BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.staging.osgeo.org//foss4g-it-2023//talk//UGT3XN
BEGIN:VTIMEZONE
TZID:GMT
BEGIN:STANDARD
DTSTART:20001029T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:GMT
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T020000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:BST
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-foss4g-it-2023-UGT3XN@talks.staging.osgeo.org
DTSTART;TZID=GMT:20230612T153000
DTEND;TZID=GMT:20230612T154500
DESCRIPTION:Autores: Abderrahim Nemmaoui\, Fernando J. Aguilar\, Manuel A. 
 Aguilar\nForests act as important carbon sinks\, therefore being key compo
 nents of the global carbon cycle. The carbon dioxide emissions account is 
 essential for climate regulation policies and the evaluation of the effect
 s of these policies\, as well as for understanding the services they provi
 de to societies.\nTraditionally\, forest inventories are completed by grou
 nd-based expert crews. These field surveys are uneconomical\, time consumi
 ng and not adequate for studies dealing with periodic data collection. Con
 sequently\, one of the key topic in forest applications is to find an effe
 ctive method to produce effective and accurate inventories.\nIn recent yea
 rs\, Remote Sensing (RS) has proven to be capable of providing independent
 \, timely and reliable forest information. RS data are used to estimate se
 veral forest variables of silvicultural interest such as crown diameter (C
 D)\, tree height (H)\, diameter at breast height (DBH) and aboveground bio
 mass (AGB). In this sense\, and due to its ability to estimate attributes 
 at tree level\, LiDAR derive point cloud data has become a valuable data s
 ource in the field of efficient and accurate detection and segmentation of
  individual trees (IT).\nState-of-the-art approaches use different algorit
 hms for individual tree segmentation (ITS). For each algorithm\, a specifi
 c methodology to create the input Canopy Height Model (CHM) and/or many pa
 rameters should be tuned to somehow adapt the segmentation algorithm to ea
 ch particular forest stand. This approach makes the results highly depende
 nt on the applied local fitting parameters\, which implies difficulties wh
 en applied for large-scale mapping. In addition\, the parameter setting pr
 ocess is quite time consuming and requires learning and understanding the 
 meaning and role of each parameter.\nThe main goal of this work aims at de
 veloping a pipeline that requires minimal user interaction when working on
  large areas of Mediterranean forests. The expected results should facilit
 ate the production of broad-extend IT maps and extract the corresponding d
 endrometric parameters from low-density airborne laser scanning (ALS) data
  without spending time tuning algorithm parameters. \nThe study area was l
 ocated in Sierra de María-Los Vélez Natural Park (Almeria\, Spain). Up t
 o 38 reference square plots of 25 m side containing reforested stands of A
 leppo pine (Pinus halepensis Mill.) with variable density\, tree height an
 d presence of shrubs and low vegetation mainly represented by little holm 
 oak trees (Quercus ilex L.). This forest composition and structure make up
  a forest typology that is very representative of the Mediterranean forest
 s.\nThree open source raster-based (i.e.\, CHM-based) were tested to extra
 ct tree location and some dendrometric parameters such as tree H and CD. T
 he first algorithm is the method proposed by Dalponte & Coomes(2016) adapt
 ed and introduced in the package lidR (Roussel et al.2020). The second one
  is the algorithm developed by Silva et al.(2016)\, which is focused on th
 e way to better approximating the intersecting canopy of multiple trees af
 ter locating treetops by local maxima. The last algorithm tested is includ
 ed in the library Digital Forestry Toolbox (DFT). In addition\, the point 
 cloud-based algorithm proposed by Li et al.(2012) was also tested. \nFor e
 very algorithm tested\, we tried different parameters to find the best pip
 eline\, finally obtaining up to 4024 combinations of all tested algorithms
  for each experimental plot. For each setting\, tree detection accuracy wa
 s assessed by computing the detection rate\, and the commission and omissi
 on errors. Some statistics\, such as median\, RMSE and relative RMSE\, wer
 e also used to quantitatively assess the accuracy of tree H and CD estimat
 es over each reference plot.\nThe IT detection accuracy rates\, in terms o
 f precision\, recall\, and F1-score\, showed the successful performance of
  the pipeline proposed in this study. The algorithm proposed by Li et al.(
 2012) showed detection F1-score average values of 82.65% (using the same p
 arameter combination for the 38 experimental plots). However\, it failed i
 n delimiting the crown diameter (relative RMSE 57.06% and Pearson r of 0.5
 5). The method developed by Silva et al.(2016)\, when applied on a CHM gen
 erated with the point-to-raster algorithm and using a LM based on a variab
 le Tree Window Size (TWS)\, presented a similar F1-score for ITS (i.e.\, 8
 2.53%)\, but being most successful delimiting the crown (relative RMSE 22.
 21% and Pearson r of 0.68). Finally\, Dalponte & Coomes(2016) and DFT meth
 ods showed slightly worse results\, with average F1-scores of 80.41% and 7
 5.66%\, respectively.\nThe results obtained confirms the usefulness of low
 -density ALS data to both detect IT and estimate H and CD\, also underlini
 ng some key aspects regarding the choice of the correct method and paramet
 ers to perform single tree detection for Aleppo pine in large areas of Med
 iterranean forests.
DTSTAMP:20260518T033510Z
LOCATION:Sala Biblioteca @ PoliBa
SUMMARY:AN AUTOMATIC AND EFFECTIVE PIPELINE FOR INDIVIDUAL TREE DETECTION A
 ND SEGMENTATION USING LOW-DENSITY AIRBORNE LASER SCANNING DATA IN LARGE AR
 EAS OF MEDITERRANEAN FOREST - Abderrahim\, Fernando J. Aguilar
URL:https://talks.staging.osgeo.org/foss4g-it-2023/talk/UGT3XN/
END:VEVENT
END:VCALENDAR