1. No category

(一社)日本リモートセンシング学会 第57回学術講演会論文集 2014年11月
A7
㧗ゎീᗘᖸ΅ SAR ⏬ീ࡜ GIS ࢹ࣮ࢱࢆ⏝࠸ࡓ㧗ᒙᘓ≀ࡢ㧗ࡉᢳฟ
Detection of high-rise building’s heights using high-resolution InSAR images and GIS data
‫ۑ‬㕥ᮌ ㈼ኴ㑻 1࣭ ࣜࣗ࢘ ࢙࢘ࣥ 2࣭ ᒣᓮ ᩥ㞝 2
Kentaro Suzuki, Wen Liu and Fumio Yamazaki
Abstract : The detection techniques of building heights using remote sensing data have been proposed in previous researches. In
this study, a detection method for high-rise building heights using high-resolution InSAR images and GIS data was proposed and
a case study was conducted. Specifically, an interferometric image was made from a pair of TerraSAR-X complex images and
layover areas were detected by an InSAR analysis. Then, building heights were measured using the detected layover areas and
GIS building footprints. Finally, detected heights were compared with the reference heights from Lidar data.
Keywords : high-rise building, height detection, TerraSAR-X, InSAR analysis
1. ࡣࡌࡵ࡟
ᘓ≀ࡢࣔࢽࢱࣜࣥࢢࡸࣔࢹࣜࣥࢢࡣ㸪㒔ᕷィ⏬ࡸ
⅏ᐖᑐ⟇ࡢ❧᱌㸪⎔ቃᙳ㡪ホ౯࡞࡝ࢆ⾜࠺ࡓࡵ࡟㔜
せ࡛࠶ࡿ㸬ࡇࡢ࡜ࡁྛࠎࡢᘓ≀ࡢᙧ≧ࢆᢕᥱࡍࡿ᪉
ἲ㸪࡜ࡃ࡟ᘓ≀㧗ࡉࡢᢳฟ᪉ἲࡀ኱ࡁ࡞ㄢ㢟࡜࡞ࡗ
࡚࠸ࡿ㸬
࣮ࣜࣔࢺࢭࣥࢩࣥࢢ࡟ࡼࡿᘓ≀㧗ࡉࡢᢳฟࡣ㸪ග
Ꮫ⏬ീࡢ❧యどࡸ࣮ࣞࢨ 㔞ࡶࡼࡃ⏝࠸ࡽࢀࡿࡀ㸪
ኳೃࡸ᫨ኪ㛫࡟㛵ࢃࡽࡎほ ࡛ࡁࡿྜᡂ㛤ཱྀ࣮ࣞ
ࢲ (SAR) ࡀ ᭷ ຠ ࡞ ᡭ ẁ ࡛ ࠶ ࡿ 㸬 ࡉ ࡽ ࡟ 㸪 ㏆ ᖺ
TerraSAR-X (TSX)ࡸ COSMO-SkyMed ࡞࡝ࡢ㧗ゎീ
⾨ᫍ SAR ⏬ീࡀ฼⏝࡛ࡁࡿࡼ࠺࡟࡞ࡾ㸪ᗈᇦ࣭ᐃ
ᮇⓗ࡟ヲ⣽࡞ᆅ⾲㠃ࡢほ ࡀྍ⬟࡜࡞ࡗࡓ㸬SAR ࡟
ࡼࡿᘓ≀㧗ࡉࡢᢳฟᡭἲࡣࡇࢀࡲ࡛ከࡃᥦ᱌ࡉࢀ
࡚࠸ࡿࡀ 1)㸪㒔ᕷࡢ㧗ᒙᘓ≀ࡢ㧗ࡉᢳฟ࡟࠾࠸࡚ࡣ
ᘓ≀ࡢࠕಽࢀ㎸ࡳࠖ࡜ࠕᖸ΅఩┦ࠖࢆ⏝࠸ࡓᡭἲࡀ
᭷⏝࡛࠶ࡿ࡜⪃࠼ࡽࢀࡿ㸬SAR ⏬ീ࡛ࡣ㸪ᗄఱᏛⓗ
࡞ṍࡳ࡟ࡼࡗ࡚ᘓ≀ࡀ⏬ീୖ࡟ಽࢀ㎸ࡴࡓࡵ㸪ࡇࡢ
ಽࢀ㎸ࡳ⠊ᅖࢆ≉ᐃࡋ㸪ᘓ≀ࡢ GIS ࢹ࣮ࢱ࡜⤌ࡳྜ
ࢃࡏࡿࡇ࡜࡛ྛࠎࡢᘓ≀ࡢ㧗ࡉࢆᢳฟࡀྍ⬟࡛࠶
ࡿ 2)㸬ࡲࡓ㸪ᚲせ࡞ฎ⌮ࢆ᪋ࡋࡓᖸ΅⏬ീ࡛ࡣᘓ≀
࡟ࡼࡗ࡚ᖸ΅⦤ࡀⓎ⏕ࡋ㸪⌮ㄽ್ࢆ⏝࠸ࡿࡇ࡜࡛ᘓ
≀ࡢಽࢀ㎸ࡳ⠊ᅖࢆ≉ᐃࡍࡿࡇ࡜ࡀ࡛ࡁࡿ 3)㸬ࡇࡢ
࡜ࡁ㸪㧗ᒙᘓ≀ࡣ఩┦ࡢኚ໬㔞ࡀ኱ࡁ࠸ࡓࡵ㸪ಽࢀ
㎸ࡳ⠊ᅖ࡟ᖸ΅⦤ࡀⓎ⏕ࡋࡸࡍࡃ㸪ಽࢀ㎸ࡳ⠊ᅖࡢ
≉ᐃࡀࡋࡸࡍ࠸≉ᚩࡀ࠶ࡿ㸬
ᮏ◊✲࡛ࡣ㸪㧗ゎീᗘᖸ΅ SAR ⏬ീ࡜ GIS ᘓ≀
㍯㒌ࢆ⏝࠸࡚㧗ᒙᘓ≀ࡢ㧗ࡉᢳฟࡍࡿࡇ࡜ࢆ┠ⓗ
࡜ࡍࡿ㸬ࡲࡎ TSX ࡢᖸ΅⏬ീࢆసᡂࡋ㸪ᖸ΅⦤ࡢ
⌮ㄽᘧ࠿ࡽᘓ≀ࡢಽࢀ㎸ࡳ⠊ᅖࢆᢳฟࡍࡿ㸬ࡉࡽ࡟㸪
ᢳฟࡋࡓಽࢀ㎸ࡳ⠊ᅖ࡜ GIS ᘓ≀㍯㒌ࢹ࣮ࢱࢆ⏝
࠸࡚ᘓ≀㧗ࡉࢆᢳฟࡋ㸪Lidar ࠿ࡽồࡵࡓ㧗ࡉ࡜ẚ
㍑ࡍࡿ㸬
1
Ꮫ⏕఍ဨ ༓ⴥ኱Ꮫ኱Ꮫ㝔 ᕤᏛ◊✲⛉ ᘓ⠏࣭㒔ᕷ⛉Ꮫᑓᨷ
ṇ఍ဨ ༓ⴥ኱Ꮫ኱Ꮫ㝔 ᕤᏛ◊✲⛉ ᘓ⠏࣭㒔ᕷ⛉Ꮫᑓᨷ
( ᡤᅾᆅ ࠛ263-8522 ༓ⴥᕷ✄ẟ༊ᘺ⏕⏫1-33 ᕤᏛ⣔⥲ྜ◊✲Ჷ4F)
(㐃⤡ඛ Tel; 043-290-3528, E-mail; [email protected])
2
2. ᑐ㇟ᆅᇦ࡜౑⏝ࢹ࣮ࢱ
ᮏ◊✲࡛ࡣ㸪࢔࣓ࣜ࢝ྜ⾗ᅜࡢ࢝ࣜࣇ࢛ࣝࢽ࢔ᕞ
ࢧࣥࣇࣛࣥࢩࢫࢥᕷࡢ㧗ᒙࣅࣝ⩌ࢆᑐ㇟ᆅᇦ࡜ࡋ
ࡓ(Fig.1)㸬౑⏝ࡍࡿ TSX ⏬ീࡣ 2007/12/05 ࡜ྠᖺ
12/27 ࡟ྲྀᚓࡉࢀࡓ HS ࣮ࣔࢻࡢ SLC ⏬ീ࡛࠶ࡿ㸬
ࡇࡢ⏬ീࡣ࡜ࡶ࡟࢔ࢪ࣐ࢫ᪉ྥࡢゎീᗘࡀ 1.10
(m/pixel) 㸪 ࢫ ࣛ ࣥ ࢺ ࣞ ࣥ ࢪ ᪉ ྥ ࡢ ゎ ീ ᗘ ࡀ 0.92
(m/pixel)࡛࠶ࡾ㸪༡ྥ㌶㐨࡟ࡼࡾྲྀᚓࡉࢀࡓ㸬ࡲࡓ㸪
Lidar ࡢ DEM ࡜ DSM ࢆᖸ΅⏬ീࡢᆅᙧ఩┦ࡢ⿵ṇ
࡜ᘓ≀㧗ࡉࡢ᳨ド⏝ࢹ࣮ࢱ࡜ࡋ࡚⏝࠸ࡓ㸬ࡇࡢࢹ࣮
ࢱࡢỈᖹ᪉ྥศゎ⬟ࡣ 2.0m㸪ᆶ┤ศゎ⬟ࡣ 0.06m ࡛
࠶ࡿ㸬౑⏝ࡍࡿ GIS ᘓ≀㍯㒌ࢹ࣮ࢱࡣ㸪ࢧࣥࣇࣛࣥ
ࢩࢫࢥᕷࡀ HP ୖ࡛බ㛤ࡋ࡚࠸ࡿࡶࡢࢆ౑⏝ࡋࡓ 4)㸬
Fig.2(a)࡟ᑐ㇟ᆅᇦࡢᘓ≀㍯㒌࡜ Lidar ࠿ࡽồࡵࡓᆅ
⾲㠃㧗ࡉࢆ♧ࡍ㸬ᮏ◊✲࡛ࡣ㸪㧗ࡉ 50m ௨ୖ࡛ಽࢀ
㎸ࡳ⠊ᅖࡀ௚ࡢᘓ≀ࡢᙳ࡟ධࡽ࡞࠸ࡼ࠺࡞ᘓ≀ 12
Ჷࢆᑐ㇟࡟㧗ࡉࡢᢳฟࢆ⾜࠺㸬ᑐ㇟ᘓ≀ࢆ Fig.2(b)
࡟♧ࡍ㸬
Fig.1. The study area in San Francisco, CA, U.S.A. and
shooting areas of TerraSAR-X (TSX) and Lidar data.
3. ᖸ΅⦤ࢆ⏝࠸ࡓᘓ≀ಽࢀ㎸ࡳ⠊ᅖᢳฟࡢᢳฟ
3-1. ᖸ΅⏬ീࡢసᡂ
ᘓ≀ࡢಽࢀ㎸ࡳ⠊ᅖࢆᢳฟࡍࡿ➨୍ẁ㝵࡜ࡋ࡚㸪
TSX ࡢᖸ΅⏬ീࢆసᡂࡋࡓ㸬12/05 ⏬ീࢆ࣐ࢫࢱ࣮
⏬ീ࡜ࡋ࡚㸪ࣝࢵࢡᩘ 2:2 ࡢึᮇᖸ΅⏬ീࢆసᡂࡍ
ࡿ㸬㌶㐨⿵ṇ࡜ Lidar ࡢ DEM ࢆ⏝࠸ࡓᆅᙧ఩┦ࡢ
㝖ཤࢆ⾜ࡗࡓᚋ㸪GoldStein ࣇ࢕ࣝࢱ࣮ࢆ㐺⏝ࡋࡓ㸬
̶ 17 ̶
Range
Azimuth
a
c
d
e
f
g
h
i
k
0
b
j
l
100m
Building Height
Fig.4. Schematic image of the InSAR observation.
(a)
(b)
Fig.2. Geocoded image of building heights measured
by Lidar data (a); slant-range phase image and target
buildings (b).
h
a
R
cos T
(2)
a : ࢫࣛࣥࢺࣞࣥࢪศゎ⬟ (m/pixel)
a
ᮏ⏬ീ࡛ࡣ a = 0.92 ࡞ࡢ࡛㸪h = 1.2R ࡜࡞ࡿ㸬
ᘧ(1)(2)࠿ࡽ㸪ࢫࣛࣥࢺࣞࣥࢪ᪉ྥࡢ㞄᥋ࡍࡿࣆ
ࢡࢭࣝࡢ఩┦ᕪ ǻ‫( ׋‬rad/pixel)ࡣ௨ୗࡢᘧ࡛⾲ࡉࢀࡿ㸬
b
c
'I
Building footprint
Fig.3. A close up of the phase image within the
yellow frame in Fig.2 (b).
Fig.2(b)ࡣᙉᗘ⏬ീ࡜ᖸ΅⏬ീࢆ㔜ࡡࡓࢫࣛࣥࢺࣞ
ࣥࢪ⏬ീ࡛࠶ࡾ㸪୍㒊ᣑ኱ࡋࡓᅗࢆ Fig. 3 ࡟♧ࡍ㸬
ࡇࡢ⏬ീ࠿ࡽᘓ≀ࡢಽࢀ㎸ࡳ᪉ྥ࡟ᖸ΅⦤ࡀⓎ⏕
ࡋ࡚࠸ࡿᵝᏊࡀ☜ㄆ࡛ࡁࡿ㸬
3-2. ᖸ΅⦤ࡢ⌮ㄽᘧ
ḟ࡟ᘓ≀㧗ࡉ࡜ᖸ΅఩┦ࡢ㛵ಀᘧࢆ⏝࠸࡚ಽࢀ
㎸ࡳ⠊ᅖࢆᢳฟࡍࡿ㸬SAR ࡟ࡼࡿᘓ≀ࡢほ ࡢᵝᏊ
ࢆ Fig. 4 ࡟♧ࡍ㸬ࡲࡎ㸪㧗ࡉ h (m)ࡢᘓ≀࡟ࡼࡿ఩┦
ࡢኚ໬㔞 ‫( ׋‬rad)ࡀ௨ୗࡢᘧ࡛⾲ࡉࢀࡿ㸬
I
4SBA
h
OH tan T
Bԋ : ⏬ീ㛫ࡢᆶ┤㌶㐨㛫㊥㞳 (m)
Ȝ : ࣮ࣞࢲࡢἼ㛗 (m)
H : ࣐ࢫࢱ࣮⏬ീྲྀᚓ᫬ࡢ⾨ᫍ㧗ࡉ (m)
ș : ධᑕゅ (rad)
(1)
4SaBA
OH sin T
(3)
ᮏ⏬ീ࡛ࡣ ǻ‫ = ׋‬0.65 rad/pixel ࡜࡞ࡿ㸬
ࡲࡓ㸪ᖸ΅⦤࡟ࡣ఩┦ࡀ ʌ ࠿ࡽ-ʌ ࡟ኚ໬ࡍࡿ఩┦
ࣛࢵࣉ(PW)ⅬࡀᏑᅾࡍࡿ㸬PW Ⅼ࡛ࡢ ǻ‫⌮ࡣ ׋‬ㄽୖ
-2ʌ ࡛࠶ࡾ㸪PW Ⅼࡣ୍ᐃ㛫㝸 T (pixel)࡛⌧ࢀࡿ㸬T
ࡣᘧ(1)࡜(2)࠿ࡽ௨ୗࡢᘧ࡛⾲ࡉࢀࡿ㸬
T
OH sin T
2aBA
(4)
ᮏ⏬ീ࡛ࡣ T = 9.66 pixels ࡜࡞ࡿ㸬
3-3. ಽࢀ㎸ࡳ⠊ᅖᢳฟ᪉ἲ
ᖸ΅⏬ീ࡟ࡘ࠸࡚㸪ᘧ(3)ࢆ฼⏝ࡋࡓ఩┦ᕪ࡟ࡼࡿ
ಽࢀ㎸ࡳ⠊ᅖᢳฟ࡜㸪ᘧ(4)ࢆ฼⏝ࡋࡓ PW Ⅼ㛫㝸࡟
ࡼࡿᢳฟࢆ⾜࠺㸬๓ฎ⌮࡜ࡋ࡚㸪ᖸ΅⏬ീ࠿ࡽࢫࣛ
ࣥࢺࣞࣥࢪ᪉ྥࡢ఩┦ᕪ⏬ീࢆసᡂࡋ㸪ǻ‫ < ׋‬-ʌ ࢆ
‶ࡓࡍࣆࢡࢭࣝࢆ PW Ⅼ࡜ࡋ࡚ᢳฟࡋࡓ㸬Fig.3 ࡢ
఩┦⏬ീ࠿ࡽᢳฟࡉࢀࡓ PW Ⅼࢆ Fig.5(a)࡟♧ࡍ㸬
ࡲࡎ㸪఩┦ᕪ࡟ࡼࡿᢳฟࢆ⾜࠺㸬఩┦ᕪ⏬ീࡢ
PW Ⅼ࡟ 2ʌ ࢆຍ࠼࡚఩┦࢔ࣥࣛࢵࣆࣥࢢࢆ⾜࠸㸪ࡑ
ࡢ⏬ീࡢ఩┦ᕪࡀ 0.325 < ǻ‫ < ׋‬0.975 ࢆ‶ࡓࡍⅬࢆ
ᮏ⏬ീ࡛ࡣ㸪Bԋ = 561.6㸪Ȝ = 3.1×10-2㸪H = 5.01×105㸪 ᢳฟࡋࡓ㸬ࡑࡋ࡚㸪ᢳฟࡋࡓࣆࢡࢭࣝࢆࢭࢢ࣓ࣥࢸ
࣮ࢩࣙࣥࡉࡏ㸪⤖ྜᚋࡢࣆࢡࢭࣝᩘࡀ 35 ௨ୗࡢࢭ
ș = 0.70 ࡛࠶ࡾ㸪‫ = ׋‬0.54h ࡜࡞ࡿ㸬
ḟ࡟㸪㧗ࡉ h ࡢᘓ≀ࡀࢫࣛࣥࢺࣞࣥࢪ⏬ീୖ࡟࠾ ࢢ࣓ࣥࢺࢆࣀ࢖ࢬ࡜ࡋ࡚㝖ཤࡋࡓ㸬᭱⤊ⓗ࡟ᢳฟࡋ
ࡅࡿಽࢀ㎸ࡳ㔞 R (pixel)ࡢ㛵ಀࡣ௨ୗࡢᘧ࡛⾲ࡏࡿ㸬 ࡓ⠊ᅖࢆ Fig. 5(b)࡟♧ࡍ㸬఩┦ᕪ࡟ࡼࡿᢳฟࡣ㸪1
̶ 18 ̶
a
b
Per 1 pixel
c
c
Fig.6. Illustration of detecting layover length.
0
PW point
1.3
ࣆࢡࢭࣝࡈ࡜࡟⠊ᅖࢆᢳฟࡍࡿࡓࡵ㸪⣽࠿࠸⠊ᅖࡲ
࡛ಽࢀ㎸ࡳࢆ᥎ᐃ࡛ࡁࡿ୍᪉㸪ࣀ࢖ࢬ࡟ࡼࡗ࡚ಽࢀ
㎸ࡳ⠊ᅖ࡛ࡶᢳฟ࡛ࡁ࡞࠸ࣆࢡࢭࣝࡶከ࠸㸬
(a)
a
Detected area
ࡑࡇ࡛㸪఩┦ᕪࡀ኱ࡁࡃ≉ᐃࡋࡸࡍ࠸ PW Ⅼ࡟ὀ
┠ࡋࡓᢳฟࡶヨࡳࡓ㸬ࡇࡢᡭἲ࡛ࡣ PW Ⅼ㛫㝸࡟ࡼ
ࡗ࡚ಽࢀ㎸ࡳ⠊ᅖࢆᢳฟࡍࡿ㸬๓ฎ⌮࡛ᢳฟࡋࡓ
PW Ⅼ࡟ࡘ࠸࡚㸪ࣞࣥࢪ᪉ྥࡢḟࡢ PW Ⅼࡲ࡛ࡢ㛫
㝸 T ࢆ⟬ฟࡋ㸪8 < T < 12 ࢆ‶ࡓࡍሙྜ㸪⮬㌟ࡢ PW
Ⅼ࠿ࡽḟࡢ PW Ⅼࡲ࡛ࡢࣆࢡࢭࣝࢆᢳฟࡍࡿ㸬ࡑࡋ
࡚㸪ྠࡌࡃࢭࢢ࣓ࣥࢸ࣮ࢩࣙࣥࢆ⾜ࡗ࡚㸪ࣆࢡࢭࣝ
ᩘ 35 ௨ୗࡢࢭࢢ࣓ࣥࢺࢆ㝖ཤࡋࡓ㸬ࡇࡢฎ⌮࡟ࡼ
ࡿᢳฟ⠊ᅖࢆ Fig. 5(c)࡟♧ࡍ㸬ࡇࡢᡭἲ࡛ࡣ㸪PW
Ⅼࡢࡳ࡟ὀ┠ࡍࡿࡓࡵ㸪PW Ⅼ㛫ࡢࣀ࢖ࢬࡢᙳ㡪ࢆ
ཷࡅࡎ࡟ಽࢀ㎸ࡳ⠊ᅖࢆᢳฟࡍࡿࡇ࡜ࡀ࡛ࡁࡿ㸬୍
᪉㸪1 ᖸ΅⦤࠶ࡓࡾࡢᢳฟ࡜࡞ࡿࡓࡵ㸪ᐇ㝿ࡢಽࢀ
㎸ࡳ⠊ᅖࢆ඲࡚ᢳฟ࡛ࡁࡿࢃࡅ࡛ࡣ࡞࠸㸬
b
c
(b)
a
Detected area
b
c
3-4. ᢳฟ⠊ᅖࡢ⤖ྜ
ࡑࡇ࡛㸪఩┦ᕪ࡜ PW Ⅼ㛫㝸࡟ࡼࡾᢳฟࡋࡓ⠊ᅖ
ࢆ⤖ྜࡍࡿࡇ࡜࡛཮᪉ࡢᢳฟ₃ࢀࢆ⿵᏶ࡋࡓ㸬Fig.
5(d)ࡣ୧ᡭἲ࡟ࡼࡿᢳฟ⠊ᅖࡢ࣮࢝ࣛྜᡂᅗ࡛࠶ࡾ㸪
ⓑࡣ୧ᡭἲ࡛ᢳฟ࡛ࡁࡓ⠊ᅖ㸪㯤Ⰽࡣ఩┦ᕪ࡟ࡼࡗ
࡚⿵᏶ࡉࢀࡓ⠊ᅖ㸪㟷ࡣ PW 㛫㝸࡟ࡼࡗ࡚⿵᏶ࡉࢀ
ࡓ⠊ᅖࢆ♧ࡍ㸬ࡇࡢᅗ࠿ࡽ㸪┦஫࡟ᢳฟ₃ࢀࢆ⿵᏶
ࡋ࡚࠸ࡿᵝᏊࡀࢃ࠿ࡿ㸬ᮏ◊✲࡛ࡣ㸪ᢳฟࡉࢀࡓࡍ
࡭࡚ࡢ⠊ᅖࢆᘓ≀ࡢಽࢀ㎸ࡳ࡜ぢ࡞ࡋ㸪㧗ࡉࡢᢳฟ
࡟⏝࠸ࡿ㸬
(c)
a
b
c
R:PD G:PD B:PW
(d)
Fig.5. Phase difference image with phase wrap
points (yellow pixel) (a); Detected layover area
from phase difference analysis (b); Detected layover
area from PW analysis (c); Color composite of two
detected areas by phase difference (PD) and phase
wrap (PW) analysis (d).
4. ಽࢀ㎸ࡳ࡜ GIS ࢹ࣮ࢱࢆ⏝࠸ࡓᘓ≀㧗ࡉࡢᢳฟ
4-1. ᘓ≀㧗ࡉᢳฟ᪉ἲ
ᢳฟࡋࡓಽࢀ㎸ࡳ⠊ᅖ࡜ GIS ᘓ≀㍯㒌ࢆ⏝࠸ࡓ
ᘓ≀㧗ࡉࡢᢳฟࡢᡭ㡰ࢆ௨ୗ࡟♧ࡍ㸬Fig.6 ࡢࡼ࠺
࡟ᘓ≀㍯㒌ࢆ 1 ࣆࢡࢭࣝẖ࡟⾨ᫍ᪉ྥ࡟⛣ືࡋ㸪
SAR ࡢධᑕഃࡢ㍯㒌ୖࣆࢡࢭࣝ࡟ྵࡲࢀࡿಽࢀ㎸
ࡳ⠊ᅖࡢ๭ྜࢆồࡵࡿ㸬ࡑࡋ࡚ࡇࡢ๭ྜࡀ 25%ࢆୗ
ᅇࡗࡓࡽࡑࡢ⛣ື⠊ᅖࡲ࡛ࢆಽࢀ㎸ࡳ㔞 R ࡜ࡋ㸪ᘧ
(2)࠿ࡽᘓ≀㧗ࡉࢆồࡵࡓ㸬
̶ 19 ̶
Table 1. Comparison table of the detected heights vs.
the Lidar heights.
Building ID
a
b
c
d
e
f
Detected Height (m)
71.8
51.4
113.6
108.8
49.0
89.7
Lidar Height (m)
77.0
67.0
116.0
115.0
55.0
90.0
Error (m)
-5.2
-15.6
-2.4
-6.2
-6.0
-0.3
Building ID
g
h
i
j
k
l
Detected Height (m)
102.9
88.5
143.5
102.9
65.8
74.2
Lidar Height (m)
106.0
90.0
125.0
99.0
67.0
84.0
-3.1
-1.5
18.5
3.9
-1.2
-9.8
Error (m)
Lidar height (m)
150
130
Average error: 6.1m
RMS error : 8.3m
110
ᖸ΅⏬ീࢆసᡂࡋ࡚㸪ᖸ΅⦤ࡢ⌮ㄽᘧ࠿ࡽ఩┦ᕪ࡜
఩┦ࣛࢵࣉⅬࡢ㛫㝸࡟ὀ┠ࡋࡓᘓ≀ࡢಽࢀ㎸ࡳ⠊
ᅖࡢᢳฟࢆ⾜ࡗࡓ㸬ࡉࡽ࡟㸪GIS ᘓ≀㍯㒌ࢆ⏝࠸࡚
ᘓ≀ࡢಽࢀ㎸ࡳ㔞ࢆỴᐃࡋ㸪ᘓ≀㧗ࡉࢆồࡵࡓ㸬᭱
ᚋ࡟ᮏᡭἲࡢᘓ≀㧗ࡉ࡜ Lidar ࡼࡾồࡵࡓᘓ≀㧗ࡉ
ࢆẚ㍑᳨ウࡋࡓ㸬
ᢳฟ⤖ᯝࡣᖹᆒㄗᕪ 6.1m ࡛࠶ࡾ㸪㧗ᒙᘓ≀ࡢ㧗
ࡉ᥎ᐃ࡜ࡋ࡚ࡣ㧗࠸⢭ᗘ࡛࠶ࡗࡓ㸬୍᪉㸪࿘㎶ࡢᵓ
㐀≀ࡸᘓ≀ࡢᙧ≧࡟ࡼࡗ࡚ᢳฟㄗᕪࡀ኱ࡁࡃ࡞ࡿ
ࡇ࡜ࡶ☜ㄆࡉࢀࡓ㸬
௒ᚋࡣ㸪࿘㎶ࡢᘓ≀❧ᆅ≧ἣࢆ⪃៖ࡋࡓᡭἲࡸ㸪
ᘓ≀ࡢᙧ≧࡟ࡶ␃ពࡋࡓᡭἲࢆ⪃࠼ࡿᚲせࡀ࠶ࡿ㸬
ࡲࡓ㸪᧜ᙳ᮲௳ࡀ␗࡞ࡿ⏬ീࢆ౑⏝ࡍࡿࡇ࡜࡛㸪ᖸ
΅⦤ࡢ⌮ㄽ್ࡀኚືࡋࡓሙྜ࡛ࡶᮏᡭἲࡢᢳฟࣇ
࣮ࣟࡀ᭷⏝࡛࠶ࡿ࠿ࢆ᳨ドࡍࡿணᐃ࡛࠶ࡿ㸬
ㅰ㎡
ᮏ◊✲࡛⏝࠸ࡓ TSX ⏬ീ࡜ Lidar ࢹ࣮ࢱࡣ 2012
IEEE Geoscience and Remote Sensing Society Data
Fusion Contest ࡼࡾᥦ౪ࡉࡏ࡚࠸ࡓࡔ࠸ࡓ㸬ࡇࡇ࡟ㅰ
ពࢆ⾲ࡍࡿ㸬
90
70
࠙ཧ⪃ᩥ⊩ࠚ
1) T. Toutin and L. Gray, : State-of-the-art of elevation
50
70
90 110 130 150
extraction from satellite SAR data, ISPRS Journal of
Detected height (m)
Photogrammetry and Remote Sensing, Volume 55,
Issue 1, pp. 13–33, 2000.
Fig.7. Comparison of the detected heights vs. the
2) W. Liu, F. Yamazaki, : Building height detection
Lidar heights.
from high-resolution TerraSAR-X imagery and GIS
data, Proceedings of 2013 Joint Urban Remote
4-2. ᢳฟ⤖ᯝ࡜᳨ドࢹ࣮ࢱ࡜ࡢẚ㍑
Sensing Event, IEEE, Sao Paulo, Brazil, CD-ROM,
ᢳฟࡋࡓᘓ≀㧗ࡉࢆ Lidar ࠿ࡽồࡵࡓ㧗ࡉࢆṇゎ
33-36, 2013.
࡜ࡋ࡚ẚ㍑ࡋࡓ⤖ᯝࢆ Table1 ࡜ Fig. 7 ࡟♧ࡍ㸬ᮏᡭ
ἲ࡟ࡼࡿᖹᆒㄗᕪࡣ 6.1m㸪RMS ࡣ 8.3m ࡜࡞ࡾ㸪
50% 3) ୖᮏ⣧ᖹ࣭ᑠᯘ㐩἞࣭బ➉ㄔ࣭ඣᓥṇ୍㑻࣭ᱵ
ࡢᘓ≀ࢆ 5m ௨ෆࡢㄗᕪ࡛㧗ࡉࡢ᥎ᐃࡀ࡛ࡁࡓ㸬㧗
ཎಇᙪ࣭ᯇᒸᘓᚿ࣭ᾆሯΎᓠ㸸SAR ࢖ࣥࢱ࣮ࣇ
ࡉ 50m ௨ୖࡢᘓ≀ࡀᑐ㇟࡛࠶ࡿࡇ࡜ࢆ⪃៖ࡍࡿ࡜㸪
࢙ࣟࢢ࣒ࣛ࠿ࡽࡢᆶ┤ᵓ㐀≀ࡢ⮬ືᢳฟ᪉ἲ㸪
⢭ᗘࡼࡃᘓ≀㧗ࡉࡀᢳฟ࡛ࡁࡓ࡜⪃࠼ࡽࢀࡿ㸬
᪥ᮏ࣮ࣜࣔࢺࢭࣥࢩࣥࢢᏛ఍➨ 56 ᅇᏛ⾡ㅮ₇఍
୍᪉ᢳฟㄗᕪࡢ኱ࡁ࠸ᘓ≀࡟ὀ┠ࡋ࡚ࡳࡿ࡜㸪ᘓ
ㄽᩥ㞟㸪pp.139-140, 2014.
≀ i ࡣ 18.5m 㐣኱࡟㧗ࡉࢆᢳฟࡋ㸪ᘓ≀ b ࡣ 15.6m 4) City and County of San Francisco WEB ࣮࣌ࢪ:
https://data.sfgov.org/Facilities-and-Structures/Build
㐣ᑠ࡟㧗ࡉࢆᢳฟࡋ࡚࠸ࡿ㸬ᘓ≀ i ࡢㄗᕪ࡟ࡘ࠸࡚
ing-Footprints-Zipped-Shapefile-Format-/jezr-5bxm
᳨ウࡋࡓ⤖ᯝ㸪ࡇࡢᘓ≀ࡢಽࢀ㎸ࡳ᪉ྥ࡟࠶ࡿ௚ࡢ
ᘓ≀ࡢಽࢀ㎸ࡳ⠊ᅖࡀΰྜࡋ࡚㧗ࡉࢆᢳฟࡋ࡚࠸
ࡓࡇ࡜ࡀศ࠿ࡗࡓ㸬ᘓ≀ b ࡟ࡘ࠸࡚ࡣ㸪࿘㎶ࡢᆅୖ
ᵓ㐀≀ࡸ✵ㄪᶵ࡞࡝ࡢᒇ᰿ୖᵓ㐀≀࡟ࡼࡿᙳ㡪࡟
ࡼࡾ㸪Ᏻᐃࡋࡓᖸ΅ࡀᚓࡽࢀ࡞࠿ࡗࡓࡇ࡜ࡀཎᅉ࡜
⪃࠼ࡽࢀࡿ㸬
50
4. ࡲ࡜ࡵ
ᮏ◊✲࡛ࡣ㸪㒔ᕷᇦࡢ㧗ᒙᘓ≀ࡢ㧗ࡉࢆ㧗ゎീᗘ
ᖸ΅ SAR ⏬ീ࡜ GIS ᘓ≀㍯㒌ࢹ࣮ࢱࢆ⏝࠸࡚ᢳฟ
ࡍࡿᡭἲࢆᥦ᱌ࡋࡓ㸬ලయⓗ࡟ࡣ㸪TerraSAR-X ࡢ
̶ 20 ̶