Mobile QR Code QR CODE : Journal of the Korean Society of Civil Engineers

  1. ๊ฒฝ๋ถ๋Œ€ํ•™๊ต ์œต๋ณตํ•ฉ์‹œ์Šคํ…œ๊ณตํ•™๊ณผ ์„์‚ฌ๊ณผ์ • (Kyungpook National University ยท kanghs0829@knu.ac.kr)
  2. ์ข…์‹ ํšŒ์› ยท ๊ฒฝ๋ถ๋Œ€ํ•™๊ต ๋ฉ€ํ‹ฐ์Šค์ผ€์ผ ์œ ยท๋ฌด๊ธฐ ๊ตฌ์กฐ๋ฌผ ์ž์œจ์ง„๋‹จ๊ธฐ์ˆ ์—ฐ๊ตฌ์†Œ ์—ฐ๊ตฌ์ดˆ๋น™๊ต์ˆ˜ (Kyungpook National University ยท geolee@knu.ac.kr)
  3. ๊ฒฝ๋ถ๋Œ€ํ•™๊ต ์—๋„ˆ์ง€์œตํ•ฉ ๋ฐ ๊ธฐํ›„๋ณ€ํ™”ํ•™๊ณผ ๋ฐ•์‚ฌ๊ณผ์ • (Kyungpook National University ยท gusrlf6695@knu.ac.kr)
  4. ์ข…์‹ ํšŒ์› ยท ์„œ์šธ์—ฐ๊ตฌ์› AI ๋น…๋ฐ์ดํ„ฐ๋žฉ ์—ฐ๊ตฌ์œ„์› (The Seoul Institute ยท jungok@si.re.kr)
  5. ์ข…์‹ ํšŒ์› ยท ๊ต์‹ ์ €์ž ยท ๊ฒฝ๋ถ๋Œ€ํ•™๊ต ์œ„์น˜์ •๋ณด์‹œ์Šคํ…œํ•™๊ณผ ๊ต์ˆ˜ (Corresponding Author ยท Kyungpook National University ยท wlee33@knu.ac.kr)



์œ„์„ฑํ•ญ๋ฒ•์‹œ์Šคํ…œ, ๋ฌด์ธํ•ญ๊ณต๊ธฐ, ๋ผ์ด๋‹ค, ๋ฐ˜์‚ฌ ๊ฐ•๋„, ๋น„์ง€๋„ํ•™์Šต, ํ† ๊ณต๋Ÿ‰
Global navigation satellite system, Unmanned aerial vehicle, Light detection and ranging, Reflectance intensity, Unsupervised learning, Earthwork volume

1. ์„œ ๋ก 

๊ฑด์„ค ํ˜„์žฅ์—์„œ์˜ ์ •ํ™•ํ•œ ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •์€ ํ‰ํƒ„ํ•˜๊ณ  ์•ˆ์ •๋œ ์ง€๋ฐ˜์„ ํ™•๋ณดํ•˜๊ธฐ ์œ„ํ•œ ํ•„์ˆ˜ ๊ณผ์ •์œผ๋กœ ํšจ์œจ์ ์ธ ํ† ๋ชฉ ๋ฐ ๊ฑด์„ค ๊ณต์ • ์ง„ํ–‰์— ์ค‘์š”ํ•œ ์—ญํ• ์„ ํ•œ๋‹ค. ์ด๋ฅผ ์œ„ํ•ด ์ˆœ์ˆ˜ํ•œ ์ง€๋ฉด ๋ฐ์ดํ„ฐ๋ฅผ ํ™•๋ณดํ•˜๋Š” ๊ฒƒ์ด ํ•ต์‹ฌ์ด์ง€๋งŒ, ํ˜„์žฅ์˜ ์ง€ํ˜• ๊ธฐ๋ณต์ด๋‚˜ ๋‹ค์–‘ํ•œ ์žฅ์• ๋ฌผ๋กœ ์ธํ•ด ์ •ํ™•ํ•œ ๋ฐ์ดํ„ฐ ์ทจ๋“์ด ์–ด๋ ค์šด ๊ฒฝ์šฐ๊ฐ€ ๋งŽ๋‹ค. GNSS(Global Navigation Satellite System) ๊ธฐ๋ฐ˜ VRS(Virtual Reference Station) ์ธก๋Ÿ‰์€ ์ด๋™๊ตญ์˜ ์œ„์น˜ ๋ฐ์ดํ„ฐ๋ฅผ ์ฃผ๋ณ€ ๊ณ ์ • ๊ธฐ์ค€์ ์˜ ๋„คํŠธ์›Œํฌ๋ฅผ ํ™œ์šฉํ•ด ์‹ค์‹œ๊ฐ„์œผ๋กœ ๋ณด์ •ํ•˜์—ฌ ๋†’์€ ์ •ํ™•๋„์˜ ์œ„์น˜ ์ •๋ณด๋ฅผ ์ˆ˜์‹  ๋ฐ›์Œ์œผ๋กœ์จ ๋†’์€ ์ •ํ™•๋„๋ฅผ ์ œ๊ณตํ•˜๋Š” ์ „ํ†ต์ ์ธ ๋ฐฉ์‹์œผ๋กœ, ๊ธฐ์กด ํ† ๋ชฉ๊ณต์‚ฌ ๋ฐ ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ • ์—ฐ๊ตฌ์—์„œ ๋„๋ฆฌ ์‚ฌ์šฉ๋˜์–ด์™”๋‹ค(Lee and Lee, 2022). ๊ทธ๋Ÿฌ๋‚˜ ๋ณต์žกํ•œ ์ง€ํ˜•์ด๋‚˜ ๋„“์€ ๋ฉด์ ์„ ๋‹ค๋ฃฐ ๊ฒฝ์šฐ ๋‹ค์ˆ˜์˜ ์ขŒํ‘œ ๋ฐ์ดํ„ฐ๋ฅผ ์ˆ˜์ง‘ํ•ด์•ผ ํ•˜๋ฏ€๋กœ, ์‹œ๊ฐ„๊ณผ ์ธ๋ ฅ์ด ๋งŽ์ด ์†Œ๋ชจ๋œ๋‹ค๋Š” ๋‹จ์ ์ด ์žˆ๋‹ค. ํŠนํžˆ ์‹์ƒ์ด ์šธ์ฐฝํ•œ ์ง€์—ญ์—์„œ๋Š” GNSS ์ธก๋Ÿ‰ ์‹œ, ์žฅ์• ๋ฌผ์ด ๋งŽ์•„ ์ž‘์—… ์ธ๋ ฅ์˜ ์ด๋™์ด ์ œ์•ฝ์„ ๋ฐ›์œผ๋ฉฐ, ์ •ํ™•ํ•œ ๋ฐ์ดํ„ฐ ์ทจ๋“์ด ์–ด๋ ค์›Œ ์ž‘์—… ์‹œ๊ฐ„์ด ๊ธธ์–ด์ง€๊ณ  ๋น„์šฉ์ด ์ฆ๊ฐ€ํ•˜๊ฒŒ ๋œ๋‹ค. ์ด๋Ÿฌํ•œ ํ•œ๊ณ„๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•ด ์ตœ๊ทผ RTK(Real-Time Kinematic)-UAV (Unmanned Aerial Vehicle)์— ํƒ‘์žฌ๋œ LiDAR(Light Detection and Ranging) ์„ผ์„œ๊ฐ€ ์ฃผ๋ชฉ๋ฐ›๊ณ  ์žˆ๋‹ค(Park and Jung, 2021). LiDAR ์„ผ์„œ๋Š” ๋ณต์žกํ•œ ์ง€ํ˜•์—์„œ๋„ ๊ณ ํ•ด์ƒ๋„์˜ ์ ๊ตฐ ๋ฐ์ดํ„ฐ๋ฅผ ํšจ์œจ์ ์œผ๋กœ ์ทจ๋“ํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด์™€ ๋น„๋กฏํ•˜์—ฌ ๋ณต์žกํ•œ ํ™˜๊ฒฝ์—์„œ ๋ฐ์ดํ„ฐ๋ฅผ ์ฒ˜๋ฆฌํ•  ์ˆ˜ ์žˆ๋Š” ๋‹ค์–‘ํ•œ ์—ฐ๊ตฌ๊ฐ€ ์ด๋ฃจ์–ด์ ธ ๋†’์€ ์ ๋ฐ€๋„๋ฅผ ๊ธฐ๋ฐ˜์˜ ์šฐ์ˆ˜ํ•œ ์ •ํ™•๋„์˜ ๋ฐ์ดํ„ฐ ํ™•๋ณด๊ฐ€ ๊ฐ€๋Šฅํ•จ์ด ์ž…์ฆ๋˜์—ˆ๋‹ค(Park and Lee, 2021a). ํŠนํžˆ White et al.(2021)์€ ๋‹ค์–‘ํ•œ ํ•ญ๊ณต LiDAR ์„ผ์„œ๋ฅผ ํ™œ์šฉํ•˜์—ฌ ์‚ฐ๋ฆผ์ง€์—ญ์—์„œ ์ทจ๋“๋œ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ๊ตฌ์ถ•๋œ ์ˆ˜์น˜ํ‘œ๊ณ ๋ชจํ˜•(Digital Elevation Model, DEM)์˜ ์ •ํ™•๋„๋ฅผ ํ‰๊ฐ€ํ•˜๋ฉฐ, LiDAR ์„ผ์„œ๊ฐ€ ์šธ์ฐฝํ•œ ์‚ฐ๋ฆผํ™˜๊ฒฝ์—์„œ ๋ฐœ์ƒํ•˜๋Š” ๊ธฐ์กด RGB ์‚ฌ์ง„์ธก๋Ÿ‰ ๋ฐฉ์‹์˜ ํ•œ๊ณ„์ธ ์‹์ƒ ํ•˜๋ถ€ ๋ฐ์ดํ„ฐ์˜ ๋ฏธ์ˆ˜์ง‘ ๋ฌธ์ œ๋ฅผ ๊ทน๋ณตํ•  ๋Œ€์•ˆ์ž„์„ ์ž…์ฆํ•˜์˜€๋‹ค. ๋˜ํ•œ, LiDAR ์„ผ์„œ๊ฐ€ ์ œ๊ณตํ•˜๋Š” ๋ฐ˜์‚ฌ ๊ฐ•๋„(Reflectance Intensity) ์ •๋ณด๋Š” ๋ ˆ์ด์ € ์‹ ํ˜ธ์˜ ๋ฌผ๋ฆฌ์  ๋ฐ ๊ด‘ํ•™์  ํŠน์„ฑ์„ ๊ธฐ๋ฐ˜์œผ๋กœ ์‹์ƒ๊ณผ ์ง€๋ฉด์„ ๊ตฌ๋ถ„ํ•˜๋Š” ๋ฐ ํšจ๊ณผ์ ์œผ๋กœ ํ™œ์šฉ๋  ์ˆ˜ ์žˆ์Œ๋„ ์ฆ๋ช…๋˜์—ˆ๋‹ค(Scaioni et al., 2018). Park and Lee(2021b)์€ ์ง€์ƒ LiDAR ์„ผ์„œ๋ฅผ ํ†ตํ•ด ํ•จ๊ป˜ ์ทจ๋“๋˜๋Š” ์š”์†Œ์ธ RGB ์ •๋ณด๋ฅผ ํ™œ์šฉํ•˜๊ฑฐ๋‚˜, LiDAR ์„ผ์„œ๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ๊ตฌ์ถ•๋˜๋Š” ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ์˜ ํ˜•ํƒœํ•™์  ์ •๋ณด๋ฅผ ์ด์šฉํ•œ ๋‹ค์–‘ํ•œ ํ•„ํ„ฐ๋ง ๊ธฐ๋ฒ•์„ ์ ์šฉํ•˜์—ฌ ์‹์ƒ์ด ๋ฌด์„ฑํ•œ ํ•˜์ฒœ์˜ ์ œ๋ฐฉ์„ ๋Œ€์ƒ์œผ๋กœ ํšจ๊ณผ์ ์œผ๋กœ ์‹์ƒ๊ณผ ์ง€๋ฉด์„ ๊ตฌ๋ถ„ํ•˜์˜€๊ณ , ์ด๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์ •ํ™•๋„ ๋น„๊ตํ‰๊ฐ€๋ฅผ ์ง„ํ–‰ํ•˜์˜€๋‹ค. ์•„์šธ๋Ÿฌ, Yilmaz et al.(2016)์€ ํ•ญ๊ณต LiDAR ์„ผ์„œ๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ RGB ์ƒ‰์ƒ ํ•„ํ„ฐ๋ฅผ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ์— ์ ์šฉํ•˜์—ฌ ์‹์ƒ์˜ ๋†’์ด ์ถ”์ถœ ๋ฐ ์‚ฐ์ •ํ•˜์—ฌ ์ด๋ฅผ ์ •ํ™•๋„ ๋น„๊ตํ‰๊ฐ€์— ํ™œ์šฉํ•˜์˜€๋‹ค. ์ด์™€ ๋”๋ถˆ์–ด LiDAR ์„ผ์„œ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์ทจ๋“๋˜๋Š” ๋‹ค์–‘ํ•œ ์š”์†Œ๋ฅผ ์ ์ ˆํžˆ ํ™œ์šฉํ•˜๋ฉด ํ† ์ง€ ํ”ผ๋ณต์„ ํšจ๊ณผ์ ์œผ๋กœ ๋ถ„๋ฅ˜ํ•  ์ˆ˜ ์žˆ๋‹ค๋Š” ์ ์ด ์—ฌ๋Ÿฌ ์—ฐ๊ตฌ๋ฅผ ํ†ตํ•ด ์ž…์ฆ๋˜๊ณ  ์žˆ๋‹ค(Lang et al., 2020). ๋‹ค๋งŒ, LiDAR ์„ผ์„œ๋กœ ์ทจ๋“๋œ ๋ฐ์ดํ„ฐ๋ฅผ ํ†ตํ•ด ์ •ํ™•ํ•œ ์ง€๋ฉด ๋ฐ์ดํ„ฐ๋ฅผ ์–ป๊ธฐ ์œ„ํ•ด์„œ๋Š” ๋ฐ˜๋“œ์‹œ ํ›„์ฒ˜๋ฆฌ ๊ณผ์ •์ด ํ•„์š”ํ•˜๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” K-Means, K-Medoids, ๊ทธ๋ฆฌ๊ณ  DBSCAN(Density-Based Spatial Clustering of Application with Noise) ํด๋Ÿฌ์Šคํ„ฐ๋ง๊ณผ ๊ฐ™์€ ๋น„์ง€๋„ ํ•™์Šต ๊ธฐ๋ฒ•์„ ์ ์šฉํ•˜์—ฌ ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋ฅผ ๋ถ„์„ํ•˜๊ณ , ์‹์ƒ์ด ์ œ๊ฑฐ๋œ ์ˆœ์ˆ˜ํ•œ ์ง€๋ฉด ๋ฐ์ดํ„ฐ๋ฅผ ์ถ”์ถœํ•˜๊ณ ์ž ํ•œ๋‹ค. K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ๊ตฐ์ง‘ ์ˆ˜๋ฅผ ๋ฏธ๋ฆฌ ์„ค์ •ํ•ด ๋ช…ํ™•ํ•œ ๊ตฌ๋ถ„์ด ํ•„์š”ํ•œ ๋ฐ์ดํ„ฐ์— ์ ํ•ฉํ•˜๋ฉฐ, DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ๋ฐ€๋„๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ๋ถ„์„ํ•˜๋ฏ€๋กœ ๊ตฐ์ง‘ ํ˜•์„ฑ ์กฐ๊ฑด์„ ์ถฉ์กฑํ•˜๋Š” ๋ฐ์ดํ„ฐ๋งŒ์„ ํด๋Ÿฌ์Šคํ„ฐ๋งํ•˜์—ฌ ๋…ธ์ด์ฆˆ ๋ฐ์ดํ„ฐ์˜ ์˜ํ–ฅ์„ ๋œ ๋ฐ›์•„ ๋‹ค์–‘ํ•œ ๋ฐ€๋„๋ฅผ ๊ฐ€์ง„ ๋ฐ์ดํ„ฐ๋ฅผ ์ฒ˜๋ฆฌํ•˜๋Š”๋ฐ ์œ ๋ฆฌํ•˜๋‹ค(Deng, 2020). ํŠนํžˆ K-Medoids ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ํ‰๊ท ๊ฐ’์ด ์•„๋‹Œ ๋ฐ์ดํ„ฐ ๋‚ด์— ์กด์žฌํ•˜๋Š” ์ค‘๊ฐ„๊ฐ’์„ ๊ธฐ์ค€์œผ๋กœ ์ฒ˜๋ฆฌํ•˜์—ฌ, ์ด์ƒ์น˜์— ๋ฏผ๊ฐํ•˜์ง€ ์•Š์•„ ๋…ธ์ด์ฆˆ ๋ฐ์ดํ„ฐ์— ๊ฐ•์ธํ•˜๋‹ค๋Š” ํŠน์ง•์„ ์ง€๋‹Œ๋‹ค(Ushakov et al., 2021). ๋ณธ ์—ฐ๊ตฌ๋Š” ํ•„ํ„ฐ๋ง ๊ธฐ๋ฒ•์„ ํ†ตํ•ด ์ถ”์ถœํ•œ ์ง€๋ฉด ๋ฐ์ดํ„ฐ๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ GNSS ์ธก๋Ÿ‰ ๊ฒฐ๊ณผ์™€ ํ† ๊ณต๋Ÿ‰์„ ๋น„๊ตํ•˜๊ณ , ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์˜ ๋น„์œจ์„ ํ™œ์šฉํ•œ ๊ฐ€์ค‘ํ‰๊ท  ๋ฐฉ์‹์„ ํ†ตํ•ด ์ •ํ™•๋„๋ฅผ ํ‰๊ฐ€ํ•  ์˜ˆ์ •์ด๋‹ค. ์ด๋Ÿฌํ•œ ์ ‘๊ทผ๋ฒ•์€ ๊ณ„์ ˆ์  ํ•œ๊ณ„๋ฅผ ๊ทน๋ณตํ•˜๊ณ , ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •์˜ ํšจ์œจ์„ฑ์„ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋Š” ์ƒˆ๋กœ์šด ๋ฐฉ๋ฒ•๋ก ์„ ์ œ์‹œํ•  ๊ฒƒ์œผ๋กœ ๊ธฐ๋Œ€๋œ๋‹ค.

2. ์—ฐ๊ตฌ๋ฐฉ๋ฒ•

2.1 ์—ฐ๊ตฌ๋ฐฉ๋ฒ•

๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ๋ฅผ ์ด์šฉํ•˜์—ฌ ์—ฐ๊ตฌ๋Œ€์ƒ์ง€์— ๋Œ€ํ•ด ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ(Point Cloud) ๋ฐ์ดํ„ฐ๋ฅผ ์ทจ๋“ํ•˜์˜€์œผ๋ฉฐ, GNSS ์ธก๋Ÿ‰์„ ๋ฐ”ํƒ•์œผ๋กœ ๋Œ€์ƒ์ง€ ๊ฐ€์žฅ์ž๋ฆฌ์™€ ๋ชจ์„œ๋ฆฌ์— ๋Œ€ํ•˜์—ฌ 5๊ฐœ์˜ ์ง€์ƒ๊ธฐ์ค€์ (Ground Control Point, GCP)๊ณผ 3๊ฐœ์˜ ๊ฒ€์‚ฌ์ (Check Point, CP)์— ๋Œ€ํ•ด ์ด 8๊ฐœ์˜ ์ง€์ ์— ๋Œ€ํ•œ 3์ฐจ์› ์ขŒํ‘œ ๋ฐ์ดํ„ฐ๋ฅผ ์ทจ๋“ํ•˜์˜€๋‹ค. ์ทจ๋“๋œ LiDAR ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•ด ์ด์ƒ์น˜ ์ œ๊ฑฐ ๋ฐ GNSS ์ธก๋Ÿ‰์„ ๊ธฐ๋ฐ˜์œผ๋กœ ์ทจ๋“๋œ ์ง€์ƒ๊ธฐ์ค€์  ๋ฐ ๊ฒ€์‚ฌ์  ์ขŒํ‘œ์— ๋”ฐ๋ผ ์ขŒํ‘œ ์ •ํ•ฉ ์‹ค์‹œ ์ดํ›„ ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ ์šฉํ•˜์—ฌ ์‹์ƒ๊ณผ ์ง€๋ฉด์„ ๋ถ„๋ฆฌํ•˜์˜€๋‹ค. ์‹์ƒ์ด ๋ถ„๋ฆฌ๋˜์–ด ์–ป์€ ์ง€๋ฉด ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•ด ํ›„์ฒ˜๋ฆฌ ์ดํ›„ GIS(Geographic Information System)๋ฅผ ํ†ตํ•ด ์ง€ํ˜• ๊ตฌ์ถ• ๋ชจ๋ธ์„ ์ œ์ž‘ํ•˜์˜€๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ์„ธ ๊ฐ€์ง€ ๋น„์ง€๋„ํ•™์Šต ๊ธฐ๋ฒ• ์ ์šฉ์„ ๋ฐ”ํƒ•์œผ๋กœ ์‚ฐ์ •๋œ ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์„ ํ•ฉ์‚ฐํ•˜์—ฌ ์‚ฐ์ •๋œ ์ตœ์ข… ํ† ๊ณต๋Ÿ‰์„ ๋น„๊ต ๊ธฐ์ค€๊ฐ’์ธ GNSS ์ธก๋Ÿ‰ ์„ฑ๊ณผ๋กœ ์‚ฐ์ •๋œ ํ† ๊ณต๋Ÿ‰๊ณผ ์ •ํ™•๋„ ๋น„๊ต ๋ฐ ํ‰๊ฐ€ํ•˜์˜€๊ณ , ๋ณธ ์—ฐ๊ตฌ์˜ ์ „์ฒด ํ๋ฆ„๋„๋Š” Fig. 1๊ณผ ๊ฐ™๋‹ค.

Fig. 1. Flow of Study

../../Resources/KSCE/Ksce.2025.45.2.0265/fig1.png

2.2 ์—ฐ๊ตฌ๋Œ€์ƒ์ง€ ๋ฐ ์—ฐ๊ตฌ ์žฅ๋น„

๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ๊ฒฝ์ƒ๋ถ๋„ ์ƒ์ฃผ์‹œ ํ™”์„œ๋ฉด ์ƒ์šฉ๋ฆฌ 312์— ์œ„์น˜ํ•œ ์ƒ์šฉ ์ €์ˆ˜์ง€ ๊ตฌ์—ญ์„ ๋Œ€์ƒ์œผ๋กœ ํ•˜์˜€์œผ๋ฉฐ, ์ €์ˆ˜์ง€ ์˜์—ญ์ด ์•„๋‹Œ ๋ฐ˜๋Œ€ํŽธ ์‚ฌ๋ฉด์„ ์—ฐ๊ตฌ๋Œ€์ƒ์ง€๋กœ ์„ ์ •ํ•˜์˜€๋‹ค. ์ด ์ง€์—ญ์€ GNSS ์œ„์„ฑ ์‹ ํ˜ธ ์ˆ˜์‹ ์„ ๋ฐฉํ•ดํ•˜๋Š” ์š”์†Œ๊ฐ€ ์—†๊ณ , ๋ฌด์ธํ•ญ๊ณต๊ธฐ ๋น„ํ–‰์— ์ ํ•ฉํ•œ ํ™˜๊ฒฝ์„ ์ œ๊ณตํ•˜์—ฌ, ์—ฐ๊ตฌ ์ˆ˜ํ–‰์— ์ ํ•ฉํ•œ ์กฐ๊ฑด์„ ๊ฐ–์ถ”๊ณ  ์žˆ๋‹ค๊ณ  ํŒ๋‹จํ•˜์˜€๋‹ค. ์—ฐ๊ตฌ๋Š” ์‹์ƒ์ด ๊ฐ€์žฅ ํ™œ๋ฐœํ•œ ์‹œ๊ธฐ์ธ 6์›” ์ค‘์ˆœ์— ์ง„ํ–‰ํ•˜์˜€์œผ๋ฉฐ, ํ•ด๋‹น ์ง€์—ญ์€ ์งง์€ ์ดˆ๋ชฉ๊ณผ ๋‹ค์–‘ํ•œ ๊ด€๋ชฉ์ด ํ˜ผ์žฌ๋œ ๋ณตํ•ฉ ์ง€ํ˜•์ด๋‹ค(Fig. 2).

๋ณธ ์—ฐ๊ตฌ์— ์‚ฌ์šฉ๋œ ๋ฌด์ธํ•ญ๊ณต๊ธฐ๋Š” DJI์‚ฌ์˜ Matrice 300 RTK์ด๋ฉฐ, ๋Œ€์ƒ์ง€ ์ดฌ์˜์— ์‚ฌ์šฉ๋œ LiDAR ์„ผ์„œ๋Š” Matrice 300 RTK์™€ ํƒ‘์žฌ ๋ฐ ํ˜ธํ™˜์ด ๊ฐ€๋Šฅํ•œ DJI Zenmuse L1 ๋ชจ๋ธ์„ ์‚ฌ์šฉํ•˜์˜€๋‹ค. ๋˜ํ•œ, ์‹ค์‹œ๊ฐ„ ๋ฌด์ธํ•ญ๊ณต๊ธฐ ์œ„์น˜ ๋ฐ์ดํ„ฐ ๋ณด์ •์„ ์œ„ํ•ด DJI D-RTK2 Mobile Station์„ ์‚ฌ์šฉํ•˜์˜€๋‹ค. ์ง€์ƒ๊ธฐ์ค€์  ๋ฐ ๊ฒ€์‚ฌ์  ์ธก๋Ÿ‰์— ๋Œ€ํ•ด์„œ๋Š” GNSS ์ˆ˜์‹ ์žฅ์น˜์ธ Trimble์‚ฌ์˜ R8S ๋ชจ๋ธ์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ฐ์ดํ„ฐ๋ฅผ ์ทจ๋“ํ•˜์˜€๋‹ค. ๋ฐ์ดํ„ฐ ์ทจ๋“์— ํ™œ์šฉํ•œ ์ดฌ์˜ ์žฅ๋น„ ๋ฐ GNSS ์ˆ˜์‹ ์žฅ์น˜์— ๋Œ€ํ•œ ์ œ์›์€ Tables 1, 2์™€ ๊ฐ™๋‹ค.

Fig. 2. Study Area

../../Resources/KSCE/Ksce.2025.45.2.0265/fig2.png

Table 1. Specification of UAV, Sensor and DGNSS Station

Spec.

UAV

Spec.

Sensor

Spec.

DGNSS station

Matrice 300 RTK

Zenmuse L1

D-RTK 2 moblie station

Weight

6.3 kg

Weight

930ยฑ10 $g$

Satellite signal (GPS)

L1, C/A, L2, L5

Hovering accuracy (GPS)

Vertical : ยฑ0.1 $m$

Resolution

3840 ร—2160 Pixel

Positioning precision (single)

Vertical : 3.0 $m$

Horizontal : ยฑ0.1 $m$

File format

JPEG

Horizontal : 1.5 $m$

Flight time (Max)

โ‰ค 55 $\min$

F-stop

f/2.8~f/11

Positioning precision (RTK)

Vertical : 2 $cm$ + 1 $ppm$

Speed (Max)

โ‰ค 17 $m/s$

FOV

(Field Of View)

84ยฐ

Horizontal : 1 $cm$ + 1 $ppm$

RTK positioning accuracy

Vertical : 1.5 $cm$ + 1 $ppm$

Focal length

8.8 $mm$/24 $mm$

Size

168 ร— 168 ร— 1708 $mm$

Horizontal : 1 $cm$ + 1 $ppm$

Wavelength

905 $nm$

Memory volume

16 $GB$

ISO

100

Table 2. Specification of GNSS Data Receiver and Controller

Spec.

GNSS data receiver

Spec.

GNSS data controller

Trimble R8S

Juno T41/5

Channels

440 Channels

Operating

system

Android 4.1

โ€œJellyBeanโ€

Satellite signals(GPS)

L1C/A, L1C, L2C, L2E, L5

RAM

512 $MB$

Network-RTK precision

Horizontal

8 $mm$ + 0.5 $ppm$ RMS

Storage

8 $GB$

Vertical

15 $mm$ + 0.5 $ppm$ RMS

Processor

800 $MHZ$

3. ๋ฐ์ดํ„ฐ ์ทจ๋“

3.1 ๋ฐ์ดํ„ฐ ์ทจ๋“

๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ๊ธฐ์กด ์ทจ๋“๋œ GNSS ์ธก๋Ÿ‰ ์„ฑ๊ณผ์— ๋”ํ•ด RTK-UAV ๊ธฐ๋ฐ˜์˜ LiDAR ์„ผ์„œ๋ฅผ ํ™œ์šฉํ•˜์—ฌ ๋Œ€์ƒ์ง€์— ๋Œ€ํ•œ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ํ˜•์‹์˜ 3์ฐจ์› ์ขŒํ‘œ ๋ฐ์ดํ„ฐ๋ฅผ ์ทจ๋“ํ•˜์˜€๋‹ค. ๋ฌด์ธํ•ญ๊ณต๊ธฐ ์˜์ƒ ์ดฌ์˜์€ ๋ฐ”๋žŒ์œผ๋กœ ์ธํ•œ ์ขŒํ‘œ ์ทจ๋“ ์˜ค์ฐจ๊ฐ€ ์ตœ์†Œํ™”๋  ์ˆ˜ ์žˆ๋Š” ์‹œ์ ์„ ์„ ์ •ํ•˜์—ฌ ์ง„ํ–‰ํ•˜์˜€๋‹ค. ์ทจ๋“๋œ ์˜์ƒ์€ DJI์‚ฌ์˜ Terra ์†Œํ”„ํŠธ์›จ์–ด๋ฅผ ์‚ฌ์šฉํ•ด ํŒŒ์ผ ํ˜•์‹์„ GIS ์†Œํ”„ํŠธ์›จ์–ด์—์„œ ์ฒ˜๋ฆฌํ•˜๊ธฐ ์šฉ์ดํ•œ Las ํ˜•์‹์œผ๋กœ ๋ณ€ํ™˜ํ•˜์˜€์œผ๋ฉฐ, ๋˜ํ•œ ์—ฐ๊ตฌ๋Œ€์ƒ์ง€๊ฐ€ ์œ„์น˜ํ•œ ์ง€์—ญ ์ขŒํ‘œ๊ณ„์ธ TM(Transverse Mercator) ์ค‘๋ถ€์›์  ์ขŒํ‘œ๊ณ„์ธ Korea 2000/ Central Belt 2010(EPSG : 5186)๊ณผ GNSS ์ธก๋Ÿ‰ ๊ธฐ์ค€ ์ง€์˜ค์ด๋“œ ๋ชจ๋ธ๊ณผ ๊ฐ™์€ KN-Geoid 18(Korea National-Geoid Model 18)๋ฅผ ์ ์šฉํ•˜์˜€๋‹ค. ๋ฐ์ดํ„ฐ ๊ฐ„์˜ ์ƒํ˜ธ ๊ฑฐ๋ฆฌ ๋ฐ€๋„๋Š” Terra ์†Œํ”„ํŠธ์›จ์–ด ์ตœ์†Œ๊ฑฐ๋ฆฌ ๋ฐ€๋„ ์„ค์ • ํ•œ๊ณ„์ธ 5 cm๋กœ ์„ค์ •ํ•˜์˜€๋‹ค. ์•„์šธ๋Ÿฌ, ์‹ค์‹œ๊ฐ„ ํ•ญ๊ณต๊ธฐ ์œ„์น˜ ๋ณด์ •์„ ์œ„ํ•ด ํ˜„์žฅ ์„ค์น˜์‹ DGNSS(Differential GNSS) ์ˆ˜์‹ ๊ธฐ์ธ D-RTK 2 Mobile Station์„ ์„ค์น˜ํ•˜์—ฌ ์‹ค์‹œ๊ฐ„ ์œ„์น˜ ์˜ค์ฐจ ์ˆ˜์ •์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค.

3.2 ์ง€์ƒ๊ธฐ์ค€์  ๋ฐ ๊ฒ€์‚ฌ์  ์ธก๋Ÿ‰

๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” DGNSS Station์˜ ์‹ค์‹œ๊ฐ„ ์œ„์น˜ ์˜ค์ฐจ ๋ณด์ •๋ฟ๋งŒ ์•„๋‹ˆ๋ผ, ์ •ํ™•ํ•œ ์ขŒํ‘œ ๋ฐ์ดํ„ฐ์˜ ์ •ํ•ฉ์„ ์œ„ํ•ด ๋Œ€์ƒ์ง€์— ์ง€์ƒ๊ธฐ์ค€์ ๊ณผ ๊ฒ€์‚ฌ์ ์„ ์ง์ ‘ ์„ค์น˜ํ•˜์—ฌ ์ดฌ์˜์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค(Seong et al., 2018). RTK ๋ฐฉ์‹๊ณผ ๋”๋ถˆ์–ด ๊ฐ€์ƒ๊ธฐ์ค€์  ๊ฐ„์˜ ์ƒํ˜ธ ์‹œํ†ต์„ ๊ธฐ๋ฐ˜์œผ๋กœ ๋†’์€ ์ •ํ™•๋„๋ฅผ ์ œ๊ณตํ•˜์—ฌ ๋„๋ฆฌ ์‚ฌ์šฉ๋˜๋Š” ๋ฐฉ์‹์ธ VRS ์ธก๋Ÿ‰ ๋ฐฉ์‹์„ ์ฑ„ํƒํ•˜์˜€๋‹ค(Schaefer and Pearson, 2021). ๋Œ€์ƒ์ง€ ์ธก๋Ÿ‰์— ์‚ฌ์šฉ๋œ ์ง€์˜ค์ด๋“œ ๋ชจ๋ธ์€ KN-Geoid 18์ด๊ณ , ์‹ค์‹œ๊ฐ„ GPS ์ˆ˜์‹  ์œ„์„ฑ ์ˆ˜๋Š” 25๊ฐœ ์ด์ƒ์„ ์œ ์ง€ํ•˜์˜€์œผ๋ฉฐ, ์ˆ˜์ง ์˜ค์ฐจ๋Š” 14 mm, ์ˆ˜ํ‰ ์˜ค์ฐจ๋Š” 9 mm๋กœ ์ง‘๊ณ„๋˜์—ˆ๋‹ค. ์ง€์ƒ๊ธฐ์ค€์ ์€ ๋Œ€์ƒ์ง€ ๊ฐ€์žฅ์ž๋ฆฌ์— 5๊ฐœ, ๊ฒ€์‚ฌ์ ์€ ๋Œ€์ƒ์ง€ ์ค‘๊ฐ„ ์ง€์ ์— 3๊ฐœ๋ฅผ ๋ฐฐ์น˜ํ•˜์—ฌ ์ด 8๊ฐœ ์ ์— ๋Œ€ํ•ด GNSS ์ธก๋Ÿ‰์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค. ๋Œ€์ƒ์ง€๋Š” ์šธ์ฐฝํ•œ ์‹์ƒ๊ณผ ๋ถˆ๊ทœ์น™์ ์ธ ์ง€ํ˜• ๊ธฐ๋ณต์„ ํŠน์ง•์œผ๋กœ ํ•˜๋ฉฐ, GNSS ์ˆ˜์‹ ์ด ์›ํ™œํ•œ ์ง€๋ฉด ๊ตฌ์—ญ ์ค‘ ์‹์ƒ ๋ถ„ํฌ๊ฐ€ ๋น„๊ต์  ์ ์€ ๋Œ€์ƒ์ง€ ๊ฐ€์žฅ์ž๋ฆฌ์™€ ์ค‘๊ฐ„ ์ง€์ ์„ ์„ ์ •ํ•˜์—ฌ ์ง€์ƒ๊ธฐ์ค€์  ๋ฐ ๊ฒ€์‚ฌ์ ์„ ์„ค์น˜ํ•˜์˜€๋‹ค. ์ด๋Ÿฌํ•œ ์œ„์น˜๋Š” ์ •ํ™•ํ•œ ์ขŒํ‘œ ๋ฐ์ดํ„ฐ ์ˆ˜์ง‘์ด ๊ฐ€๋Šฅํ•˜๋‹ค๊ณ  ํŒ๋‹จ๋˜์–ด, ์ด๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์ธก๋Ÿ‰ ์ž‘์—…์„ ์ˆ˜ํ–‰ํ•˜์˜€๋‹ค. ๊ณต๊ณต์ธก๋Ÿ‰ ์ž‘์—… ๊ทœ์ •์— ๋”ฐ๋ผ ํ•ด๋‹น ์ง€์ ๋งˆ๋‹ค 10์ดˆ ์ด์ƒ ๊ด€์ธกํ•˜์˜€์œผ๋ฉฐ ๋Œ€์ƒ์ง€๊ฐ€ ์‹์ƒ์œผ๋กœ ์ธํ•œ ๊ธฐ๋ณต๊ณผ ์žฅ์• ๋ฌผ๋กœ ์ธํ•ด ์ธก๋Ÿ‰์„ ์ง„ํ–‰ํ•œ ์‹œ๊ฐ„์€ ์ด 20๋ถ„์ด ์†Œ์š”๋˜์—ˆ๋‹ค.

3.3 RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ ์ดฌ์˜

LiDAR ์„ผ์„œ๋Š” ์งง์€ ์ฃผ๊ธฐ์˜ ๋ ˆ์ด์ € ํŽ„์Šค๋ฅผ ๋ฐœ์‚ฌํ•˜์—ฌ ๋ฐ˜์‚ฌ ์‹ ํ˜ธ๋ฅผ ๊ฐ์ง€ํ•˜๊ณ , ๋ฐ˜์‚ฌ๋œ ์‹œ๊ฐ„์„ ๊ธฐ๋ฐ˜์œผ๋กœ ๋ฌผ์ฒด์™€์˜ ๊ฑฐ๋ฆฌ๋ฅผ ๊ณ„์‚ฐํ•œ๋‹ค. ์ด ๊ณผ์ •์—์„œ ๊ด‘์ž๊ฐ€ ๋ฌผ์ฒด์— ๋ฐ˜์‚ฌ๋˜๊ฑฐ๋‚˜ ํ‹ˆ์„ ํ†ตํ•ด ์ง€๋ฉด์œผ๋กœ ํˆฌ๊ณผ๋˜์–ด ๋‹ค์ค‘๋ฐ˜์‚ฌ๊ฐ€ ๋ฐœ์ƒํ•˜๋ฉฐ, ์ด๋ฅผ ํ†ตํ•ด ์ •๋ฐ€ํ•œ 3์ฐจ์› ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋ฐ์ดํ„ฐ๋ฅผ ์ƒ์„ฑํ•œ๋‹ค. LiDAR๋Š” ๊ธฐ์กด ์‚ฌ์ง„์ธก๋Ÿ‰ ๋ฐฉ์‹์— ๋น„ํ•ด ์ •ํ™•ํ•˜๊ณ  ์„ธ๋ฐ€ํ•œ ๋ฐ์ดํ„ฐ๋ฅผ ์ œ๊ณตํ•˜๋ฉฐ, ๋ณต์žกํ•œ ์ง€ํ˜•์—์„œ๋„ ์šฐ์ˆ˜ํ•œ ์ •๋ณด ์ˆ˜์ง‘ ๋Šฅ๋ ฅ์„ ๋ณด์—ฌ ๋‹ค์ˆ˜์˜ ๊ด€๋ จ ์—ฐ๊ตฌ๊ฐ€ ์ง„ํ–‰๋˜์—ˆ๋‹ค(Langhammer et al., 2018). RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ๋ฅผ ํ™œ์šฉํ•œ ์ธก๋Ÿ‰์€ ๋ฐ”๋žŒ์œผ๋กœ ์ธํ•œ ๊ธฐ์ฒด ํ”๋“ค๋ฆผ์„ ๋ฐฉ์ง€ํ•˜๊ธฐ ์œ„ํ•ด ํ’์† 1 m/s ์ดํ•˜ ์กฐ๊ฑด์—์„œ ์ˆ˜ํ–‰๋˜์—ˆ์œผ๋ฉฐ, ๋น„ํ–‰ ์ „ ์ž์„ธ ๋ณด์ • ๋ฐ ์™œ๊ณก ์ˆ˜์ •์„ ์œ„ํ•œ ์บ˜๋ฆฌ๋ธŒ๋ ˆ์ด์…˜ ๊ณผ์ •์„ ์™„๋ฃŒํ•˜์˜€๋‹ค. ์ •ํ™•ํ•œ ๊ณ ๋„ ๊ฐ’ ํ™•๋ณด๋ฅผ ์œ„ํ•ด ๋น„ํ–‰๊ณ ๋„๋Š” 50 m๋กœ ์„ค์ •ํ•˜์˜€๊ณ , ์ข…์ค‘๋ณต๋„์™€ ํšก์ค‘๋ณต๋„๋Š” ์‚ฌ์ง„์ธก๋Ÿ‰ ๊ถŒ์žฅ ๊ธฐ์ค€์ธ ๊ฐ๊ฐ 70 %, 80 %๋กœ ์„ค์ •ํ•˜์˜€๋‹ค. ์ดฌ์˜ ๋ฐ์ดํ„ฐ์˜ ์ •ํ•ฉ์„ฑ์„ ํ™•๋ณดํ•˜๊ธฐ ์œ„ํ•ด ๋น„ํ–‰์†๋„๋Š” 4 m/s๋กœ ์œ ์ง€ํ•˜์˜€์œผ๋ฉฐ, ์‚ฌ์ „์— ์„ค์ •ํ•œ ์ฝ”์Šค๋ฅผ ๋”ฐ๋ผ ์ž๋™๋น„ํ–‰ ์ธก๋Ÿ‰์„ ์ˆ˜ํ–‰ํ•˜์˜€๋‹ค. ๋Œ€์ƒ์ง€๋Š” ๊ฒฝ์‚ฌ๊ฐ์ด ์กด์žฌํ•˜๋Š” ์ง€ํ˜•์œผ๋กœ, ์„ธ๋ฐ€ํ•œ ๋ฐ์ดํ„ฐ ์ทจ๋“์„ ์œ„ํ•ด ์ฝ”์Šค๊ฐ์„ 240๋„๋กœ ์„ค์ •ํ•˜์—ฌ ๋Œ€๊ฐ์„  ๋ฐฉํ–ฅ ๊ฒฝ๋กœ๋กœ ๋น„ํ–‰์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค. ๋˜ํ•œ, LiDAR ์„ผ์„œ์™€ ์ง€๋ฉด ์š”์†Œ ๊ฐ„์˜ ์ž…์‚ฌ๊ฐ ์˜ํ–ฅ์„ ์ตœ์†Œํ™”ํ•˜๊ธฐ ์œ„ํ•ด ์นด๋ฉ”๋ผ ์„ผ์„œ๋ฅผ ์ง€๋ฉด์— ์ˆ˜์ง์œผ๋กœ ์„ค์ •ํ•˜์—ฌ ์ดฌ์˜์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค. ์ด 7๋ถ„๊ฐ„์˜ ๋น„ํ–‰์„ ํ†ตํ•ด ๋ฐ์ดํ„ฐ๋ฅผ ์ทจ๋“ํ•˜์˜€์œผ๋ฉฐ, Fig. 3์€ ๋ฌด์ธํ•ญ๊ณต๊ธฐ LiDAR ์ธก๋Ÿ‰์„ ํ†ตํ•ด ์ดฌ์˜๋œ ๋ฐ์ดํ„ฐ์˜ ์ทจ๋“์ฝ”์Šค์˜ ๊ฒฝ๋กœ๋ฅผ ๋ณด์—ฌ์ค€๋‹ค.

Fig. 3. Data Acquisition Course of RTK-UAV LiDAR Sensor

../../Resources/KSCE/Ksce.2025.45.2.0265/fig3.png

4. ์ขŒํ‘œ ์ •ํ•ฉ ๋ฐ ๋ฐ์ดํ„ฐ ์ „์ฒ˜๋ฆฌ

4.1 ๋ฐ์ดํ„ฐ ์ƒ˜ํ”Œ๋ง

LiDAR ์„ผ์„œ๋กœ ์ทจ๋“ํ•œ ๋ฐ์ดํ„ฐ๋Š” ๋†’์€ ์ ๋ฐ€๋„์˜ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ๋กœ ๊ตฌ์„ฑ๋˜์–ด ์žˆ์ง€๋งŒ, ๋ถ„์„ํ•˜๋ ค๊ณ  ํ•˜๋Š” ๋Œ€์ƒ ์ด์™ธ์˜ ๋ถˆํ•„์š”ํ•œ ๋ฐ์ดํ„ฐ๋„ ํฌํ•จํ•˜๊ณ  ์žˆ์–ด ๋ฐ์ดํ„ฐ ์ฒ˜๋ฆฌ ์‹œ๊ฐ„์˜ ์ฆ๊ฐ€์™€ ์ฒ˜๋ฆฌ ์†๋„๊ฐ€ ์ €ํ•˜๋˜์–ด ๊ฒฐ๊ณผ์ ์œผ๋กœ ๋ฐ์ดํ„ฐ ์ฒ˜๋ฆฌ ํšจ์œจ์„ฑ์ด ๋–จ์–ด์ง€๋Š” ๊ฒฐ๊ณผ๋ฅผ ์œ ๋ฐœํ•˜๊ฒŒ ๋œ๋‹ค. ์ด๋Ÿฌํ•œ ๋น„ํšจ์œจ์ ์ธ ์ธก๋ฉด์„ ๊ฐœ์„ ํ•˜๊ธฐ ์œ„ํ•ด ๋ฐ์ดํ„ฐ ์ƒ˜ํ”Œ๋ง ๊ธฐ๋ฒ•์ด ์‚ฌ์šฉ๋˜๋ฉฐ, ์ด๋Š” ์ „์ฒด ๋ฐ์ดํ„ฐ ์„ธํŠธ์—์„œ ๋Œ€ํ‘œ์ ์ธ ๋ถ€๋ถ„ ์ง‘ํ•ฉ์„ ์ถ”์ถœํ•จ์œผ๋กœ์จ ๋ฐ์ดํ„ฐ ๋ถ„์„ ์†๋„์™€ ์ฒ˜๋ฆฌ ํšจ์œจ์„ ํ–ฅ์ƒ์‹œํ‚ค๋Š” ๋ฐ ํ™œ์šฉ๋œ๋‹ค(Guo et al., 2010). ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” Cloud Compare ์†Œํ”„ํŠธ์›จ์–ด์˜ Subsampling ๋„๊ตฌ๋ฅผ ํ™œ์šฉํ•˜์˜€์œผ๋ฉฐ, ํฌ์ธํŠธ ๊ฐ„ ๊ฑฐ๋ฆฌ ๊ฐ„๊ฒฉ 0.01 m๋กœ ์„ค์ •ํ•˜์—ฌ ๋ฐ์ดํ„ฐ ์ƒ˜ํ”Œ๋ง์„ ์ˆ˜ํ–‰ํ•˜์˜€๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ์ทจ๋“๋œ ๋Œ€์ƒ์ง€ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ์—์„œ ๋ถ„์„์— ํ•„์š”ํ•œ ์ฃผ์š” ํŠน์ง•์„ ๋‚˜ํƒ€๋‚ด๋Š” ํฌ์ธํŠธ ์œ„์ฃผ๋กœ ์ƒ˜ํ”Œ๋งํ•จ์œผ๋กœ์จ ๋ถˆํ•„์š”ํ•œ ๋ฐ์ดํ„ฐ๋ฅผ ์ œ๊ฑฐํ•˜๊ณ , ๋ฉ”๋ชจ๋ฆฌ ์‚ฌ์šฉ๋Ÿ‰ ์ ˆ๊ฐ๊ณผ ๋Œ€ํ‘œ์„ฑ ํ™•๋ณด๋ฅผ ํ†ตํ•ด ๋ฐ์ดํ„ฐ ๋ถ„์„ ํšจ์œจ์„ฑ์˜ ํ–ฅ์ƒ์„ ๋‹ฌ์„ฑํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค.

4.2 ์ง€์ƒ๊ธฐ์ค€์  ๋ฐ ๊ฒ€์‚ฌ์  ์ขŒํ‘œ ์ •ํ•ฉ

๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ๋ฌด์ธํ•ญ๊ณต๊ธฐ์˜ ์‹ค์‹œ๊ฐ„ ์ž์„ธ ๋ณด์ • ๋ฐ ์œ„์น˜ ์˜ค์ฐจ ์ˆ˜์ •์„ ์œ„ํ•ด ํ˜„์žฅ์— ์ด๋™์‹ GNSS ์ˆ˜์‹ ๊ธฐ๋ฅผ ์„ค์น˜ํ•˜์—ฌ RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ๋ฅผ ํ†ตํ•ด ๋Œ€์ƒ์ง€์˜ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋ฐ์ดํ„ฐ๋ฅผ ์ทจ๋“ํ•˜์˜€๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ์ด๋™์‹ GNSS ์ˆ˜์‹ ๊ธฐ๋งŒ์œผ๋กœ๋Š” 100 % ์ •ํ™•๋„์— ๊ทผ์ ‘ํ•˜๋Š” ๋ฐ์ดํ„ฐ๋ฅผ ์ทจ๋“ํ•˜๋Š” ๊ฒƒ์€ ์–ด๋ ค์šด ์‹ค์ •์ด๋ฉฐ, ์‹ค์‹œ๊ฐ„ ์™œ๊ณก ๋ณด์ •์€ ๊ฐ€๋Šฅํ•˜์ง€๋งŒ, RTK ์ˆ˜์‹  ํŠน์„ฑ์ƒ ๊ฑฐ๋ฆฌ๊ฐ€ ๋ฉ€์–ด์งˆ์ˆ˜๋ก ์˜ค์ฐจ์œจ์ด ์ฆ๊ฐ€ํ•œ๋‹ค๋Š” ํ•œ๊ณ„๊ฐ€ ์กด์žฌํ•œ๋‹ค. ์ด์— ๋”ฐ๋ผ ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ์ •ํ™•ํ•œ ์ขŒํ‘œ ๋ณด์ •์„ ์œ„ํ•ด ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋ฐ์ดํ„ฐ ํ›„์ฒ˜๋ฆฌ ํ”„๋กœ๊ทธ๋žจ์ธ Cloud Compare ์†Œํ”„ํŠธ์›จ์–ด๋ฅผ ํ†ตํ•ด GNSS ์ธก๋Ÿ‰์œผ๋กœ ์ทจ๋“ํ•œ ์ขŒํ‘œ์™€ RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ๋ฅผ ํ†ตํ•ด ์ทจ๋“๋œ ์ง€์ƒ๊ธฐ์ค€์ ๊ณผ ๊ฒ€์‚ฌ์  ์ขŒํ‘œ์— ๋Œ€ํ•˜์—ฌ ๊ฐ๊ฐ ์ขŒํ‘œ ์ •๋ฐ€ ์ •ํ•ฉ์„ ํ•˜์˜€๋‹ค(Fig. 4). RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์ทจ๋“๋œ ์ขŒํ‘œ์™€ GNSS ์ธก๋Ÿ‰์œผ๋กœ ์ทจ๋“๋œ ์ขŒํ‘œ ๊ฐ„์˜ ์ตœ์ข… ์‚ฐ์ถœ ํ‰๊ท ์ œ๊ณฑ๊ทผ์˜ค์ฐจ(Root Mean Square Error, RMSE)๋Š” 0.42 m๋กœ ์‚ฐ์ถœ๋˜์—ˆ๋‹ค. ์‚ฐ์ถœ๋œ RMSE๋Š” ์ •ํ•ฉ๋œ ์ขŒํ‘œ๋“ค์˜ ํ‰๊ท ์ ์ธ ์˜ค์ฐจ๋ฅผ ๋‚˜ํƒ€๋‚ด๋ฉฐ, ๋ชจ๋“  ํฌ์ธํŠธ๊ฐ€ ๋™์ผํ•œ ์˜ค์ฐจ๋ฅผ ๊ฐ€์ง€๋Š” ๊ฒƒ์€ ์•„๋‹ˆ๋‹ค. ์ด์— ๋”ฐ๋ผ ์ผ๋ถ€ ํฌ์ธํŠธ์—์„œ ํŠน์ • ์˜ค์ฐจ๊ฐ€ ๋‚˜ํƒ€๋‚˜๋Š” ๊ฒƒ์„ ๋ฐฐ์ œํ•  ์ˆ˜ ์—†์—ˆ๋‹ค. ๋˜ํ•œ, ์—ฐ๊ตฌ๋Œ€์ƒ์ง€์— ์„ค์น˜ํ•œ ์ง€์ƒ๊ธฐ์ค€์  ๋ฐ ๊ฒ€์‚ฌ์ ์— ๋Œ€ํ•ด RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ๋กœ ์ธก๋Ÿ‰ํ•œ ๊ฒฐ๊ณผ์™€ ๊ธฐ์ค€๊ฐ’์ธ GNSS ์ธก๋Ÿ‰์œผ๋กœ ์ง์ ‘ ์ทจ๋“ํ•œ ์„ฑ๊ณผ๋ฅผ ๋น„๊ตํ•œ ๊ฒฐ๊ณผ๋Š” Table 3๊ณผ ๊ฐ™๋‹ค.

Fig. 4. GCP/CP Location of Study Area

../../Resources/KSCE/Ksce.2025.45.2.0265/fig4.png

Table 3. Maximum Error between GNSS Survey and RTK-UAV LiDAR (Unit: m)

RTK-UAV LiDAR

GNSS-VRS

Maximum error

X

Y

Z

X

Y

Z

283044.13

428480.53

266.06

283044.75

428481.03

266.10

0.60 m

283023.37

428525.13

265.88

283023.27

428525.33

266.03

0.25 m

282992.38

428588.41

266.02

282992.60

428588.92

266.11

0.43 m

282979.85

428583.85

261.32

282979.68

428583.93

261.28

0.17 m

282975.19

428568.50

256.46

282975.24

428568.48

256.542

0.31 m

283011.65

428513.00

260.50

283011.28

428512.63

260.08

0.73 m

282974.53

428545.44

252.44

282974.23

428545.83

252.39

0.33 m

282990.85

428510.68

252.28

282990.70

428510.65

252.17

0.18 m

4.3 ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋ฐ์ดํ„ฐ ์ „์ฒ˜๋ฆฌ

๋ฌด์ธํ•ญ๊ณต๊ธฐ LiDAR ์„ผ์„œ๋ฅผ ํ†ตํ•ด ์ทจ๋“๋œ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋ฐ์ดํ„ฐ๋Š” ์‚ฌ๋žŒ, ์ƒˆ, ์ž๋™์ฐจ ๋“ฑ ์›€์ง์ด๋Š” ๋Œ€์ƒ๋ฌผ๋กœ ์ธํ•ด ๋‹ค์–‘ํ•œ ๋…ธ์ด์ฆˆ๊ฐ€ ํฌํ•จ๋  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด๋ฅผ ์ถ”๊ฐ€์ ์ธ ์ž‘์—… ์—†์ด ํ™œ์šฉํ•  ๊ฒฝ์šฐ ์˜ค์ฐจ๊ฐ€ ๋ฐœ์ƒํ•  ๊ฐ€๋Šฅ์„ฑ์ด ์žˆ๋‹ค. ์ด๋Ÿฌํ•œ ๋…ธ์ด์ฆˆ ๋ฐ์ดํ„ฐ๋ฅผ ์ œ๊ฑฐํ•˜๊ธฐ ์œ„ํ•ด ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ํ›„์ฒ˜๋ฆฌ ์†Œํ”„ํŠธ์›จ์–ด์ธ Cloud Compare๋ฅผ ํ™œ์šฉํ•˜์—ฌ, ์ทจ๋“๋œ ์˜์ƒ์— ๋Œ€ํ•œ ์ „์ฒ˜๋ฆฌ ๊ณผ์ •์„ ์ˆ˜ํ–‰ํ•ด ์ฃผ์—ˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” Noise filter ๋ฐ SOR filter๋ฅผ ์ˆœ์ฐจ์ ์œผ๋กœ ์ ์šฉํ•˜์˜€๋‹ค. Noise filter๋Š” k-NN(k-Nearest Neighbors) ์•Œ๊ณ ๋ฆฌ์ฆ˜ ๊ธฐ๋ฐ˜์œผ๋กœ ํ•œ ์ด์ƒ์น˜ ์ œ๊ฑฐ ๊ธฐ๋ฒ•์œผ๋กœ, ์ง€์—ญ ๋ฐ€๋„๋ฅผ ๋ถ„์„ํ•˜์—ฌ ๋ฐ€๋„๊ฐ€ ๋‚ฎ์€ ์˜์—ญ์€ ์ž๋™์œผ๋กœ ๋…ธ์ด์ฆˆ๋กœ ๊ฐ์ง€ํ•˜๊ณ  ์ œ๊ฑฐํ•˜๋ฉฐ, SOR(Statistical Outlier Remover) filter๋Š” ํ†ต๊ณ„์  ๋ฐฉ๋ฒ•์„ ์ด์šฉํ•˜์—ฌ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋‚ด ์ด์ƒ์น˜๋ฅผ ์ œ๊ฑฐํ•˜๋Š” ํ•„ํ„ฐ๋ง ๊ธฐ๋ฒ•์œผ๋กœ ๊ฐ ํฌ์ธํŠธ๋“ค๊ณผ ์ด์›ƒ ํฌ์ธํŠธ๋“ค ๊ฐ„์˜ ํ‰๊ท  ๊ฑฐ๋ฆฌ๋ฅผ ๊ณ„์‚ฐํ•˜์—ฌ ํŠน์ • ์ž„๊ณ„๊ฐ’์„ ์ดˆ๊ณผํ•˜๋Š” ํฌ์ธํŠธ๋ฅผ ๋…ธ์ด์ฆˆ๋กœ ๊ฐ„์ฃผํ•˜๊ณ  ์ œ๊ฑฐํ•œ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ๋Œ€์ƒ์ง€ ์‚ฌ๋ฉด์—์„œ ํŠน์ • ์ž„๊ณ„ ๋ฒ”์œ„๋ฅผ ์ดˆ๊ณผํ•˜๋Š” ๋…ธ์ด์ฆˆ๋ฅผ ์ œ๊ฑฐํ•˜์˜€๊ณ , ๊ฒฐ๊ณผ์ ์œผ๋กœ ์ด์ƒ์น˜๊ฐ€ ๊ฐ์†Œ๋œ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋ฐ์ดํ„ฐ๋ฅผ ํ™•๋ณดํ•˜์—ฌ ๋ฐ์ดํ„ฐ ๊ฐ€๋…์„ฑ์„ ํ–ฅ์ƒ์‹œํ‚ค๊ณ  ๋ถˆํ•„์š”ํ•œ ๋ฐ์ดํ„ฐ๋ฅผ ์‚ญ์ œํ•จ์œผ๋กœ์จ ์ „๋ฐ˜์ ์ธ ์ •ํ™•๋„๋ฅผ ๋†’์˜€๋‹ค(Fig. 5).

Fig. 5. Result of Preprocessed LiDAR Point Cloud

../../Resources/KSCE/Ksce.2025.45.2.0265/fig5.png

5. ๋น„์ง€๋„ ํ•™์Šต ๊ธฐ๋ฒ• ์ ์šฉ

5.1 LiDAR Reflectance Intensity

๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ์˜ ์ทจ๋“ ์š”์†Œ ์ค‘ ํ•˜๋‚˜์ธ ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋ฅผ ํ™œ์šฉํ•˜์—ฌ ์‹์ƒ์„ ๋ถ„๋ฆฌํ•˜๊ณ  ์ˆœ์ˆ˜ํ•œ ์ง€๋ฉด ์ •๋ณด๋ฅผ ์ถ”์ถœํ•˜์˜€๋‹ค. ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋ž€ ๋ ˆ์ด์ € ํŽ„์Šค์˜ ์†ก์ˆ˜์‹ ์„ ํ†ตํ•ด ์–ป์„ ์ˆ˜ ์žˆ๋Š” ์š”์†Œ๋กœ ๋ ˆ์ด์ €๊ฐ€ ํ‘œ๋ฉด์—์„œ ๋ฐ˜์‚ฌ๋  ๋•Œ ์„ผ์„œ๊ฐ€ ์ˆ˜์‹ ํ•˜๋Š” ๋ฐ˜์‚ฌ๋œ ๋น›์˜ ์„ธ๊ธฐ๋ฅผ ๋‚˜ํƒ€๋‚ด๋ฉฐ 0~255 ๋ฒ”์œ„์— ์กด์žฌํ•˜๋Š” ๊ฐ’์„ ๊ฐ€์ง„๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ LiDAR ์„ผ์„œ๋กœ ์ฑ„ํƒํ•œ DJI Zenmuse L1 ๋ชจ๋ธ์€ ๊ทผ์ ์™ธ์„ (Near Infrared, NIR) ๋Œ€์—ญ์— ํ•ด๋‹นํ•˜๋Š” 905 nm ํŒŒ์žฅ์„ ์‚ฌ์šฉํ•˜๋Š” ๋ ˆ์ด์ €๋ฅผ ํƒ‘์žฌํ•˜๊ณ  ์žˆ๋‹ค. ์‹๋ฌผ์˜ ์—ฝ๋ก์†Œ๊ฐ€ ๊ทผ์ ์™ธ์„  ๋Œ€์—ญ์˜ ํŒŒ์žฅ์—์„œ ๋†’์€ ๋ฐ˜์‚ฌ์œจ์„ ๋ณด์ด๊ธฐ ๋•Œ๋ฌธ์— ์ด๋Ÿฌํ•œ ํŠน์ง•์„ ํ™œ์šฉํ•˜๋ฉด ํšจ์œจ์ ์œผ๋กœ ์‹์ƒ ํƒ์ง€ ๋ฐ ๊ฑด๊ฐ• ์ƒํƒœ ํ‰๊ฐ€ ๋“ฑ์„ ํ•  ์ˆ˜ ์žˆ๋‹ค(Gong et al., 2022). ๋˜ํ•œ, 905 nm ๋Œ€์—ญ์˜ ๋ ˆ์ด์ €๋Š” ์žฅ๊ฑฐ๋ฆฌ ๊ณ ๋ฐ€๋„ ์ธก์ • ๋ฐ ๋Œ€๊ธฐ ์‚ฐ๋ž€์— ๋Œ€ํ•œ ์ €ํ•ญ์„ฑ์ด ๊ฐ•ํ•ด, ์ง€ํ˜• ๋ฐ ๊ตฌ์กฐ๋ฌผ์˜ ๊ณ ๋„์™€ ๊ฑฐ๋ฆฌ ์ •๋ณด๋ฅผ ์ •ํ™•ํ•˜๊ณ  ์•ˆ์ •์ ์œผ๋กœ ์ทจ๋“ํ•  ์ˆ˜ ์žˆ๋‹ค.

5.2 ๋น„์ง€๋„ํ•™์Šต ๊ธฐ๋ฐ˜ ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ ํ•„ํ„ฐ๋ง

LiDAR ์„ผ์„œ๋Š” ๋ ˆ์ด์ €๋ฅผ ๋ฐœ์‚ฌํ•˜์—ฌ ๋ฐ˜์‚ฌ๋œ ์‹ ํ˜ธ๋ฅผ ๋ถ„์„ํ•˜๋Š” ๋ฐฉ์‹์œผ๋กœ ์ž‘๋™ํ•˜๋ฉฐ, ๋ฌผ์ฒด์˜ ๊ด‘ํ•™์  ๋ฐ˜์‚ฌํŠน์„ฑ, ์„ผ์„œ์™€ ๋ฌผ์ฒด ๊ฐ„์˜ ์ž…์‚ฌ๊ฐ, ํ‘œ๋ฉด์ƒํƒœ ๋“ฑ์ด ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ ํ’ˆ์งˆ์— ์˜ํ–ฅ์„ ๋ฏธ์นœ๋‹ค. ๋น›์„ ํก์ˆ˜ํ•˜๊ฑฐ๋‚˜ ์‚ฐ๋ž€์ด ์ ์€ ์ง€๋ฉด์€ ๊ด‘ํ•™์  ๋ฐ˜์‚ฌ ๊ฐ•๋„๊ฐ€ ๋‚ฎ๊ฒŒ ์ธก์ •๋˜๋Š” ๋ฐ˜๋ฉด, ๋ณต์žกํ•˜๊ณ  ๋ถˆ๊ทœ์น™ํ•œ ํ‘œ๋ฉด์„ ๊ฐ€์ง„ ์‹์ƒ์€ ๋ ˆ์ด์ € ์‹ ํ˜ธ๊ฐ€ ์—ฌ๋Ÿฌ ๋ฐฉํ–ฅ์œผ๋กœ ์‚ฐ๋ž€๋˜๋ฉด์„œ ๊ด‘ํ•™์ ์œผ๋กœ ์šฐ์ˆ˜ํ•œ ๊ฒฐ๊ณผ๋ฅผ ๋‚ด์–ด ๋†’์€ ๋ฐ˜์‚ฌ๊ฐ•๋„๋ฅผ ๋‚˜ํƒ€๋‚ธ๋‹ค(Zhao et al., 2020). ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋Š” ๋ฌผ์ฒด์˜ ํ‘œ๋ฉด ํŠน์„ฑ์— ๋”ฐ๋ผ ๋‹ค๋ฅด๊ฒŒ ์ธก์ •๋˜๋ฉฐ, ํŠน์ • ๊ตฌ๊ฐ„์— ๋ฐ€๋„๊ฐ€ ์ง‘์ค‘์ ์œผ๋กœ ํ˜•์„ฑ๋˜๋Š” ํŠน์„ฑ์„ ๋ณด์—ฌ์ค€๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ ์ฑ„ํƒํ•œ ๋น„์ง€๋„ํ•™์Šต ๊ธฐ๋ฒ• ์ค‘ ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์€ ์ด๋Ÿฌํ•œ ๋ฐ์ดํ„ฐ๋ฅผ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ๊ทธ๋ฃนํ™”ํ•˜๋Š” ๋ฐ ์ ํ•ฉํ•˜๋ฉฐ, ๋ฌผ์ฒด์˜ ํ‘œ๋ฉด ํŠน์„ฑ์— ๋”ฐ๋ผ ๋‹ค๋ฅด๊ฒŒ ์ˆ˜์ง‘๋œ ๋ฐ€๋„ ๊ธฐ๋ฐ˜์˜ ๋ฐ์ดํ„ฐ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์‹์ƒ๊ณผ ์ง€๋ฉด์„ ์ž๋™์œผ๋กœ ๊ทธ๋ฃนํ™”ํ•  ์ˆ˜ ์žˆ๋‹ค. ๋˜ํ•œ, ๋น„์ง€๋„ํ•™์Šต ๊ธฐ๋ฒ•์€ ์‚ฌ์ „์— ํด๋ž˜์Šค ์ •๋ณด๊ฐ€ ์ •์˜๋˜์ง€ ์•Š์€ ์ƒํƒœ์—์„œ๋„ ๋ฐ์ดํ„ฐ์˜ ํŠน์„ฑ์„ ์ •๊ตํ•˜๊ฒŒ ๋ฐ˜์˜ํ•˜๋ฉฐ, ์ž”์กด ๋…ธ์ด์ฆˆ ๋ฐ์ดํ„ฐ๋ฅผ ํšจ๊ณผ์ ์œผ๋กœ ํ•„ํ„ฐ๋งํ•  ์ˆ˜ ์žˆ๋Š” ์žฅ์ ์ด ์žˆ๋‹ค. ๋‹จ์ˆœ Thresholding ๋ฐฉ์‹๊ณผ ๋‹ฌ๋ฆฌ, ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์€ ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๊ฐ’์˜ ๋ถ„ํฌ์™€ ๋ฐ€๋„๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ๊ทธ๋ฃนํ™”๋ฅผ ์ˆ˜ํ–‰ํ•˜์—ฌ, ๋ณด๋‹ค ์œ ์˜๋ฏธํ•œ ๊ฒฐ๊ณผ๋ฅผ ๋„์ถœํ•  ์ˆ˜ ์žˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” K-Means, DBSCAN, K-Medoids ๋“ฑ์˜ ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์„ ๋น„๊ต ๋ถ„์„ํ•˜์˜€์œผ๋ฉฐ, ์ด๋ฅผ ํ†ตํ•ด ๋‹จ์ผ ๋ฐด๋“œ ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋ฅผ ํ™œ์šฉํ•œ ๋ถ„๋ฅ˜์˜ ํšจ์šฉ์„ฑ๊ณผ ๊ฐ€๋Šฅ์„ฑ์„ ํ™•์ธํ•˜์˜€๋‹ค.

5.3 K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง

K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์€ ๋ฐ์ดํ„ฐ๋ฅผ ์‚ฌ์ „ ์ง€์ •๋œ ํด๋Ÿฌ์Šคํ„ฐ ์ˆ˜(K)๋กœ ๊ตฐ์ง‘ํ™”ํ•˜๋Š” ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์œผ๋กœ, ๊ฐ ๋ฐ์ดํ„ฐ ํฌ์ธํŠธ๋ฅผ ๊ฐ€์žฅ ๊ฐ€๊นŒ์šด ์ค‘์‹ฌ์ ์— ํ• ๋‹นํ•œ ํ›„, ์ค‘์‹ฌ์ ์œผ๋กœ ๋ฐ˜๋ณต ์ˆ˜์ •ํ•˜์—ฌ ๋ฐ์ดํ„ฐ ํฌ์ธํŠธ๋“ค์ด ์„œ๋กœ ์œ ์‚ฌํ•œ ๊ฐ’์œผ๋กœ ๋ฌถ์ด๋„๋ก ํ•œ๋‹ค. K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ์ดˆ๊ธฐ ์ค‘์‹ฌ์ (Centroid)์„ ์‚ฌ์ „์— ์„ค์ •ํ•œ ๋’ค ๊ฐ ๋ฐ์ดํ„ฐ ํฌ์ธํŠธ๋ฅผ ๊ฐ€์žฅ ๊ฐ€๊นŒ์šด ์ค‘์‹ฌ์ ์— ํ• ๋‹นํ•˜๊ณ , ํด๋Ÿฌ์Šคํ„ฐ ๋‚ด ๋ฐ์ดํ„ฐ์˜ ํ‰๊ท ๊ฐ’์„ ์ค‘์‹ฌ์ ์œผ๋กœ ๋ฐ˜๋ณต ๊ฐฑ์‹ ํ•˜๋ฉด์„œ ๋ฐ์ดํ„ฐ ํฌ์ธํŠธ๋“ค์ด ์„œ๋กœ ์œ ์‚ฌํ•œ ๊ฐ’์œผ๋กœ ๋ฌถ์ด๋„๋ก ํ•œ๋‹ค(Ghazal, 2021). ์ด๋Ÿฌํ•œ ๊ณผ์ •์„ ์ˆ˜๋ ดํ•  ๋•Œ๊นŒ์ง€ ๋ฐ˜๋ณตํ•จ์œผ๋กœ์จ ์ตœ์ ์˜ ํด๋Ÿฌ์Šคํ„ฐ๋ฅผ ๋„์ถœํ•œ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š”, LiDAR ์„ผ์„œ ์ทจ๋“ ์š”์†Œ ์ค‘ ํ•˜๋‚˜์ธ ๋ฐ˜์‚ฌ ๊ฐ•๋„๋ฅผ ๊ธฐ์ค€์œผ๋กœ ํด๋Ÿฌ์Šคํ„ฐ๋ง์„ ์ˆ˜ํ–‰ํ•˜์˜€์œผ๋ฉฐ, ์ง€ํ‘œ๋ฉด์—๋Š” ์‹์ƒ๊ณผ ์ง€๋ฉด๋ฟ๋งŒ ์•„๋‹ˆ๋ผ ๋‹ค์–‘ํ•œ ์š”์†Œ๊ฐ€ ์กด์žฌํ•  ์ˆ˜ ์žˆ๊ธฐ ๋•Œ๋ฌธ์— ํด๋Ÿฌ์Šคํ„ฐ ์ˆ˜๋Š” 5๊ฐœ๋กœ ์„ค์ •ํ•˜์˜€๋‹ค. ํด๋Ÿฌ์Šคํ„ฐ๋ง ์ˆ˜ํ–‰ ๊ฒฐ๊ณผ, 46.33, 57.94, 31.98, 70.22, 88.04์˜ 5๊ฐœ ํด๋Ÿฌ์Šคํ„ฐ๋กœ ๊ตฐ์ง‘ํ™”ํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค(Fig. 6). K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง์˜ ๊ฒฝ์šฐ ์‚ฌ์ „์— ์„ค์ •๋œ ํด๋Ÿฌ์Šคํ„ฐ ์ˆ˜์— ๋”ฐ๋ผ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ ๋‚ด ๋ชจ๋“  ๋ฐ์ดํ„ฐ๋ฅผ ํฌํ•จํ•˜์—ฌ ์ฒ˜๋ฆฌํ•˜๋ฏ€๋กœ, ์ž…๋ ฅ ๋ฐ์ดํ„ฐ์™€ ์ถœ๋ ฅ ๊ฒฐ๊ณผ์˜ ํฌ์ธํŠธ ๊ฐœ์ˆ˜๊ฐ€ ๋™์ผํ•˜๊ฒŒ ์œ ์ง€๋œ๋‹ค. LiDAR ์„ผ์„œ์—์„œ ๋ฐœ์‚ฌ๋˜๋Š” ๋ ˆ์ด์ € ํŽ„์Šค๋Š” ๊ธฐ์กด ์ง€๋ฉด๋ฟ๋งŒ ์•„๋‹ˆ๋ผ ์‹์ƒ์˜ ์ž‘์€ ํ‹ˆ์„ ํŒŒ๊ณ  ๋“ค์–ด๊ฐ€ ์‹์ƒ์˜ ํ•˜๋ถ€์— ์กด์žฌํ•˜๋Š” ์ง€๋ฉด ์ •๋ณด๊นŒ์ง€ ์ทจ๋“ํ•  ์ˆ˜ ์žˆ๋‹ค. ์ด๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์ง€๋ฉด์— ํ•ด๋‹นํ•˜๋Š” ๋ฐ์ดํ„ฐ ๋ถ„ํฌ๊ฐ€ ๊ฐ€์žฅ ๋งŽ์ด ์ˆ˜์ง‘๋˜๊ฒŒ ๋œ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ๊ฐ€์žฅ ๋†’์€ ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ ๋ฐ€์ง‘๋„์˜ ํ‰๊ท  ์ˆ˜์น˜์ธ 46.33์œผ๋กœ ๋ถ„๋ฅ˜๋œ ํด๋Ÿฌ์Šคํ„ฐ ๊ตฐ์ง‘์— ๋”ฐ๋ผ ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๊ธฐ๋ฐ˜ ํ•„ํ„ฐ๋ง์„ ์ˆ˜ํ–‰ํ•˜์˜€๋‹ค.

Fig. 6. Histogram of K-Means Clustering Result by Intensity

../../Resources/KSCE/Ksce.2025.45.2.0265/fig6.png

5.4 K-Medoids ํด๋Ÿฌ์Šคํ„ฐ๋ง

K-Medoids ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ๋Œ€ํ‘œ์ ์ธ ๋น„์ง€๋„ํ•™์Šต ๊ธฐ๋ฒ• ์ค‘ ํ•˜๋‚˜๋กœ, K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•๊ณผ ์œ ์‚ฌํ•˜์ง€๋งŒ, K-Means๊ฐ€ ํ‰๊ท ๊ฐ’์„ ๊ธฐ์ค€์œผ๋กœ ํด๋Ÿฌ์Šคํ„ฐ๋งํ•˜๋Š” ๊ฒƒ๊ณผ ๋‹ฌ๋ฆฌ ๊ฐ ํด๋Ÿฌ์Šคํ„ฐ์˜ ์ค‘์‹ฌ์„ ๋ฐ์ดํ„ฐ ํฌ์ธํŠธ์™€ ํด๋Ÿฌ์Šคํ„ฐ ์ค‘์‹ฌ ๊ฐ„์˜ ์ด ๊ฑฐ๋ฆฌ๋ฅผ ์ตœ์†Œํ™”ํ•˜๋„๋ก ์ค‘์‹ฌ์ ์„ ๋ฐ˜๋ณต์ ์œผ๋กœ ์—…๋ฐ์ดํŠธํ•˜๋ฉด์„œ ๋™์ž‘ํ•œ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด, ํด๋Ÿฌ์Šคํ„ฐ ๋‚ด ๋ฐ์ดํ„ฐ ๊ฐ„์˜ ๊ฑฐ๋ฆฌ์™€ ์œ ์‚ฌ๋„๋ฅผ ์ง€์†์ ์œผ๋กœ ์ค„์—ฌ ์ตœ์ ์˜ ์ค‘์‹ฌ์ ์„ ์ฐพ๋Š” ๋ฐฉ์‹์ด๋‹ค. ํŠนํžˆ, ํ‰๊ท ๊ฐ’ ๋Œ€์‹  ์‹ค์ œ ๋ฐ์ดํ„ฐ ํฌ์ธํŠธ๋ฅผ ์ค‘์‹ฌ์œผ๋กœ ์‚ฌ์šฉํ•˜๊ธฐ ๋•Œ๋ฌธ์— ์ด์ƒ์น˜์™€ ๋…ธ์ด์ฆˆ์— ๊ฐ•๊ฑดํ•˜๋‹ค๋Š” ํŠน์ง•์ด ์žˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง์„ ์ˆ˜ํ–‰ํ•œ ๊ณผ์ •๊ณผ ๊ฐ™์ด ๊ธฐ์กด K-Medoids ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์—์„œ ๋ณดํŽธ์ ์œผ๋กœ ์‚ฌ์šฉ๋˜๋Š” ๋ฐฉ์‹์ธ 5๊ฐœ์˜ ํด๋Ÿฌ์Šคํ„ฐ๋ฅผ ์„ค์ •ํ•˜์—ฌ ์ง„ํ–‰ํ•˜์˜€๊ณ , ์ˆ˜ํ–‰ ๊ฒฐ๊ณผ 43.10, 53.72, 58.49, 30.29, 77.17์˜ ์ˆ˜์น˜๋ฅผ ๊ฐ€์ง€๋Š” ๋ฐ˜์‚ฌ๊ฐ•๋„ ํด๋Ÿฌ์Šคํ„ฐ ๊ตฐ์ง‘์œผ๋กœ ๋ถ„๋ฅ˜ํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค(Fig. 7). K-Medoids ํด๋Ÿฌ์Šคํ„ฐ๋ง์˜ ๊ฒฝ์šฐ ํด๋Ÿฌ์Šคํ„ฐ์˜ ์ค‘์‹ฌ์„ ์‹ค์ œ ๋ฐ์ดํ„ฐ ํฌ์ธํŠธ ์ค‘์—์„œ ์„ ์ •ํ•˜๊ณ , ํด๋Ÿฌ์Šคํ„ฐ ๋‚ด ์ด ๊ฑฐ๋ฆฌ๋ฅผ ์ตœ์ ํ™”ํ•˜๋Š” ๊ณผ์ •์„ ๋ฐ˜๋ณต์ ์œผ๋กœ ์ˆ˜ํ–‰ํ•œ๋‹ค. ์ด ๊ณผ์ •์—์„œ ๊ฒฝ๊ณ„์— ์œ„์น˜ํ•œ ์ผ๋ถ€ ํฌ์ธํŠธ๊ฐ€ ํŠน์ • ํด๋Ÿฌ์Šคํ„ฐ์— ์†ํ•˜์ง€ ์•Š์•„ ๊ตฐ์ง‘ ๋Œ€์ƒ์—์„œ ์ œ์™ธ๋  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด๋กœ ์ธํ•ด, K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฐ˜์˜ ๋ถ„๋ฅ˜ ๊ฒฐ๊ณผ์™€ ๋น„๊ตํ•  ๋•Œ, ์ถœ๋ ฅ๋˜๋Š” ํฌ์ธํŠธ ์ˆ˜๊ฐ€ ๋‹ค๋ฅด๊ฒŒ ๋‚˜ํƒ€๋‚œ๋‹ค. ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ ๋ถ„ํฌ๊ฐ€ ์ฃผ๋กœ ํด๋Ÿฌ์Šคํ„ฐ 0๊ณผ ํด๋Ÿฌ์Šคํ„ฐ 2๊ตฌ๊ฐ„์— ์ฃผ๋กœ ๋ถ„ํฌํ•˜์˜€์œผ๋ฉฐ, ๊ฐ€์žฅ ๋งŽ์€ ๋ถ„ํฌ๋ฅผ ๋ณด์ด๋Š” ํด๋Ÿฌ์Šคํ„ฐ 0(43.10)์— ๋”ฐ๋ผ ๋ถ„๋ฅ˜๋ฅผ ์ˆ˜ํ–‰ํ•˜์˜€๋‹ค.

Fig. 7. Histogram of K-Medoids Clustering Result by Intensity

../../Resources/KSCE/Ksce.2025.45.2.0265/fig7.png

5.5 DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง

DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ๋ฐ์ดํ„ฐ์˜ ๋ฐ€๋„๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ํด๋Ÿฌ์Šคํ„ฐ๋ฅผ ํ˜•์„ฑํ•˜๋ฉฐ, ๋ฐ€์ง‘๋œ ํฌ์ธํŠธ๋“ค์€ ํด๋Ÿฌ์Šคํ„ฐ๋กœ ๋ฌถ๊ณ , ๋ฐ€๋„๊ฐ€ ๋‚ฎ์€ ํฌ์ธํŠธ๋Š” ๋…ธ์ด์ฆˆ๋กœ ์ธ์‹ ๋ฐ ์ฒ˜๋ฆฌํ•˜๋Š” ์•Œ๊ณ ๋ฆฌ์ฆ˜์ด๋‹ค. ๋ณธ ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ํด๋Ÿฌ์Šคํ„ฐ ๊ฐœ์ˆ˜๋ฅผ ๋ฏธ๋ฆฌ ์ง€์ •ํ•  ํ•„์š” ์—†์ด, ๋‘ ํฌ์ธํŠธ ๊ฐ„์˜ ๊ฑฐ๋ฆฌ์™€ ๋ฐ€๋„๋ฅผ ๊ธฐ์ค€์œผ๋กœ ํด๋Ÿฌ์Šคํ„ฐ๋ฅผ ์ž๋™์œผ๋กœ ๊ฒฐ์ •ํ•˜๋ฉฐ, ๋ฐ€๋„ ์กฐ๊ฑด์„ ์ถฉ์กฑํ•˜์ง€ ๋ชปํ•˜๋Š” ํฌ์ธํŠธ๋Š” ๋…ธ์ด์ฆˆ๋กœ ํŒ๋‹จํ•˜์—ฌ ํด๋Ÿฌ์Šคํ„ฐ์— ํฌํ•จ์‹œํ‚ค์ง€ ์•Š๊ณ  ๊ฒฝ๊ณ„ ๊ตฌ๋ถ„์„ ํ™•์‹คํ•˜๊ฒŒ ํ•  ์ˆ˜ ์žˆ์–ด, ๋ณต์žกํ•œ ๋ฐ์ดํ„ฐ ๋ถ„ํฌ์—์„œ ์œ ์—ฐํ•˜๊ฒŒ ์ž‘๋™ํ•œ๋‹ค. DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ K-Means ๋ฐ K-Medoids ํด๋Ÿฌ์Šคํ„ฐ๋ง๊ณผ ๋‹ฌ๋ฆฌ, ๋ฐ€๋„ ์กฐ๊ฑด์„ ์ถฉ์กฑํ•˜์ง€ ๋ชปํ•˜๋Š” ํฌ์ธํŠธ๋ฅผ ๋…ธ์ด์ฆˆ๋กœ ๊ฐ„์ฃผํ•˜์—ฌ ์ˆ˜์ง‘๋Œ€์ƒ์—์„œ ๋ฐฐ์ œํ•˜๊ณ  ์ž๋™์œผ๋กœ ๊ตฐ์ง‘์—์„œ ์ œ์™ธํ•œ๋‹ค. ์ด๋กœ ์ธํ•ด ์•ž์„  ๋‘ ๊ธฐ๋ฒ•์— ๋น„ํ•ด ์ถœ๋ ฅ๋˜๋Š” ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ์˜ ๊ฐœ์ˆ˜๊ฐ€ ๋Œ€ํญ ๊ฐ์†Œ๋˜์–ด ์ƒ๋‹นํžˆ ์ ๊ฒŒ ์‚ฐ์ถœ๋œ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ์ง€ํ˜•์˜ ํŠน์„ฑ์— ๋”ฐ๋ผ ์ผ์ • ๊ตฌ๊ฐ„์— ํŠน์ง•์ ์œผ๋กœ ๋ถ„ํฌํ•˜๊ณ  ์žˆ๋Š” ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋ฅผ ๋Œ€์ƒ์œผ๋กœ ํ•˜์˜€๊ธฐ ๋•Œ๋ฌธ์— ํ™˜๊ฒฝ์— ๋”ฐ๋ฅธ ๋‹ค์–‘ํ•œ ์—ฌ๋Ÿฌ ์„ค์ •๊ฐ’์„ ๋น„๊ตํ•ด๋ณด์•˜์œผ๋ฉฐ, ๊ฒฐ๋ก ์ ์œผ๋กœ ํด๋Ÿฌ์Šคํ„ฐ๋ง ๋ณ€์ˆ˜์ด์ž ๋ฐ์ดํ„ฐ ์ด์›ƒ ์ตœ๋Œ€๊ฑฐ๋ฆฌ(Epsilon)๋ฅผ 3์œผ๋กœ ์„ค์ •ํ•˜๊ณ , ๋ณต์žกํ•œ ์ง€ํ˜• ๊ตฌ์กฐ์™€ LiDAR ์„ผ์„œ์˜ ๋‹ค์ค‘๋ฐ˜์‚ฌ ํŠน์„ฑ์„ ๋ฐ˜์˜ํ•˜์—ฌ ์ตœ์†Œ ์ด์›ƒ ํฌ์ธํŠธ(Min_samples) ๋ณ€์ˆ˜๋ฅผ 42๋กœ ์„ค์ •ํ•˜์—ฌ ๋ถ„๋ฅ˜๋ฅผ ์ง„ํ–‰ํ•˜์˜€๋‹ค(Fig. 8). ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ฒฐ๊ณผ 3๊ฐœ์˜ ํด๋Ÿฌ์Šคํ„ฐ ๊ตฐ์ง‘์ด ํ˜•์„ฑ๋˜์—ˆ๊ณ , Cluster 0์€ 164,304์˜ ํฌ์ธํŠธ๋กœ ๊ฐ€์žฅ ๋งŽ์€ ๋ฐ์ดํ„ฐ๊ฐ€ ์ˆ˜์ง‘๋˜์–ด, ์ฃผ๋กœ ์ง€๋ฉด์œผ๋กœ ์ด๋ฃจ์–ด์ง„ ๊ฒƒ์œผ๋กœ ์ถ”์ •๋œ๋‹ค. Cluster 1์€ 2,250๊ฐœ์˜ ํฌ์ธํŠธ๊ฐ€ ์ˆ˜์ง‘๋˜์–ด ๊ณ ๋„ ์ฐจ์ด๊ฐ€ ์กด์žฌํ•˜๋Š” ์‹์ƒ์ด ๋ถ„๋ฅ˜๋œ ๊ฒƒ์œผ๋กœ ์ถ”์ •๋˜๋ฉฐ, ์‚ฌ์ „์— Cluster -1์€ ์‚ฌ์ „์— ๋…ธ์ด์ฆˆ ๋ฐ์ดํ„ฐ๋กœ ์„ค์ •ํ•˜์˜€๊ณ , ๋ฐ€๋„๊ฐ€ ๋‚ฎ์•„ ๊ตฐ์ง‘์กฐ๊ฑด์„ ์ถฉ์กฑํ•˜์ง€ ๋ชปํ•˜์—ฌ ์ด๋กœ ์ธํ•ด ์ง€๋ฉด๊ณผ ์‹์ƒ์œผ๋กœ ํŒ๋‹จ๋˜๋Š” ๋‘ ํด๋Ÿฌ์Šคํ„ฐ์— ๋ชจ๋‘ ํฌํ•จ๋˜์ง€ ์•Š์•„ ํฌ์ธํŠธ ์ˆ˜์ง‘๋Œ€์ƒ์—์„œ ์ œ์™ธ๋˜์—ˆ๋‹ค. ๋‹ค์Œ์€ ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ํ†ตํ•ด ์ตœ์ข…์ ์œผ๋กœ ์‹์ƒ๊ณผ ์ง€๋ฉด์„ ํ•„ํ„ฐ๋งํ•œ ํฌ์ธํŠธ ํด๋ผ์šฐ๋“œ๋ฅผ ๋‚˜ํƒ€๋‚ธ๋‹ค(Fig. 9).

Fig. 8. Histogram of DBSCAN Clustering Result by Intensity

../../Resources/KSCE/Ksce.2025.45.2.0265/fig8.png

Fig. 9. Point Cloud of Unsupervised Learning-based Model Filtering. (a) K-Means, (b) K-Medoids, (c) DBSCAN

../../Resources/KSCE/Ksce.2025.45.2.0265/fig9.png

6. ๋ฐ์ดํ„ฐ ๋ณด๊ฐ„ ๋ฐ ๋ณ€ํ™˜

6.1 ํฌ๋ฆฌ๊น… ๋ณด๊ฐ„๋ฒ•(Kriging Interpolation)

ํฌ๋ฆฌ๊น… ๋ณด๊ฐ„๋ฒ•์€ ํ†ต๊ณ„์  ๊ธฐ๋ฒ•์„ ์ด์šฉํ•˜์—ฌ ๊ณต๊ฐ„์ ์œผ๋กœ ๋ถ„ํฌ๋œ ๋ฐ์ดํ„ฐ ๊ฐ’์„ ์˜ˆ์ธกํ•˜๋Š” ์ง€๋ฆฌ๋ฐ์ดํ„ฐ ๋ถ„์„๊ธฐ๋ฒ•์ด๋‹ค. ์ด ๊ธฐ๋ฒ•์€ ์ฃผ๋ณ€ ์ธก์ •๊ฐ’๋“ค์˜ ๊ณต๊ฐ„์  ์ƒ๊ด€์„ฑ์„ ๊ณ ๋ คํ•˜์—ฌ ๋ฏธ์ธก์ • ์ง€์ ์˜ ๊ฐ’์„ ์˜ˆ์ธกํ•˜๋ฉฐ, ํŠนํžˆ ์ง€๋ฆฌ์  ๋ฐ์ดํ„ฐ๋ฅผ ๋ถ„์„ํ•˜๋Š” ๊ฒฝ์šฐ ๋ณดํŽธ์ ์œผ๋กœ ์‚ฌ์šฉ๋œ๋‹ค. ์ด ๊ธฐ๋ฒ•์€ ์˜ˆ์ธก๊ฐ’๋ฟ๋งŒ ์•„๋‹ˆ๋ผ ์˜ˆ์ธก ์˜ค์ฐจ๊นŒ์ง€ ๊ณ„์‚ฐํ•˜์—ฌ ์˜ˆ์ธก ์‹ ๋ขฐ์„ฑ๊นŒ์ง€ ๋„์ถœ ๊ฐ€๋Šฅํ•˜๋‹ค๋Š” ์žฅ์ ์ด ์žˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” Cloud Compare ์†Œํ”„ํŠธ์›จ์–ด์— ๋‚ด์žฅ๋œ Rasterization - Kriging Interpolation ๋„๊ตฌ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ, ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ํ†ตํ•ด ์ œ๊ฑฐ๋œ ์‹์ƒ์„ ๊ธฐ๋ฐ˜์œผ๋กœ ๋ถ„๋ฆฌ๋œ ์ˆœ์ˆ˜ํ•œ ์ง€๋ฉด ๋ฐ์ดํ„ฐ์— ์กด์žฌํ•˜๋Š” ๋ฏธ์ธก์ • ๊ตฌ์—ญ์— ํ•ด๋‹นํ•˜๋Š” ์ขŒํ‘œ๋ฅผ ๋ณด๊ฐ„ํ•˜์—ฌ ๊ธฐ์กด ์ง€ํ˜•์˜ 3์ฐจ์› ์ขŒํ‘œ ๋ฐ์ดํ„ฐ๋ฅผ ์˜ˆ์ธกํ•˜์˜€๋‹ค. ๋‹ค์Œ์€ ํฌ๋ฆฌ๊น… ๋ณด๊ฐ„๋ฒ•์„ ํ†ตํ•ด ๋ฏธ์ธก์ • ์ง€ํ˜• ์ขŒํ‘œ ๋ฐ์ดํ„ฐ๊ฐ€ ์˜ˆ์ธก ๋ฐ ๋ณด๊ฐ„๋œ ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ๊ฒฐ๊ณผ ๋ชจ๋ธ์„ ๋‚˜ํƒ€๋‚ธ๋‹ค(Fig. 10).

Fig. 10. Kriging Interpolation of Unsupervised Learning-based Models. (a) K-Means, (b) K-Medoids, (c) DBSCAN

../../Resources/KSCE/Ksce.2025.45.2.0265/fig10.png

6.2 ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ • ๊ธฐ์ดˆ ์ž๋ฃŒ๋ฅผ ์œ„ํ•œ ๋ฐ์ดํ„ฐ ๋ณ€ํ™˜

ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •์„ ์œ„ํ•œ ๊ธฐ์ดˆ์ž๋ฃŒ๋ฅผ ์ œ์ž‘ํ•˜๋Š”๋ฐ GIS ์†Œํ”„ํŠธ์›จ์–ด๋ฅผ ํ™œ์šฉํ•œ ๋ถ„์„์€ ๋งค์šฐ ์œ ์šฉํ•˜๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ํฌ๋ฆฌ๊น… ๋ณด๊ฐ„๋ฒ•์„ ํ†ตํ•ด ๋ฏธ์ธก์ • ์ขŒํ‘œ๊ฐ’์— ๋Œ€ํ•ด ๋ณด๊ฐ„๋œ ๋ฐ์ดํ„ฐ๋ฅผ ArcMap 10.1 ์†Œํ”„ํŠธ์›จ์–ด์— ํˆฌ์˜ํ•œ ํ›„ ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •์„ ์œ„ํ•œ ๊ธฐ์ดˆ์ง€ํ˜•์ž๋ฃŒ๋ฅผ ์ œ์ž‘ํ•˜์˜€๋‹ค. ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •์„ ์œ„ํ•ด์„œ๋Š” ๋ž˜์Šคํ„ฐ ํ˜•์‹์ธ DEM ๋ชจ๋ธ๊ณผ ๋ฒกํ„ฐ ํ˜•์‹์ธ ๋ถˆ๊ทœ์น™ ์‚ผ๊ฐ๋ง(Triangulated Irregular Network, TIN) ๋ชจ๋ธ์„ ํ™œ์šฉํ•  ์ˆ˜ ์žˆ๋‹ค. ๋ถˆ๊ทœ์น™ ์‚ผ๊ฐ๋ง ๋ชจ๋ธ์€ ์ˆ˜์น˜ํ‘œ๊ณ ๋ชจํ˜•์— ๋น„ํ•ด ์ง€ํ˜•์˜ ๊ธฐ๋ณต ๋“ฑ ์„ธ๋ฐ€ํ•œ ํ‘œํ˜„์— ๋ฏผ๊ฐํ•˜๋ฉฐ ๊ธ‰๊ฒฝ์‚ฌ๋‚˜ ๊ตด๊ณก ๋“ฑ์˜ ๋ถˆ๊ทœ์น™์ ์ธ ๊ณ ๋„ ๋ฐ์ดํ„ฐ์—์„œ ์ •๋ฐ€๋„๊ฐ€ ๋†’๋‹ค. ๋”ฐ๋ผ์„œ ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ๋ถˆ๊ทœ์น™ ์‚ผ๊ฐ๋ง ๋ชจ๋ธ์„ ์ฑ„ํƒํ•˜์—ฌ ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •์— ํ™œ์šฉํ•˜์˜€๋‹ค. ๋‹ค์Œ์€ ArcMap 10.1 ์†Œํ”„ํŠธ์›จ์–ด์˜ Create TIN ๋„๊ตฌ๋ฅผ ์ด์šฉํ•˜์—ฌ ์ œ์ž‘๋œ ๊ฐ ๊ธฐ๋ฒ•์— ๋”ฐ๋ฅธ ๋ถˆ๊ทœ์น™ ์‚ผ๊ฐ๋ง ๋ชจ๋ธ์„ ๋‚˜ํƒ€๋‚ธ๋‹ค(Fig. 11).

Fig. 11. Triangulated Irregular Network Model. (a) K-Means, (b) K-Medoids, (c) DBSCAN

../../Resources/KSCE/Ksce.2025.45.2.0265/fig11.png

7. ์ตœ์ข… ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ • ๋ฐ ์ •ํ™•๋„ ๋น„๊ตยทํ‰๊ฐ€

7.1 ๋น„์ง€๋„ํ•™์Šต ๋ชจ๋ธ๋ณ„ ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •

์•ž์„œ ์ œ์ž‘๋œ ๋ถˆ๊ทœ์น™ ์‚ผ๊ฐ๋ง ๋ชจ๋ธ์„ ๋ฐ”ํƒ•์œผ๋กœ ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์„ ํ•ฉ์‚ฐํ•œ ์ตœ์ข… ํ† ๊ณต๋Ÿ‰์„ ๊ณ„์‚ฐํ•  ์ˆ˜ ์žˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ArcMap 10.1 ์†Œํ”„ํŠธ์›จ์–ด์˜ ์ฒด์ ๊ณ„์‚ฐ ๋„๊ตฌ์ธ Surface Volume ๋„๊ตฌ๋ฅผ ์ด์šฉํ•˜์—ฌ ๊ธฐ์ค€๋ฉด์„ ๊ธฐ์ค€์œผ๋กœ ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์„ ์‚ฐ์ •ํ•˜์˜€๋‹ค. Surface Volume ๋„๊ตฌ๋Š” TIN ๋ชจ๋ธ์„ ๊ธฐ๋ฐ˜์œผ๋กœ ๋ถˆ๊ทœ์น™ํ•œ ์ง€ํ‘œ๋ฉด์œผ๋กœ๋ถ€ํ„ฐ ํšจ์œจ์ ์ธ ๊ณ„์‚ฐ์ด ์šฉ์ดํ•œ ํ”„๋ฆฌ์ฆ˜ ์ฒด์  ๊ณ„์‚ฐ๋ฒ•์„ ๋ฐ”ํƒ•์œผ๋กœ ์ง€์ •๋œ ํ‘œ๊ณ  ๊ธฐ์ค€๋ฉด๊ณผ ์ž…๋ ฅ๋œ ํ‘œ๋ฉด ๊ฐ„์˜ ๊ณ ๋„ ์ฐจ์ด๋ฅผ ๊ณ„์‚ฐํ•˜์—ฌ ํ† ๊ณต๋Ÿ‰์„ ์‚ฐ์ •ํ•œ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ ๊ธฐ์ค€ ๊ณ ๋„๋Š” ๋Œ€์ƒ์ง€๊ฐ€ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ์ „์ฒด ๊ณ ๋„๊ฐ’ ์ค‘ ์ค‘์•™๊ฐ’์„ ์‚ฐ์ •ํ•˜์—ฌ ํ™œ์šฉํ•˜์˜€๋‹ค. ์ค‘์•™๊ฐ’์€ ํ‰๊ท ๊ฐ’์— ๋น„ํ•ด ๋น„๋Œ€์นญ์ ์ธ ํŠน์ง•์„ ๊ฐ€์ง€๋Š” ์ง€ํ˜•์—์„œ ํ˜„์‹ค์ ์œผ๋กœ ๊ณ ๋„ ์ •๋ณด๋ฅผ ๋ฐ˜์˜ํ•  ์ˆ˜ ์žˆ๋‹ค. ๋‹ค์Œ Eq. (1)์€ ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์„ ํ•ฉ์‚ฐํ•˜์—ฌ ์ตœ์ข… ํ† ๊ณต๋Ÿ‰์„ ์‚ฐ์ •ํ•˜๋Š” ๊ณต์‹์„ ๋‚˜ํƒ€๋‚ธ๋‹ค.

(1)
$ V_{cut}=\sum_{i=1}^{n}(A_{i}\times h_{i}),\: {if}{h}_{{i}}<0 \;\;\;\;\;\;\;\;\;\;\;\;:{Cut}\;{Volume}\\\\ {V}_{{fill}}=\sum_{{i}=1}^{{n}}({A}_{{i}}\times{h}_{{i}}),\: {if}{h}_{{i}}>0 \;\;\;\;\;\;\;\;\;\;\;\;:{Fill}\;{Volume}\\\\ {V}_{{total}}={V}_{{cut}}+{V}_{{fill}}\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;:{Total}\;{Volume}\\ $

๏ผŠ$A_{i}$ : Area of each pixel ($m^{2}$)

๏ผŠ$h_{i}$ : Elevation difference between the reference plane and the surface for each pixel ($m$)

๏ผŠ$n$ : Total number of each pixel

7.2 ์ •ํ™•๋„ ๋น„๊ต ๋ฐ ํ‰๊ฐ€

์—ฐ๊ตฌ๋Œ€์ƒ์ง€๋Š” ํ˜„์žฌ ํ•œ๊ตญ๋†์–ด์ดŒ๊ณต์‚ฌ๊ฐ€ ๊ด€๋ฆฌ ๊ฐ๋…ํ•˜์— ์žˆ์œผ๋‚˜, ์ •ํ™•ํ•œ ํ† ๊ณต๋Ÿ‰ ์ •๋ณด๋Š” ์ œ๊ณต๋˜์ง€ ์•Š์•˜๋‹ค. ์ด์— ๋”ฐ๋ผ ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ๊ธฐ์กด์— ๋„๋ฆฌ ์‚ฌ์šฉ๋˜๋Š” ๋ฐฉ์‹์ธ GNSS ์ธก๋Ÿ‰ ๋ฐฉ์‹์„ ๋ฐ”ํƒ•์œผ๋กœ ์‚ฐ์ •๋œ ํ† ๊ณต๋Ÿ‰์„ ๊ธฐ์ค€์œผ๋กœ ์ •ํ™•๋„ ๋น„๊ต ๋ฐ ํ‰๊ฐ€๋ฅผ ์ˆ˜ํ–‰ํ•˜์˜€๋‹ค. ์ •ํ™•๋„ ๋น„๊ต์˜ ๊ธฐ์ค€๊ฐ’์ด ๋˜๋Š” ํ† ๊ณต๋Ÿ‰์€ ์‹์ƒ์ด ๊ฐ€์žฅ ์ ์–ด ์‹์ƒ์˜ ์˜ํ–ฅ์ด ์ตœ์†Œํ™”๋˜๋Š” ์‹œ๊ธฐ์ธ 2024๋…„ 1์›”์„ ์„ ์ •ํ•˜์—ฌ ์‚ฌ์ „์ธก๋Ÿ‰์„ ํ†ตํ•ด ์‚ฐ์ •ํ•˜์˜€๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ๊ธฐ์ค€๊ฐ’์„ ๋ฐ”ํƒ•์œผ๋กœ ๊ฐ ๊ธฐ๋ฒ•์—์„œ ์‚ฐ์ •๋œ ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์ด ์ „์ฒด ํ† ๊ณต๋Ÿ‰์—์„œ ์ฐจ์ง€ํ•˜๋Š” ๋น„์œจ์„ ๊ณ„์‚ฐํ•˜๊ณ , ์ด๋ฅผ ๊ฐ€์ค‘์น˜๋กœ ์ ์šฉํ•˜์—ฌ ๊ฐ€์ค‘ํ‰๊ท ์„ ๊ณ„์‚ฐํ•œ ๋’ค, ์ตœ์ข… ์ •ํ™•๋„๋ฅผ ์‚ฐ์ •ํ•˜์˜€๋‹ค(Eq. (2)). ๋‹ค์Œ์€ RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ๋ฅผ ํ†ตํ•ด ์ทจ๋“๋œ ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•ด ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ ์šฉํ•˜์—ฌ ์ตœ์ข… ์‚ฐ์ •๋œ ํ† ๊ณต๋Ÿ‰ ๋ฐ ์ •ํ™•๋„ ๊ฒฐ๊ณผ์ด๋‹ค(Table 4). ๋น„์ง€๋„ํ•™์Šต ๊ธฐ๋ฐ˜ ํ•„ํ„ฐ๋ง ๊ธฐ๋ฒ•์„ ์ ์šฉํ•˜์ง€ ์•Š์€ ์ƒํƒœ์—์„œ ์‚ฐ์ •๋œ ๊ธฐ์กด ์ง€ํ˜•์˜ ์ตœ์ข… ํ† ๊ณต๋Ÿ‰ ๋น„์œจ์€ GNSS ์ธก๋Ÿ‰ ๊ฒฐ๊ณผ๋ฅผ ๊ธฐ์ค€์œผ๋กœ 104.6 %๋ฅผ ๊ธฐ๋กํ•˜๋ฉฐ, ์ „์ฒด ํ•„์š” ํ† ๊ณต๋Ÿ‰ ๋Œ€๋น„ 4.6 % ๊ณผ๋Œ€ ์‚ฐ์ •๋˜์—ˆ์ง€๋งŒ, ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰ ๊ฐ„์˜ ํŽธ์ฐจ๋Š” ๋ฌด๋ ค 26.3 %๋‚˜ ๋ฐœ์ƒํ•˜์˜€๋‹ค. ์ด๋Š” ๊ฐ€์ค‘์น˜๊ฐ€ ๋†’์€ ์„ฑํ† ๋Ÿ‰ ๋น„์œจ์ด 93 %๋กœ ๊ณผ์†Œ ์‚ฐ์ •๋˜๊ณ , ๊ฐ€์ค‘์น˜๊ฐ€ ๋‚ฎ์€ ์ ˆํ† ๋Ÿ‰ ๋น„์œจ์ด 119.3 %๋กœ ๊ณผ๋Œ€ ์‚ฐ์ •๋œ ๋ฐ ๋”ฐ๋ฅธ ๋ถˆ๊ท ํ˜•์ด ์ตœ์ข… ๊ฒฐ๊ณผ์— ์˜ํ–ฅ์„ ๋ฏธ์นœ ๊ฒƒ์ด๋‹ค. K-Medoids ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์„ ๊ธฐ๋ฐ˜์œผ๋กœ ์‚ฐ์ •๋œ ํ† ๊ณต๋Ÿ‰ ๋น„์œจ์€ ๊ธฐ์ค€๊ฐ’ ๋Œ€๋น„ 100.3 %๋กœ ๋ฏธ์„ธํ•œ ๊ณผ๋Œ€ ์‚ฐ์ •์„ ๋ณด์—ฌ์ฃผ์—ˆ๋‹ค. ์„ฑํ† ๋Ÿ‰ ๋น„์œจ์€ 98.7 %๋กœ 1.3 % ๊ณผ์†Œ ์‚ฐ์ •๋˜์—ˆ๊ณ , ์ ˆํ† ๋Ÿ‰ ๋น„์œจ์€ 102.8 %๋กœ 2.8 % ๊ณผ๋Œ€์‚ฐ์ •๋˜์–ด, ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰ ๊ฐ„ ํŽธ์ฐจ๋Š” 4.1 %๋กœ ์ƒ๋Œ€์ ์œผ๋กœ ์ž‘์•„ ๊ท ํ˜• ์žกํžŒ ๊ฒฐ๊ณผ๋ฅผ ์ œ๊ณตํ•˜์˜€๋‹ค. ์ตœ์ข… ํ† ๊ณต๋Ÿ‰ ์ •ํ™•๋„ ์ธก๋ฉด์—์„œ ๋ณด์•˜์„ ๋•Œ, DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์€ ์ตœ์ข… ํ† ๊ณต๋Ÿ‰ ๋น„์œจ์ด ๊ธฐ์ค€๊ฐ’(100 %)์— ๊ทผ์ ‘ํ•œ 99.5 %๋ฅผ ๊ธฐ๋กํ•˜์—ฌ ์•ˆ์ •์„ฑ ์ธก๋ฉด์—์„œ ์šฐ์ˆ˜ํ•˜๋‹ค๊ณ  ๋ณผ ์ˆ˜ ์žˆ๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ๊ฐ€์ค‘์น˜๊ฐ€ ๋†’์€ ์„ฑํ† ๋Ÿ‰์€ 96.4 %๋กœ ๊ณผ์†Œ ์‚ฐ์ •๋˜์—ˆ๊ณ , ๊ฐ€์ค‘์น˜๊ฐ€ ๋‚ฎ์€ ์ ˆํ† ๋Ÿ‰์€ 104.8 %๋กœ ๊ณผ๋Œ€ ์‚ฐ์ •๋˜์–ด ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰ ๊ฐ„์˜ ํŽธ์ฐจ๊ฐ€ ๋น„๊ต์  ํฐ 8.4 %๋กœ ์ง‘๊ณ„๋˜์—ˆ๋‹ค. ์ด๋Š” DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง์ด ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์˜ ๊ฐ€์ค‘์น˜ ๋น„์ค‘์„ ๊ท ํ˜• ์žˆ๊ฒŒ ๋ฐ˜์˜ํ•˜์ง€ ๋ชปํ•œ ๊ฒฐ๊ณผ๋กœ, ์ƒ๋Œ€์ ์œผ๋กœ ๋ถˆ๊ท ํ˜•ํ•œ ๊ฒฐ๊ณผ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ํ† ์‚ฌ ์šด๋ฐ˜ ๋ฐ ๋ถˆํ•„์š”ํ•œ ์ž‘์—…๊ณผ์ •์„ ์ดˆ๋ž˜ํ•  ๊ฐ€๋Šฅ์„ฑ์ด ์žˆ๋‹ค. ๊ฒฐ๋ก ์ ์œผ๋กœ, K-Medoids ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ๋ฏธ์„ธํ•œ ๊ณผ๋Œ€ ์‚ฐ์ •(0.3 %)๊ณผ ์ ์€ ํŽธ์ฐจ(4.1 %)๋กœ ๋น„๊ต์  ๊ท ํ˜• ์žกํžŒ ๊ฒฐ๊ณผ๋ฅผ ์ œ๊ณตํ•œ ๋ฐ˜๋ฉด, DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ์ตœ์ข… ๊ธฐ์ค€๊ฐ’ ๊ทผ์ ‘์„ฑ(99.5 %)์€ ์šฐ์ˆ˜ํ–ˆ์œผ๋‚˜, ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰ ๊ฐ„์˜ ํŽธ์ฐจ๊ฐ€ ํฌ๋‹ค๋Š” ํ•œ๊ณ„๋ฅผ ๋ณด์˜€๋‹ค. ์ด๋Š” ๊ฐ ๊ธฐ๋ฒ•์˜ ํŠน์„ฑ๊ณผ ๊ฐ€์ค‘์น˜ ๊ธฐ๋ฐ˜ ๋ถ„์„์ด ์ตœ์ข… ๊ฒฐ๊ณผ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ์‹œ์‚ฌํ•˜๋ฉฐ, ์ƒํ™ฉ์— ๋งž๋Š” ๊ธฐ๋ฒ• ์„ ํƒ์˜ ์ค‘์š”์„ฑ์„ ๊ฐ•์กฐํ•œ๋‹ค. ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ • ์‹œ ์ ˆํ† ๋Ÿ‰๊ณผ ์„ฑํ† ๋Ÿ‰ ์‚ฐ์ • ๊ฒฐ๊ณผ๊ฐ€ ๊ท ์ผํ•˜์ง€ ์•Š๋‹ค๋ฉด, ์‹ค์ œ ๊ณต์‚ฌํ˜„์žฅ์—์„œ ๊ณ„ํš๋ณด๋‹ค ๋งŽ์€ ํ† ๊ณต๋Ÿ‰์„ ์š”๊ตฌํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด๋ฅผ ํ†ตํ•ด ์ถ”๊ฐ€์ ์ธ ํ† ์‚ฌ์ฒ˜๋ฆฌ ๊ณผ์ •์„ ์ดˆ๋ž˜ํ•˜์—ฌ ๊ณต์‚ฌ ํšจ์œจ์„ฑ์„ ์ €ํ•˜์‹œํ‚ค๋Š” ์›์ธ์ด ๋  ์ˆ˜ ์žˆ๋‹ค.

(2)
$ V_{total}=V_{cut}+V_{fill}\\\\ W_{cut}=\dfrac{V_{cut}}{V_{total}}\\\\ W_{fill}=\dfrac{V_{fill}}{V_{total}} $

$Accuracy_{total}=$$(W_{cut}\times Accuracy_{cut})+(W_{fill}\times Accuracy_{fill})$

๏ผŠ$W_{cut}$ : The weight based on the proportion of cut volume within the total earthwork volume

๏ผŠ$W_{fill}$ : The weight based on the proportion of fill volume within the total earthwork volume

Table 4. Comparing Accuracy of Overall Earthwork Volume

Surveying type

Cut volume

Ratio

GNSS(VRS)

21252.50 $m^{3}$

100 %

RTK-UAV LiDAR

Without Filtering

25362.60 $m^{3}$

119.3 %

K-Means

21999.66 $m^{3}$

103.5 %

K-Medoids

21844.08 $m^{3}$

102.8 %

DBSCAN

22108.46 $m^{3}$

104.0 %

Surveying type

Fill volume

Ratio

GNSS(VRS)

34549.19 $m^{3}$

100 %

RTK-UAV LiDAR

Without Filtering

32116.97 $m^{3}$

93.0 %

K-Means

34554.27 $m^{3}$

100.1 %

K-Medoids

34103.94 $m^{3}$

98.7 %

DBSCAN

33319.42 $m^{3}$

96.4 %

Surveying type

Total volume

Ratio

GNSS(VRS)

55801.69 $m^{3}$

100 %

RTK-UAV LiDAR

Without Filtering

57479.57 $m^{3}$

104.6 %

K-Means

56553.93 $m^{3}$

101.4 %

K-Medoids

55948.02 $m^{3}$

100.3 %

DBSCAN

55427.88 $m^{3}$

99.5 %

8. ๊ฒฐ ๋ก 

๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ์‹์ƒ ๋ฐ ์ง€ํ˜•์ง€๋ฌผ ๋“ฑ์œผ๋กœ ์ธํ•œ ๊ณ„์ ˆ์  ์ œ์•ฝ๊ณผ ์ธ๊ณต์  ์‹์ƒ ์ œ๊ฑฐ์— ๋”ฐ๋ฅธ ์ž‘์—… ๋น„ํšจ์œจ์„ฑ์˜ ํ•œ๊ณ„๋ฅผ ๊ทน๋ณตํ•˜๊ธฐ ์œ„ํ•ด RTK-UAV ๊ธฐ๋ฐ˜ LiDAR ์„ผ์„œ๋ฅผ ์ด์šฉํ•˜์—ฌ ๋Œ€์ƒ์ง€๋ฅผ ์ธก๋Ÿ‰ํ•˜๊ณ , ์ทจ๋“๋œ ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ ์šฉ

ํ•˜์—ฌ ์‹์ƒ๊ณผ ์ง€๋ฉด์„ ๋ถ„๋ฆฌํ•˜์˜€๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ์–ป์€ ์ง€๋ฉด ๋ฐ์ดํ„ฐ๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ํฌ๋ฆฌ๊น… ๋ณด๊ฐ„๋ฒ•์„ ์ˆ˜ํ–‰ํ•˜์—ฌ ๋ฏธ์ธก์ • ์ง€์ ์˜ 3์ฐจ์› ๋ฐ์ดํ„ฐ๋ฅผ ํ†ต๊ณ„์ ์œผ๋กœ ์˜ˆ์ธก ๋ฐ ๋ณด๊ฐ„ํ•˜๊ณ , GIS ์†Œํ”„ํŠธ์›จ์–ด๋ฅผ ํ†ตํ•ด ๋ถˆ๊ทœ์น™ ์‚ผ๊ฐ๋ง์„ ์ œ์ž‘ํ•˜์—ฌ ์ ˆํ† ๋Ÿ‰๊ณผ ์„ฑํ† ๋Ÿ‰์„ ๊ณ„์‚ฐํ•œ ๋’ค ์ตœ์ข… ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •์— ํ™œ์šฉํ•˜์˜€๋‹ค. ์ด๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์‚ฐ์ถœ๋œ ์ •ํ™•๋„๋ฅผ ๊ฐ๊ด€์  ์ง€ํ‘œ์ธ GNSS ์ธก๋Ÿ‰ ์„ฑ๊ณผ์™€ ์ •ํ™•๋„ ๋น„๊ต ๋ฐ ํ‰๊ฐ€๋ฅผ ์ˆ˜ํ–‰ํ•˜์˜€๋‹ค. ๊ทธ ๊ฒฐ๊ณผ, ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ ์šฉํ•˜์—ฌ ๊ณ„์‚ฐ๋œ ๊ฐ ๊ธฐ๋ฒ•์— ๋”ฐ๋ผ ๊ณ„์‚ฐ๋œ ์ ˆํ† ๋Ÿ‰๊ณผ ์„ฑํ† ๋Ÿ‰ ๋น„์œจ์€ ๊ธฐ์ค€๊ฐ’์— ๋Œ€๋น„ํ•˜์—ฌ ๊ณผ๋Œ€ ์‚ฐ์ • ํ˜น์€ ๊ณผ์†Œ ์‚ฐ์ •๋˜๋Š” ๊ฒฝํ–ฅ์„ ๋ณด์˜€๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ๊ฐ ๊ธฐ๋ฒ•์—์„œ ์‚ฐ์ •๋œ ์ „์ฒด ํ† ๊ณต๋Ÿ‰ ๋Œ€๋น„ ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์ด ์ฐจ์ง€ํ•˜๋Š” ์š”์†Œ์˜ ๋น„์œจ์— ๋”ฐ๋ผ ๊ณ„์‚ฐ๋œ ๊ฐ€์ค‘์น˜๋ฅผ ๊ณ ๋ คํ•˜์—ฌ ๋ถ„์„์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค. ํ† ๊ณต๋Ÿ‰์€ ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰์„ ํ•ฉ์‚ฐํ•˜์—ฌ ์‚ฐ์ถœ๋˜๋ฉฐ, ๋‘ ๊ฐ€์ง€ ์š”์†Œ ์ค‘ ์ฒด์ ๊ฐ’์ด ๋†’์€ ํ•ญ๋ชฉ์ด ๊ฐ€์ค‘์น˜๊ฐ€ ํฌ๊ฒŒ ๋ถ€์—ฌ๋˜์–ด ์ „์ฒด ๊ฒฐ๊ณผ์— ๋งŽ์€ ์˜ํ–ฅ์„ ๋ฏธ์น˜๊ฒŒ ๋œ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ํ† ๊ณต๋Ÿ‰์— ์ฐจ์ง€ํ•˜๋Š” ๋น„์œจ ์ค‘ ์„ฑํ† ๋Ÿ‰์ด ์ ˆํ† ๋Ÿ‰๋ณด๋‹ค ๋†’๊ฒŒ ์‚ฐ์ •๋˜์–ด, ๊ณ„์‚ฐ ๊ฒฐ๊ณผ ์„ฑํ† ๋Ÿ‰ ์ •ํ™•๋„๊ฐ€ ๊ฐ€์ค‘์น˜๊ฐ€ ํฌ๊ฒŒ ๋ถ€์—ฌ๋˜์—ˆ๊ณ , ์ „์ฒด ์ •ํ™•๋„์— ๊ฒฐ์ •์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์ณค๋‹ค. DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์˜ ๊ฒฝ์šฐ ์ข…ํ•ฉ์ ์œผ๋กœ ์‚ฐ์ถœ๋œ ํ† ๊ณต๋Ÿ‰ ๋น„์œจ์ด 100 %์— ๊ฐ€์žฅ ๊ทผ์ ‘ํ–ˆ์Œ์—๋„ ๋น„์ค‘์ด ๋†’์€ ์„ฑํ† ๋Ÿ‰์ด ๊ณผ์†Œ ์‚ฐ์ •๋˜์–ด ๊ฐ€์ค‘์น˜ ๊ธฐ๋ฐ˜ ํ‰๊ฐ€์—์„œ ๋ถˆ๋ฆฌํ•˜๊ฒŒ ์ž‘์šฉํ•˜์˜€๋‹ค. ๋ฐ˜๋ฉด K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง ๊ธฐ๋ฒ•์€ ๊ธฐ์ค€๊ฐ’์„ ์•ฝ๊ฐ„ ์ดˆ๊ณผํ•œ 101.4 %์˜ ๊ฒฐ๊ณผ๋ฅผ ๋ณด์˜€์ง€๋งŒ, ๊ฐ€์ค‘์น˜ ๋น„์ค‘์ด ๋†’์€ ์„ฑํ† ๋Ÿ‰ ์ •ํ™•๋„๊ฐ€ 100.1 %๋กœ ์‚ฐ์ถœ๋˜์–ด ์ „์ฒด์ ์ธ ์ •ํ™•๋„์— ๊ธ์ •์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์ณค๋‹ค. ๋˜ํ•œ, ์ ˆํ† ๋Ÿ‰ ์ •ํ™•๋„์™€์˜ ํŽธ์ฐจ๊ฐ€ 3.4 %๋กœ ๊ฐ€์žฅ ์ ์–ด, ๊ท ํ˜• ์žกํžŒ ์„ฑ๊ณผ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ 8.4 %์˜ ํŽธ์ฐจ๋ฅผ ๊ฐ€์ง€๋Š” DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง์„ ๋„˜์–ด ๊ฐ€์žฅ ์šฐ์ˆ˜ํ•œ ์ •ํ™•๋„๋ฅผ ์‚ฐ์ถœํ•˜์˜€์Œ์„ ๋ณด์—ฌ์ฃผ์—ˆ๋‹ค. K-Means ํด๋Ÿฌ์Šคํ„ฐ๋ง์€ ์„ฑํ† ๋Ÿ‰๊ณผ ์ ˆํ† ๋Ÿ‰ ๊ฐ„์˜ ๊ท ํ˜•์„ ์œ ์ง€ํ•˜๋ฉด์„œ ์ถ”๊ฐ€์ ์ธ ํ† ์‚ฌ์ฒ˜๋ฆฌ ๊ณผ์ •์˜ ๊ทœ๋ชจ๋ฅผ ์ตœ์†Œํ™”ํ•  ์ˆ˜ ์žˆ์–ด, ์‹ค์ œ ๊ณต์‚ฌํ˜„์žฅ์—์„œ์˜ ์ ์šฉ์„ฑ๊ณผ ํšจ์œจ์„ฑ์ด ๋†’์€ ๊ธฐ๋ฒ•์œผ๋กœ ํŒ๋‹จ๋˜์—ˆ๋‹ค. ์ด๋Ÿฌํ•œ ๊ฒฐ๊ณผ๋ฅผ ํ†ตํ•ด LiDAR ๋ฐ˜์‚ฌ ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ๋น„์ง€๋„ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ˆ˜ํ–‰ํ•œ ๋ฐฉ์‹์ด ํ–ฅํ›„ ๊ณ„์ ˆ์  ์ œ์•ฝ๊ณผ ์‹œ๊ฐ„๊ณผ ์ธ๋ ฅ ์†Œ๋ชจ๋ฅผ ์ค„์—ฌ ์ž‘์—… ๋น„ํšจ์œจ์„ฑ์„ ๊ทน๋ณตํ•˜๊ณ , ํšจ์œจ์ ์ธ ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ • ๋ฐฉ์‹์„ ์ œ์‹œํ•  ์ˆ˜ ์žˆ๋Š” ์ž ์žฌ์  ๊ฐ€๋Šฅ์„ฑ์„ ํ™•์ธํ•˜์˜€๋‹ค. ํ–ฅํ›„ ์—ฐ๊ตฌ์—์„œ๋Š” DBSCAN ํด๋Ÿฌ์Šคํ„ฐ๋ง์˜ ๋ณ€์ˆ˜ ์„ค์ •์— ๋Œ€ํ•œ ์‹ฌํ™” ์—ฐ๊ตฌ ์ง„ํ–‰ ๋ฐ ๋น„์ง€๋„ํ•™์Šต ๊ธฐ๋ฒ• ์ค‘ ํ•˜๋‚˜์ธ ์ฃผ์„ฑ๋ถ„ ๋ถ„์„(Principal Component Analysis, PCA)์„ ํ™œ์šฉํ•˜์—ฌ LiDAR ๋ฐ˜์‚ฌ๊ฐ•๋„ ๋ฐ์ดํ„ฐ๋ฅผ ์ •์ œํ•œ ํ›„, ํ† ๊ณต๋Ÿ‰ ์‚ฐ์ •์˜ ์ •ํ™•๋„๋ฅผ ํ‰๊ฐ€ํ•  ๊ณ„ํš์ด๋‹ค. ์ฃผ์„ฑ๋ถ„ ๋ถ„์„์€ ๋ฐ์ดํ„ฐ์˜ ์ฐจ์›์„ ์ถ•์†Œ ๋ฐ ์žก์Œ์„ ์ œ๊ฑฐํ•˜์—ฌ ์ค‘์š”ํ•œ ํŠน์ง•๋งŒ์„ ์ถ”์ถœํ•  ์ˆ˜ ์žˆ์–ด ๋น„๊ต์  ์ •์ œ๋œ ๋ฐ์ดํ„ฐ๋ฅผ ์–ป์„ ์ˆ˜ ์žˆ์–ด, ์ด๋ฅผ ํ†ตํ•ด ๋ฐ์ดํ„ฐ ์ฒ˜๋ฆฌ ํšจ์œจ์„ฑ์ด ๋†’์€ ๊ฒฐ๊ณผ๋ฅผ ์‚ฐ์ถœํ•  ์ˆ˜ ์žˆ๋Š” ๋Œ€์•ˆ์ด ๋  ์ˆ˜ ์žˆ์„ ๊ฒƒ์œผ๋กœ ์‚ฌ๋ฃŒ๋œ๋‹ค. ๋˜ํ•œ, ๋ณธ ์—ฐ๊ตฌ๊ฐ€ ์‹์ƒ์˜ ๊ตฌ์กฐ์™€ ๊ด‘ํ•™์  ์š”์†Œ๊ฐ€ ๋ณ€ํ™”ํ•˜๋Š” ๊ฐ€์„์ฒ ์—๋„ ํšจ๊ณผ์ ์œผ๋กœ ์ ์šฉ๋  ์ˆ˜ ์žˆ๋Š”์ง€์— ๋Œ€ํ•œ ๊ณ„์ ˆ์  ์—ฐ๊ตฌ๋„ ์ถ”๊ฐ€์ ์œผ๋กœ ํ•„์š”ํ•ด ๋ณด์ธ๋‹ค.

Acknowledgements

This Research was supported by Kyungpook National University Research Fund, 2024.

This paper has been written by modifying and supplementing the KSCE 2024 CONVENTION paper.

References

1 
"Deng, D. (2020). โ€œDBSCAN clustering algorithm based on density.โ€ Proc. 2020 7th Int. forum on electrical engineering and automation (IFEEA), IEEE, Hefei, China, pp. 949-953."DOI
2 
"Ghazal, T. M. (2021). โ€œPerformances of k-means clustering algorithm with different distance metrics.โ€ Intelligent Automation & Soft Computing, Tech Science Press, Vol. 30, No. 2, pp. 735-742, https://doi.org/10.32604/iasc.2021.019067."DOI
3 
"Gong, Y., Li, X., Du, H., Zhou, G., Mao, F., Zhou, L., Zhang, B., Xuan, J. and Zhu, D. (2022). โ€œTree species classifications of urban forests using UAV-LiDAR intensity frequency data.โ€ Remote Sensing, MDPI, Vol. 15, No. 1, pp. 110, https://doi.org/10.3390/rs15010110."DOI
4 
"Guo, Q., Li, W., Yu, H. and Alvarez, O. (2010). โ€œEffects of topographic variability and lidar sampling density on several DEM interpolation methods.โ€ Photogrammetric Engineering & Remote Sensing, Ingenta Connect, Vol. 76, No. 6, pp. 701-712, https://doi.org/10.14358/PERS.76.6.701."DOI
5 
"Lang, M. W., Kim, V., McCarty, G. W., Li, X., Yeo, I. and Huang, C. (2020). โ€œImproved detection of inundation below the forest canopy using normalized LiDAR intensity data.โ€ Remote Sensing, MDPI, Vol. 12, No. 4, pp. 707, https://doi.org/10.3390/rs12040707."DOI
6 
"Langhammer, J., Janskรฝ, B., Kocum, J. and Minaล™รญk, R. (2018). โ€œ3-D reconstruction of an abandoned montane reservoir using UAV photogrammetry, aerial LiDAR and field survey.โ€ Applied Geography, Elsevier, Vol. 98, pp. 9-21, https://doi.org/10.1016/j.apgeog.2018.07.001."DOI
7 
"Lee, K. and Lee, W. H. (2022). โ€œEarthwork volume calculation, 3D model generation, and comparative evaluation using \vertical and high-oblique images acquired by unmanned aerial vehicles.โ€ Aerospace, MDPI, Vol. 9, No. 10, pp. 606, https://doi.org/10.3390/aerospace9100606."DOI
8 
"Park, J. K. and Jung, K. Y. (2021). โ€œAccuracy evaluation of earthwork volume calculation according to terrain model generation method.โ€ Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography, KSGPC, Vol. 39, No. 1, pp. 47-54, https://doi.org/10.7848/ksgpc.2021.39.1.47 (in Korean)."DOI
9 
"Park, H. and Lee, D. H. (2021b). โ€œComparative analysis of filtering techniques for vegetation points removal from photogrammetric point clouds at the stream levee.โ€ Ecology and Resilient Infrastructure, KSIE, Vol. 8, No. 4, pp. 233-244, https://doi.org/10.17820/eri.2021.8.4.233 (in Korean)."DOI
10 
"Park, J. K. and Lee, K. W. (2021a). โ€œPointcloud density analysis for data acquisition condition for utilization of UAV LiDAR.โ€ Journal of the Korea Academia-Industrial, KAIS, Vol. 22, No. 7, pp. 131-136, https://doi.org/10.5762/KAIS.2021.22.7.131 (in Korean)."DOI
11 
"Scaioni, M., Hรถfle, B., Baungarten Kersting, A. P., Barazzetti, L., Previtali, M. and Wujanz, D. (2018). โ€œMethods from information extraction from lidar intensity data and multispectral lidar technology.โ€ The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, ISPRS, Vol. 42, pp. 1503-1510, https://doi.org/10.5194/isprs-archives-XLII-3-1503-2018."DOI
12 
"Schaefer, M. and Pearson, A. (2021). Accuracy and precision of GNSS in the field, Elsevier, Amstedam."DOI
13 
"Seong, J. H., Han, Y. K. and Lee, W. H. (2018). โ€œEarth-volume measurement of small area using low-cost UAV.โ€ Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography, KSGPC, Vol. 36, No. 4, pp. 279-286, https://doi.org/10.7848/ksgpc.2018.36.4.279 (in Korean)."DOI
14 
"Ushakov, A. V. and Vasilyev, I. (2021). โ€œNear-optimal large-scale k-medoids clustering.โ€ Information Sciences, Elsevier, Vol. 545, pp. 344-362, https://doi.org/10.1016/j.ins.2020.08.121."DOI
15 
"White, J. C., Woods, M., Krahn, T., Papasodoro, C., Bรฉlanger, D., Onafrychuk, C. and Sinclair, I. (2021). โ€œEvaluating the capacity of single photon lidar for terrain characterization under a range of forest conditions.โ€ Remote Sensing of Environment, Elsevier, Vol. 252, pp. 112169, https://doi.org/10.1016/j.rse.2020.112169."DOI
16 
"Yilmaz, V., Konakoglu, B., Serifoglu, C., Gungor, O. and Gรถkalp, E. (2018). โ€œImage classification-based ground filtering of point clouds extracted from UAV-based aerial photos.โ€ Geocarto Internationa, Taylor & Francisl, Vol. 33, No. 3, pp. 310-320, https://doi.org/10.1080/10106049.2016.1250825."DOI
17 
"Zhao, X., Yang, Z. and Schwertfeger, S. (2020). โ€œMapping with reflection-detection and utilization of reflection in 3d lidar scans.โ€ Proc. 2020 IEEE Int. Symp on Safety, Security, and Rescue Robotics (SSRR), IEEE, Abu Dhabi, United Arab Emirates, pp. 27-33."DOI