of Single Tree Stems in Forested Areas From High Density Als Point Clouds Using 3d Shape Descriptors

Understanding Forest Health with Remote Sensing-Part II—A Review of Approaches and Data Models

Contact Info

Product

Resources

About