Machine Vision based Quality Management System for Tele-operated Concrete Surface Grinding Machine

(1)

Received April 24 2013, Revised May 14 2013, Accepted May 21 2013

Copyright ⵑ 2013 by the Korean Society of Civil Engineers

Ⲏቧ⸮⹃#㔖㘪Ὢ㡶#㨚′ⷆ♫#ⵣ␂Ἲ#Ⳃ㬚#ῶ⢞␂ⷂ#Ꮾ⇖#㩆⾆ኾὪ#⢚⡢㜚

׌୨ฅȵ඿਎૴ȵছஂ଀

Kim, Jeonghwan, Phi, Seung Woo, Seo, Jongwon

Machine Vision based Quality Management System for Tele-operated Concrete Surface Grinding Machine

ABSTRACT

Concrete surface grinding is frequently used for flatness of concrete surface, concrete pavement rehabilitation, and adhesiveness in pavement construction. The procedure is, however, labor intensive and has a hazardous work condition. Also, the productivity and the quality of concrete surface grinding highly depend on the skills of worker. Thus, the development of remote controlled concrete surface grinding equipment is necessary to prevent the environmental pollution and to protect the workers from hazardous work condition. However, it is difficult to evaluate the grinded surface objectively in a remote controlled system. Also, The machine vision system developed in this study takes the images of grinded surface with the network camera for image processing. Then, by representing the quality test results to the integrated program of the remote control station, the quality control system is constructed. The machine vision algorithm means the image processing algorithm of grinded concrete surface and this paper presents the objective quality control standard of grinded concrete surface through the application of the suggested algorithm.

Key words : Machine vision, Concrete grinding, Image processing, GPS

Ⅹಾ

⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᮡ⡍ᰆ໕᮹י⪵ੱ۵❭ᗱᮝಽᯙ⦽ᅕᙹ᯲ᨦŝə൉ኺ(Grooving) ᜽Ŗᮥ☖⦽⡍ᰆ໕᮹႑ᙹ܆ಆᮥv⪵⦹Ñӹ⠪

┥ᖒᮥ⪶ᅕෝ᭥⦹ᩍᯱᵝᱢᬊࡹ۵Ŗჶᯕ݅. ə్ӹə᯲ᨦ✚ᖒᯕי࠺Ḳ᧞ᱢᯕŁᇥḥ, ᜍ్ḡ, ᗭᮭ॒ᮝಽᯙ⦽ᮁ⧕⦽᯲ᨦ⪹Ğᮥᅕᮁ

⦹Łᯩᮝ໑ᰆእෝ݅൉۵ʑ܆Ŗ᮹ᙺಉࠥᨱ঑௝ᔾᔑᖒၰᱩᔎ⣩ḩ᮹⠙₉aⓑĞ⨆ᯕᯩ݅. ঑௝ᕽᰆእ᳦᳑ᯱab᳦᭥⨹ᨱי⇽ࡹḡ

ᦫࠥಾ⦹ʑ᭥⦽ᬱĊ᳦᳑⎹Ⓧญ✙⢽໕ᱩᔎᰆእ}ၽᯕ⦥᫵⦹݅. ᬱĊ᳦᳑⪹Ğᨱᕽ۵᳦᳑ᯱa~šᱢᯙᱩᔎ⣩ḩᮥ⪶ᯙ⧉ŝ࠺᜽ᨱᰆ እaĥ⫮Ğಽᨱ঑௝᯲ᨦᯕ᪍ၵ෕íᙹ⧪ࡹŁᯩ۵ḡෝ⪶ᯙ⧁ᙹᯩࠥಾ⦹۵ḡᬱ᜽ᜅ▽ᯕ⦥᫵⍡ࡹ۵ၵ, ᅙᩑǍᨱᕽ۵ນᝁእᱥ᜽ᜅ▽

(Machine Vision System)ŝGPSෝᱢᬊ⦹ᩍօ✙ᬭⓍ⋕ີ௝ಽ↍ᩢ⦽ᱩᔎ໕᮹ᯕၙḡෝॵḡ▙ᩢᔢ⃹ญ(Image Processing)ŝᱶᮥÑ ℱ~šᱢᯕ໑⣩ḩšญ⥥ಽᖙᜅaᯱ࠺⪵ࡽ᜽ᜅ▽ᮥǍ⇶⦹ᩡ݅. ੱ⦽ᰆእ᮹⩥ᰍ᭥⊹᪡ႊ⨆, ᗮࠥ, ĥ⫮ࡽĞಽ᪡᮹᪅₉ᱶᅕəญŁ᯲

ᨦ᮹ḥ⃺॒ࠥᮥ᳦⧊ᱢᮝಽᔑ⇽⦹ᩍᬭⓍᜅ▭ᯕᖹᨱ⢽᜽⧉ŝ࠺᜽ᨱນᝁእᱥ᜽ᜅ▽ᨱ᮹⦽᯲ᨦ⣩ḩᱶᅕ᪡᮹☖⧊ᮥ᭥⦽⥥ಽəఉᮥ

}ၽ⦹ᩡᮝ໑, ⩥ᰆᱢᬊ▭ᜅ✙ෝ☖⧕ᅙʑᚁᮥá᷾⦹ᩡ݅.

áᔪᨕ ນᝁእᱥ, ⎹Ⓧญ✙ᱩᔎ, ᩢᔢ⃹ญ, GPS

֨ėěν

(2)

Fig. 1. Integrated Operation Program Interface

1. ᕽು

⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᮡ⡍ᰆ໕᮹י⪵ੱ۵❭ᗱᮝಽᯙ⦽

ᅕᙹ᯲ᨦŝə൉ኺ(Grooving) ᜽Ŗᮥ☖⦽⡍ᰆ໕᮹႑ᙹ܆ಆᮥ

v⪵⦹Ñӹ⠪┥ᖒᮥ⪶ᅕෝ᭥⦹ᩍᯱᵝᱢᬊࡹ۵Ŗჶᯕ݅. ə్

ӹʑ᳕᮹⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᮡᱩᔎ᯲ᨦ᜽ၽᔾࡹ۵ᇥḥᯕ ӹᜍ్ḡಽᯙ⦹ᩍๅᬑᩕᦦ⦽᯲ᨦ⪹Ğᨱᕽᯕ൉ᨕḡŁᯩ۵

⩥ᝅᯕ݅. ޵ᇩᨕ᯲ᨦᯱ᮹ᙺಉᱶࠥᨱ঑௝᯲ᨦ⣩ḩ᮹⠙₉a

⍅⣩ḩšญᨱᨕಅᬡᯕ঑෕໑᜖᜾ᱩᔎ᮹Ğᬑၽᔾࡹ۵ᜍ్ḡ ᮹⃹ญᨱ᮹⦽⪹Ğᱢྙᱽၰእ⬉ᮉᱢᯙ᯲ᨦᮝಽᯙ⦽eᱲᱢᯙ

ᗱ⧕ၽᔾ᮹ᬑಅaᯩ݅. ᯕ్⦽ᮁ⧕⦽᯲ᨦ⪹Ğᨱᕽၽᔾ⦹۵

݅᧲⦽⦝⧕ෝ↽ᗭ⪵⦹ʑ᭥⦹ᩍ᯲ᨦᯱෝ⩥ᰆᨱᕽມญਉᨕḥ

Ŕᨱ ᭥⊹᜽┅໕ᕽᱩᔎ᯲ᨦᮥ ᪅₉ ᨧᯕᙹ⧪⧁ ᙹ ᯩᮥᐱอ

ᦥܩ௝ᰆእ᳦᳑ᯱ᮹᮹ᔍđᱶᨱݡ⦽ḡᬱʑ܆ᮥw⇹ᬱĊ᳦᳑

⎹Ⓧญ✙ ⢽໕ᱩᔎ ʑĥ᮹ }ၽᯕ ⦥᫵⦹݅.

ᯕ్⦽ ᰆእ᮹ }ၽᮥ ᭥⧕ᕽ۵ ⢽໕ᱩᔎ ᯲ᨦ᮹ ↽ᱢ Ğಽ

ĥ⫮, ᜽Ŗᱶᅕၰᰆእ᮹ᔢ┽ᱶᅕᐱอᦥܩ௝ᬱĊᨱᕽᱩᔎ

⣩ḩᮥ⊂ᱶ· ❱݉⧁ᙹᯩ۵᜽ᜅ▽ŝᬱĊ᳦᳑ᮥ᭥⦽᯲ᨦᱶᅕ

☖⧊ ᜽ᜅ▽ Ǎ⇶ᯕ ⦥ᙹᱢᯕ݅. ᅙ ᩑǍᨱᕽ۵ ᯕ్⦽ ༊ᱢŝ

⦥᫵ᖒᮝಽນᝁእᱥᮥ⪽ᬊ⦽⣩ḩᱽᨕ᪡ᰆእᬕᬊᨱᯩᨕ᫵Ǎ

ࡹ۵༉ुᱶᅕ᪡ᯕෝၵ┶ᮝಽ᳦᳑ᯱ᮹᮹ᔍđᱶ, ᷪᰆእ᳦᳑

໦ಚᮥԕตᙹᯩ۵᯲ᨦᱶᅕ☖⧊ᬕᬊ᜽ᜅ▽ᮥᱽ᜽⦹ᩡ݅.

ᯕෝ᭥⧕ນᝁእᱥ᜽ᜅ▽ᮥ⪽ᬊ⦽ᔍಡᩑǍෝᇥᕾ⦹ᩍ↽ᱢ᮹

ᩢᔢ⃹ญ᦭Łญ᷹ᮥ}ၽ⧉ŝ࠺᜽ᨱ, Łᱶၡ᮹᭥⊹⊂ᱶʑᚁᯙ

RTK(Real-Time Kinematic) GPS ᪡᮹ᩑĥෝ☖⧕Ğಽၰ᯲ᨦ

⣩ḩđŝෝ᳦⧊ᱢᮝಽ⢽⩥⧁ᙹᯩ۵⦹ऽᭉᨕᖁᱶၰ⥥ಽəఉ

}ၽᮥᙹ⧪⦹ᩡ݅. ੱ⦽᜽ᜅ▽☖⧊݉ĥᨱᕽ⥥ಽəఉe࠺ʑ⪵

ྙᱽෝ⧕đ⦹ʑ᭥⦹ᩍ ⥥ಽ☁⎽ᮥᱶ᮹⦹ᩡᮝ໑⩥ᰆᝅ⨹ᮥ

☖⧕ ᖒ܆ᮥ á᷾⦹ᩡ݅.

2. ᬱĊ᳦᳑⎹Ⓧญ✙⢽໕ᱩᔎᰆእ}ၽ

2.1 ෇݁ଆઘऀࢂ

⎹Ⓧญ✙⢽໕ᱩᔎᰆእ᮹༉ℕ۵⡎110cm, ʙᯕ۵210cm,

׳ᯕ۵115cm, ⅾᵲప1tonᮝಽǍᖒ⦹ᩍᗭ⩶✙౎ᨱࠥᱢᰍa

a܆☁ಾ⦹ᩡᮝ໑, ॵᲅᨵḥ᮹⇽ಆᮥᯕᬊ⦹ᩍ⎹Ⓧญ✙⢽໕ᮥ

ᱩᔎ⦹۵ 65cm ⡎᮹ Õ᜾ ݅ᯕᦥ།ऽ ə௝ᯙ޵ෝ ⫭ᱥ᜽┅໑, 24V ႑░ญಽǍ࠺ࡹ۵2}᮹⬥ල, ᳑⨆ᮥ᭥⦽ᱥලᰆ⊹, ᰆእ

׳ԏᯕ᳑ᱩᮥ᭥⦽ᅖ࠺᜾ᮁᦶᝅฑ޵ෝ⡍⧉⦽݅. ੱ⦽2}᮹

GPS ᙹᝁʑෝᯕᬊ⦽ႊ⨆❭ᦦ(GPS ӹ⋉ၹ), b᳦ྕᖁᘂᙹᝁ

༉ऩ, ⋕ີ௝༉ऩ, ⎹Ⓧญ✙ᇥḥᮥ⯂᯦⦹ʑ᭥⦽Ḳḥᰆ⊹a

⇵a ᰆ₊ࡽ݅(Lee ॒, 2006).

2.2 ীඹൈଆઘऀࢂ

⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᮥᬱĊᨱᕽᙹ⧪⦹ʑ᭥⦹ᩍ2.1ᱩᨱᕽ

ᗭ}⦽⦹ऽᭉᨕෝᱽᨕ⦹Łᬕᬊ⧁ᙹᯩ۵ᗭ⥥✙ᭉᨕ᮹}ၽᯕ

⦥ᙹᱢᯕ݅. ᯕᨱ ᅙ ᩑǍᨱᕽ۵ MFC (Microsoft Foundation Class)ෝᯕᬊ⦹ᩍ⥥ಽəఉᮥ}ၽ⦹ᩡᮝ໑, ⥥ಽəఉ᮹⧖ᝍᱢ ᯙ ʑ܆ŝ ə ༊⢽۵ ᯲ᨦ ᙹ⧪ᯱಽ ⦹ᩍɩ ༉ु ᯲ᨦŝ ᯲ᨦ

ᱶᅕa ☖⧊ ᗭ⥥✙ᭉᨕᨱᕽ ᙹ⧪ ၰ ⢽⩥ࡹࠥಾ ⦹ᩡ݅. ੱ⦽

ᷪ, Ğಽĥ⫮, ⣩ḩšญ, ᰆእᱽᨕ᪡šಉࡽ༉ु᯲ᨦᱶᅕa⦽

⪵໕ᨱ⢽⩥ࡽ݅. Fig. 1ᨱᕽⴗᮡ᯲ᨦĞಽ᪡⩥ᰍʑĥᰆእ᮹

᭥⊹ෝ, ⴘ۵ ḥ⧪ ႊ᭥bᮥ, ⴙᮡ ⩥ᰍ ᗮࠥ᪡ ĥ⫮ Ğಽᨱᕽ

ჸᨕӽÑญ॒ᮥ, ⴚ۵ນᝁእᱥʑၹ⣩ḩᱽᨕđŝෝ⢽⩥⦹໑, ⴛ۵᜽ᜅ▽ᖅᱶᮥ⧁ᙹᯩíࡹᨕᯩ݅. ☖⧊ᗭ⥥✙ᭉᨕᨱᕽ

⣩ḩšญᨱݡ⦽ᱶᅕෝ⢽⩥⦹۵ⴚ᮹ʑ܆ᮡᱩᔎ᯲ᨦđŝෝ

Ğಽᨱ঑௝ʑᵡ⊹ᯕᔢ/ᯕ⦹ᩍᇡෝॵᜅ⥭౩ᯕ⧁ᙹᯩ۵ʑ܆ᮥ

w⇵ᨕ᧝⦽݅. ݅ᮭᰆᨱᕽ۵ᱩᔎ⢽໕᮹⊂ᱶႊ᜾ᨱݡ⦽á☁ෝ

☖⧕ນᝁ እᱥ ʑᚁᮥ⣩ḩšญ ႊ᜾ᨱ ᖁᱶ⦽ᯕᮁᨱ š⦹ᩍ

ʑᚁ⦹ࠥಾ ⦹ā݅.

3. ⎹Ⓧญ✙ᱩᔎ⢽໕᮹⣩ḩ⊂ᱶႊჶ

ᬱĊ᳦᳑⎹Ⓧญ✙⢽໕ᱩᔎᰆእ᮹ĞᬑᰆእᬕᱥᯱaᬱÑญ ᨱᕽᰆእෝ ᱽᨕ⦹ʑ ভྙᨱᱩᔎࡽ ᯲ᨦ᮹ ⣩ḩᮥ⊂ᱶ⧁ ᙹ

ᨧᮝအಽᬱÑญᨱᕽ᯲ᨦࡽᱩᔎ⣩ḩᮥ⠪a⧁ᙹᯩ۵⣩ḩšญ

᜽ᜅ▽ᯕ᫵Ǎࡽ݅. ᬱÑญᨱᕽᱩᔎ໕ᨱݡ⦽⣩ḩᮥ⊂ᱶ⦹ʑ

᭥⧕ᕽ ݅ᮭ Table 1ŝ zᯕ ᱢᬊ a܆⦽ ⣩ḩ ⊂ᱶ ʑᚁॅᮥ

á☁⦹ᩡ݅.

bႊ᜾ᄥᰆ݉ᱱၰᝅԕᖒ܆ᝅ⨹ᇥᕾđŝ᯦ℕᩢᔢ᜽ᜅ▽ᮥ

ᔍᬊ⦹۵äᯕaᰆ୑ᨕӽᖒ܆ᮥၽ⭹⦹ᩡᮝӹ⩥ᰍᔢᬊ⪵ࡹᨕ

ᯩ۵᯦ℕᩢᔢ᜽ᜅ▽ᮡࠥಽᗭᖒᄡ⩶⊂ᱶᮥ᭥⧕ᝅ᫙ᔍᬊᨱ

(3)

Table 1. Quality Control Method

Method Advantage Disadvantage

3D Camera Easy Implementation GRequires a Target/Pattern Laser

Scanner High Precision GRequires Some Time to Measure

GAffected by Vibration InfraRed (IR)

Sensor

Very Cheap

High Competitiveness GCause an Error by Dust

GLarge Error Image

Processing (Sensor + Camera)

GCheap

GOutdoor Adaptive

GLess Affected by Vibration

GLower Precision

GImage Processing Required

ᱢᬊ⦽ᖁ⧪}ၽᔍಡ۵ᯩᮝӹᵝಽᝅԕᨱᕽ᪥ᱽ⣩ॅ᮹⣩ḩá ᔍෝ᭥⦽ᬊࠥಽᯕᬊࡹŁᯩᨕᝅ᫙ᨱᕽ᮹ᔍᬊᨱ۵ḢᔍŲᖁᮥ

₉݉⧁ᙹᯩ۵₉᧲ส॒᮹ᄥࠥ᮹⇵aᱢᯙ᳑⊹a⦥᫵⦹݅.

ੱ⦽3D Camera᪡Laser Scannerෝᔍᬊ⦽᯦ℕᩢᔢ᜽ᜅ▽᮹

౩ᯕᱡŲᬱᮡḥ࠺ᨱԕǍᖒᯕᔢݚ⯩≉᧞⦹အಽḥ࠺ᯕᝍ⦽

ᅙ}ၽᰆእ᮹ᱢᬊᨱ۵ᄥࠥ᮹ḥ࠺ᱽᨕෝ᭥⦽ᰆ⊹᮹}ၽᯕ

᫵Ǎࡹᨕ ḥ݅.

᯦ℕᩢᔢ᜽ᜅ▽ᮡ40~200⽄᮹Łᇥ⧕܆ᮥᯕᬊ⦽᪥ᱽ⣩⢽໕ ᮹ᇩపᮥ⊂ᱶ⦹۵ߑᔍᬊࡹᨕḡ۵ᰆእಽ⎹Ⓧญ✙⢽໕᮹1~2 Ŕ᮹ᇩపǍeᮥ❭ᦦ⦹ʑᨱ۵ๅᬑᮁญ⦹ḡอᱩᔎ໕᮹ᱥℕᱢᯙ

ᱩᔎᩍᇡෝ❱݉ᮥ᭥⧕ᕽ۵ᨕಅᬡᯕᯩ݅. ᯕᨱၹ⧕ᖝᕽ᪡⋕ີ

௝ෝᯕᬊ⦽ᯕၙḡ⥥ಽᖙᝒᮡᝅ᫙ᱢᬊᨱᄥ݅ෙᱽ᧞ᮡᨧᮝ໑, ᰆእᨱᕽၽᔾ⦹۵ḥ࠺ᮝಽᯙ⦽ྙᱽᱱᮡⓍḡᦫᮡäᮝಽӹ┡ԍ ᮝ໑, ᯕၙḡ⥥ಽᖙᝒᨱᔍᬊࡹ۵ນᝁእᱥੱ۵օ✙ᬭⓍ⋕ີ௝۵

ᯝᱶ⦽Ⓧʑ᮹⎹Ⓧญ✙ᱩᔎ໕ᩢᔢᮥ≉ा⦹Ł, ≉ाࡽᩢᔢჵ᭥

ԕ᮹ᱩᔎᩍᇡෝᯝťᱢᮝಽ❱݉⧁ᙹᯩᮝအಽᅙᰆእ᮹⣩ḩšญ

᜽ᜅ▽ᮝಽ ᔍᬊ⦹ʑᨱ ๅᬑ ᱢ⧊⦽ äᮝಽ ❱݉ࡹᨩ݅.

4. ນᝁእᱥʑၹ⎹Ⓧญ✙ᱩᔎ໕᮹⣩ḩᱽᨕ

4.1 ࡉ਑णୢଭՍডंઉୡ૳ॷߢՑഠ

⎹Ⓧญ✙ᱩᔎ໕᮹⣩ḩšญෝ᭥⦹ᩍᩢᔢ⃹ญʑᚁᮥ᯦ࠥ⦹

ᩡ݅. ᩢᔢ⃹ญʑᚁᮡ݅᧲⦽ᇥ᧝ᨱᕽ⣩ḩᱽᨕၰ~ℕᯙ᜾

॒ᨱᔍᬊࡹŁᯩᮝ໑, Õᖅᇥ᧝ᨱᕽࠥᯕෝᱢᬊ⦽ᔍಡෝ₟ᦥᅝ

ᙹ ᯩ݅. ✚⯩ ⎹Ⓧญ✙ ၰ Ʊపᨱᕽ᮹ Ⓧ௺, ⦹ᙹšÑ, ᱩญ໕

॒᮹ᮁḡ/ᅕᙹŝᱶᨱᕽ, áᔍš᮹ᮂᦩ(Naked Eye)ᮝಽšₑᮥ

☖⧕ᅕᙹᩍᇡෝ❱ᄥ⦹۵ᇡᇥᮥݡᝁ⦹ʑ᭥⦹ᩍᩢᔢ⃹ญෝ

⪽ᬊ⦽ᯱ࠺đ⧉ĥ⊂᜽ᜅ▽ᨱݡ⦽ᩑǍaฯᯕḥ⧪ࡹŁᯩ݅.

Leu ॒ᮡ(2005) ░ձ ᜽Ŗჶ NATMᨱᕽ Ǖ₊໕᮹ đ⧉ᨱ

ݡ⦽ áᔍෝ ḥ⧪ ⦹۵ Ğᬑ, ᯕෝ ᯕၙḡ ᱡᰆ, šญ, ᯕၙḡ

⃹ญ, ᅕeၰᰍǍᖒᮥ☖⦽3D ᜽b⪵ෝ⦹ᩍḡ⩶ᇥᕾᮥᬊᯕ⦹

í⦹ᩡ݅. Lee ॒ᮡ(2006) Ʊపᨱᕽၽᔾ⦹۵ך(Rust)ᮥᯙ᜾⦹

ʑ᭥⦽ॵḡ▙⍍్ᩢᔢ⃹ญෝᱢᬊ⦹ᩍ᨜ᨕԙđŝෝ݅ᄡప

☖ĥᇥᕾ(Multivariate Statistical Analysis)ᮥ☖⦹ᩍᱶᅕߑᯕ

░ෝǍ⇶⦹ᩡ݅. Yu ॒ᮡ(2007) ░ձ໕᮹ᯕၙḡෝCCD ⋕ີ௝

ෝᯕᬊ⦹ᩍ░ձԕᇡ໕᮹Ⓧ௺ᨱݡ⦽ᯕၙḡෝ⫮ा⦹Łᯕෝ

Ⓧ௺áᔍ᦭Łญ᷹ᮥ☖⦹ᩍⓍ௺᮹᳕ᰍᩍᇡෝ❱a෥⧁ᙹ

ᯩ۵ᯱ࠺⪵᜽ᜅ▽ᮥ}ၽ⦹ᩡ݅. Woo ॒ᮡ(2008) ࠥಽ⡍ᰆ

᭥₉ᖁ⟹ᯙ❦ᯱ࠺⪵ෝ᭥⦹ᩍᔢᬊ✙౎ᨱᛞíᖅ⊹a܆⦹Ł

₉ᖁᯕḡᬭḥᇡᇥᮥᯙ᜾⧁ᙹᯩ۵ᩢᔢ⃹ญ᦭Łญ᷹ᮥ┲ᰍ⦽

༉ऩᮥ}ၽ⦹ᩡ݅. Region of Interest(ROI)᮹}ֱᮥ᯦ࠥ, ⥥ಽ ᖙᝒ┡ᯥᮥᵥᯕŁᗭትᩑᔑᯱ᪡2₉݅⧎᜾᮹༉ߙยၰ⋝อ

⦥░ยᮥ ᱢᬊ⦽ ᯕၙḡ יᯕᷩ ᱽÑ ᦭Łญ᷹ᮥ }ၽ⦹ᩡ݅.

Haran ॒ᮡ(2006) ࠥಽ᮹ᝅ᜽eᩢᔢᯕၙḡෝ⫮ा⦹ᩍ⃹ญ⧉

ᮝಽᕽࠥಽ⪹Ğᮥᯙ᜾⧁ᙹᯩ۵ᯕၙḡ⥥ಽᖙᝒ᦭Łญ᷹ᮥ

}ၽ⦹ᩡ݅.

ḡɩʭḡᔕ⠕ᅙນᝁእᱥ᮹Õᖅᇡྙᱢᬊᔍಡෝ☖⧕༊ᱢྜྷ

᮹ᔢ┽ෝ❭ᦦ⦹۵ߑᵝಽᔍᬊࡹŁᯩᮭᮥ⪶ᯙ⧁ᙹᯩᨩ݅.

ᱩᔎŝᱶᨱᕽ⦥ᩑᱢᮝಽၽᔾ⦹۵⎹Ⓧญ✙⢽໕᮹ ➉▕॒ᮥ

ᯕᬊ⦽݅໕, ᅙᩑǍᨱᕽࠥນᝁእᱥʑᚁᯕ∊ᇥ⯩ᱢᬊa܆⦹݅

Ł ❱݉ࡽ݅.

4.2 ࡉ਑णୢ෇݁ଆઘ֜নࢫ૶ઽ

ᬱĊ ⍉✙೅్ෝ ☖⧕ ⎹Ⓧญ✙ ⢽໕ᮥ ᱩᔎ⦽ ⬥ᨱ۵ ᰆእ

᳦᳑ᯱa⎹Ⓧญ✙ ⢽໕ᱩᔎ ᩍᇡၰ ⣩ḩᮥ ⪶ᯙ⧁ᙹ ᯩᨕ᧝

⦽݅. ᯕෝ᭥⧕ᅙᩑǍŝᱽᨱᕽ۵ນᝁእᱥʑᚁᮥᔍᬊ⦹ᩡᮝ໑,

Ⓧí2 ݉ĥಽǍᇥࡹᨕᬕᩢࡽ݅. ℌჩṙ݉ĥ۵ᱩᔎ໕᮹⣩ḩ⊂

ᱶᮥ ᭥⦽ ᩢᔢ ⫮ा ݉ĥᯕ໑, ࢱ ჩṙ ݉ĥ۵ ⫮ा⦽ ᩢᔢᮥ

⃹ญ⦹ᩍᱩᔎ໕᮹⣩ḩᮥ⊂ᱶ⧁ᙹᯩ۵ᩢᔢ⃹ญ݉ĥಽᯕ൉ᨕ ḡ໑ᯕෝ᭥⧕⋕ີ௝ᨱᕽ⫮ाࡽᩢᔢᮥ⎹✙೅ᖝ░ಽᘂᝁ⦹ʑ

᭥⦽ྕᖁ☖ᝁ᜽ᜅ▽Ǎ⇶ᯕᖁ⧪ࡹᨕ᧝⦽݅. ᯕၙḡ⥥ಽᖙᝒᮥ

᭥⦽⦹ऽᭉᨕၰݡఖᱢᯙᬕᩢᱩ₉۵Fig. 2᪡zᮝ໑, ᔍᬊࡽ

(4)

(a) Network camera (b) Converter (c) Lighting Fig. 3. Hardware configuration

Fig. 4. Image Processing Program Interface

ᩢᔢʑʑ ၰ ᱥᘂᰆ⊹۵ Fig. 3ŝ z݅.

ᯕၙḡ⥥ಽᖙᝒᵲᨱᵝŲᨱ᮹⦽əฝᯱၽᔾ᜽⣩ḩšญ

᪅₉ᨱᱢḡᦫᮡᩢ⨆ᮥၙ⊹íࡽ݅. əฝᯱᨱݡ⦽᪅₉۵ḡᩎᱢ

ᯕḥ⪵ಽ᪅₉ෝᵥᯝᙹᯩᮝӹ, ᵝŲᯕv⦽Ğᬑ໦ᇡ᪡ᦵᇡ᮹

ݡ᳑₉ᯕaᝍ⦹ᩍ׳ᮡݡ᳑(High-Contrast) ᩢᔢᨱ᮹⦽ᦵᇡᨱ ᕽ᮹ᱩᔎ໕ŝእᱩᔎ໕᮹Ǎᇥᯕᇩ໦⪶⦹íࡹ۵⩥ᔢᯕၽᔾ⦽

݅. ᯕ౑⩥ᔢᮥᨖᱽ⦹ʑ᭥⦹ᩍFig. 3᮹(c)᪡zᯕᰆእ⬥໕ᨱ

150W ⧁ಽñᱥǍࢱ}ෝᖅ⊹⦹ᩡ݅. ໦ᇡ᪡ᦵᇡ᮹ݡ᳑₉ᯕෝ

ᵥᯥᮝಽ៉ ԏᮡ ݡ᳑(Low-Contrast) ᩢᔢ ≉ा⪹Ğᮥ อॅᨕ, CCDa ኼᮥ ໦⪶⯩ ၼᦥॅᯝ ᙹ ᯩࠥಾ ⦹ᩡ݅.

⎹Ⓧญ✙ ⢽໕᮹ ᱶᅕ۵ ᰆእ ⬥໕ᨱ ⋕ີ௝ෝ ☖⧕ ᱩᔎࡽ

⢽໕᮹ ᯕၙḡෝ ≉ा⦽݅. ᰆእ ᯕ࠺ ᗮࠥෝ Łಅ⦹ᩍ ᯕၙḡ

≉ाeĊFPM(Frame Per Minute)ᮥđᱶ⦽⬥, ≉ाࡽᩢᔢᮥ

ྕᖁ௽⍉ქ░ෝ☖⦹ᩍ⎹✙೅ᖝ░ಽྕᖁᘂᝁࡽ݅. ᯕෝ᭥⦽

⦹ऽᭉᨕ۵ᱩᔎᔢ┽ෝᩢᔢᱶᅕಽ⇵⇽⦹ʑ᭥⦽⋕ີ௝᪡ᩢᔢ

ᮥᱽᨕᰆ⊹ಽᘂᝁ⦹ʑ᭥⦽ྕᖁ௽⍉ქ░, ᱽᨕᰆ⊹ᨱᕽ᮹ྕᖁ

௽⋕ऽa⦥᫵⦹݅. ᅙᩑǍᨱᕽ۵CPUaԕᰆࡹᨕᯩŁ, Łᮁ᮹

IPෝaḡŁᯩᨕᯕ޵֘(Ethernet)ᮝಽߑᯕ░ෝᘂ⇽ᯕa܆⦽

օ✙ᬭⓍ ⋕ີ௝(AXIS㫝, Model No. 210)ෝ ᔍᬊ⦹ᩡ݅. ᅙ

ᩑǍᨱᕽᔍᬊ⦽⋕ີ௝۵⥥ಽə౩᜽ቭႊ᜾ᮥᯕᬊ⦹Łᯩᨕ, NTSC ႊ᜾᮹ ᯕၙḡ ᖝᕽᨱ እ⦹ᩍ ᖁ໦⦽ ᩢᔢᮥ ≉ा⧁ ᙹ

ᯩ݅. ᅙ⋕ີ௝۵᯲ᨦ᜽᯲ŝ࠺᜽ᨱᩢᔢᮥ⫮ा⦹ŁJPEGಽ

ᦶ⇶⦹ᩍྕᖁᱥᘂ⦽݅. ᯝၹRAW ᯕၙḡᨱእ⦹ᩍJPEGᮡ

ᦶ⇶ࡽ⩶┽ᯕအಽᱥᘂ⧁ߑᯕ░᮹᧲ᯕᵥᨕᱥᘂ᜽eᮥ݉⇶⧁

ᙹ ᯩ݅.

ᱩᔎ໕ᯕၙḡ᮹ྕᖁᘂᙹᝁᮡAd-hoc ༉ऽྕᖁօ✙ᬭⓍෝ

☖⧕ ᯕ൉ᨕḥ݅. ᱩᔎ ᰆእ۵ ᯕ࠺᜽ ⫭ᱥᯕ ᰇᮝအಽ ḡ⨆ᖒ

ᦩ▭ӹ۵ᝁ഑⧁ᙹᨧíࡽ݅. ঑௝ᕽᰆእ⬥໕ᨱእḡ⨆ᖒᦩ▭ӹ

ෝ ᖅ⊹⦹ᩡᮝ໑360ࠥ, 100ၙ░ၹĞ, 2.4 GHz᮹ݡᩎ⡎ᮝಽ

ᘂᝁ⧉ᮝಽ៉⎹✙೅ᖝ░ᨱᕽ᮹ߑᯕ░ᙹᝁᝁ഑ࠥෝ⨆ᔢ᜽⎑

݅. ᙹᝁ⊂ᯙᱽᨕᰆ⊹ᨱ۵ḡ⨆ᖒᦩ▭ӹෝᖅ⊹, ᙹᝁbࠥ(30ࠥ

ԕ᫙)ᦩᨱ ᰆእa ᭥⊹⦹໕ ᬱ⪽⦽ ☖ᝁ a܆⦹݅.

≉ाࡽᩢᔢᱶᅕෝGPS ᜽eŝ࠺ʑ⪵⦽⬥ᝅ᜽eᩢᔢ⃹ญෝ

ᙹ⧪⦹໑⣩ḩšญෝḥ⧪⦽݅. ᩢᔢ⃹ญ݉ĥᨱᕽ۵MFC (Microsoft Foundation Class Library) ʑၹᮝಽ⯩ᜅ☁əఉᄡ⪹, ᯕḥ⪵ၰ

יᯕᷩᱽÑ᦭Łญ᷹ᮥǍ⇶⦹ᩍ↽᳦ᱢᮝಽᱩᔎ໕ᨱݡ⦽⣩ḩ đŝෝ▮ᜅ✙❭ᯝ(.txt)ŝJPEG ❭ᯝಽ⇽ಆ⦽݅. ⇽ಆࡽ⣩ḩđ ŝ۵☖⧊ᗭ⥥✙ᭉᨕ᪡ᩑĥࡹᨕ⢽໕ᱩᔎᬱĊ᳦᳑ᰆእa᯲ᨦ

⦽ ᭥⊹ᨱ ⣩ḩ đŝෝ ⢽᜽⦹í ࡽ݅(Fig. 4).

4.3 ઽঃंজੵճࠤஶԹࢳ

≉ाࡽᱩᔎ໕᮹ᔍḥᮥၵ┶ᮝಽᱩᔎᩍᇡෝ❱݉⧕᧝⦹۵ߑ

ᅙᩑǍᨱᕽ۵ᩢᔢ⃹ญʑᚁᮥ᯦ࠥ⦹ᩍᯕෝ⧕đ⦹Łᯱ⦹ᩡ݅.

⎹Ⓧญ✙⢽໕ᱩᔎ໕ᨱᕽ⦥ᩑᱢᮝಽၽᔾ⦹۵ኸɩྕ܍⩶┽ෝ

ᯕᬊ⦹ᩍᱩᔎᮁྕෝ❱݉⧁ᙹᯩ۵ᩢᔢ⃹ญ᦭Łญ᷹ᮥ}ၽ⦹

ᩡᮝ໑əŝᱶᮡᅙםྙ᮹4.3.1ᱩᇡ░4.3.3ᱩʭḡʑᚁࡹᨕ

ᯩ݅.

(5)

(a) Original Image (b) Image after the histogram stretch

(c) Original Image Histogram (d) Histogram after Stretching 4.3.1 ํਆഠֻ޽ਆൈߑಈ

⯩ᜅ☁əఉᄡ⪹ᯕ௡ᩢᔢᯕᨕࣂÑӹၾᮡĞᬑ, ੱ۵໦ᦵݡእ aԏᦥᖁ໦⦹ḡᦫᮡĞᬑᨱ⯩ᜅ☁əఉᮥմᮡᩢᩎᨱŉŁ൉

⟝✙ಅ໦ᦵݡእෝ׳ᯥᮝಽ៉ᩢᔢᮥᅕ݅ᖁ໦⦹íอऽ۵᦭Ł ญ᷹ᯕ໑, ᩢᔢ⃹ญᨱᕽᵲ᫵⦽ᱥ⃹ญŝᱶᵲ⦹ӹᯕ݅. ᅙᩑǍᨱ ᕽ۵ᱩᔎ໕ŝእᱩᔎ໕᮹໦ࠥ₉ᯕෝմ⩡ᯕḥ⪵ᨱᕽᔍᬊࡹ۵

ʑᵡsᯕ᳑ɩ޵ᮁ⬉⦹ࠥಾ⦹۵᮹ၙಽ៉⣩ḩšญ᜽ᜅ▽ᨱ

⯩ᜅ☁əఉ ᜅ✙౩⋎ ᦭Łญ᷹ᮥ ⡍⧉⦹ࠥಾ ⦹ᩡ݅.

⯩ᜅ☁əఉ ᄡ⪹ᨱ۵ Ⓧí ⯩ᜅ☁əఉ ⠪⪽⪵᪡ ⯩ᜅ☁əఉ

ᜅ✙౩⋎ᮝಽӹ٭ᙹᯩ۵ߑ, ⯩ᜅ☁əఉ⠪⪽⪵۵ኩࠥᙹ᮹༉ु

⦞ᖡ᮹ ٥ᱢsᮥ ᯕᬊ⦹ᩍ b⦞ᖡᮥ ᱶȽ⪵ࡽ ᔩಽᬕ⦞ᖡಽ

ݡℕ⦹۵݉ĥಽᯕ൉ᨕḥ݅. ⯩ᜅ☁əఉ⠪⪽⪵᦭Łญ᷹ᮡኩ᧞

⦽ ⯩ᜅ☁əఉᮥ aḥ ᩢᔢᨱ ᥑᩍ ໦ᦵݡእෝ ɚݡ⪵ ᜽┅۵

⬉ŝaᯩᮝӹ, ᅙᩑǍᨱᕽ۵໦ᦵݡእෝɚݡ⪵⧁Ğᬑᱩᔎ໕᮹

⩶┽a᪉ᱥ⯩ᅕ᳕ࡹḡ༜⦹۵Ğᬑaၽᔾ⦹အಽᱢ⧊⦹ḡᦫ݅.

঑௝ᕽ⯩ᜅ☁əఉᜅ✙౩⋎᦭Łญ᷹ᮥᱢᬊ⦹ᩡ݅. ⯩ᜅ☁əఉ

ᜅ✙౩⋎᦭Łญ᷹ᮡ⯩ᜅ☁əఉᮝಽᇡ░P (x, y)ෝx, y ᳭⢽᮹

⦞ᖡs, ᩢᔢ᮹⦞ᖡs᮹↽ᗭsŝ↽ݡsᮥǍ⦹ᩍ݅ᮭŝzᮡ

ᩑᔑŝᱶᮥ Ñ⊽݅.

©ÞƖì Ɨß á ć ©

à ©

©ÞƖìƗß à ©

Z ÏÒÒ (1)

ᯕᩑᔑᮥ☖⦹ᩍ໦ࠥ0ᇡ░↽ᗭ໦ࠥsŝ↽ݡ໦ࠥsᇡ░

໦ࠥ255ʭḡ᮹໦ࠥsᮥᩢᔢᨱᕽ ᔎᱽ⦹ᩍ, ↽ᗭ໦ࠥsŝ

↽ݡ໦ࠥsᮥbb໦ࠥ0ŝ໦ࠥ255ಽᄡ⪹⦽݅ᮭӹນḡ

sॅᮥŉŁ൉⟝✙ญíࡽ݅. ⯩ᜅ☁əఉᜅ✙౩⋎ᱥ⬥ᔍḥᮥ

እƱ⦹໕ Fig. 5᮹ (b)᪡ zᯕ ૽ಘ⧕ḥ ᩢᔢᮥ Ǎ⧁ ᙹ ᯩ݅.

⯩ᜅ☁əఉᄡ⪹ᮝಽᯙ⦹ᩍᅙᩑǍᨱᕽ᪡zᯕ⯩ᜅ☁əఉᯕ

✚ᱶǍeᨱ༑ಅᯩ۵Ğᬑ, ᱩᔎ໕ŝእᱩᔎ໕᮹໦ࠥ₉ᯕa⍅Კ

⬥ᗮᯕḥ⪵ŝᱶᨱᕽᥑᯕ۵ᯕḥ⪵sᯕ޵ᬒᮁ⬉⦹íࡽ݅.

4.3.2 ଲ஼ฃ

ᅙᩑǍᨱᕽᔍᬊࡽօ✙ᬭⓍ⋕ີ௝۵640*480᮹⧕ᔢࠥಽ

ᖅᱶࡹᨕ ᔍᬊࡹᨩᮝ໑, bb᮹ ⦞ᖡᨱ۵ 3ᬱᔪ(Red, Green, Blue) ᨱ ݡ⦽ ᱶᅕa 0ᇡ░ 255 ᔍᯕ᮹ ᱶᙹ sᮝಽ ⢽⩥ࡹᨕ

ᱡᰆࡹᨕᯩ݅. ᩢᔢᇥᕾ᮹ʑⅩ݉ĥಽ៉, ᱩᔎ໕᮹ᱩᔎᔢ┽ෝ

❭ᦦ⦹ʑ᭥⦹ᩍᯕḥ⪵(Binary thresholding) ᦭Łญ᷹ᯕaᰆ

⬉ŝᱢᯕ݅. ᯕḥ⪵۵✚ᱶ໦ࠥsᮥʑᵡ⦹ᩍᩢᔢᨱ᳕ᰍ⦹۵

༉ु⦞ᖡॅᮥ໦ࠥs0(⮲) ੱ۵255(႒)ᮝಽอ⢽⩥⦹۵᦭Łญ

᷹ᯕ݅.

ᅙᩑǍᨱᕽ۵᳦᳑ᯱaᱩᔎ᯲ᨦ᜽ᱶᅕ⪽ᬊ܆ಆᮥɚݡ⪵⦹

ʑ᭥⦹ᩍ≉ाᩢᔢᮥᝅ᜽eᮝಽ⪶ᯙ⧁ᙹᯩࠥಾ⦹ᩡ݅. ੱ⦽

᳦᳑ᯱ᮹᜽bᱢᯙ⠙᮹ෝ᭥⦹ᩍRGB sᮥaḡ۵⍍్ᯕၙḡෝ

ᱥᘂ⦹ࠥಾ ⦹ᩡᮝအಽ, ᯕḥ⪵ෝ ᭥⦹ᩍ ⮲႒ᩢᔢ(Gray-scale

(6)

Fig. 6. Basic Concept of Otsu Algorithm

image) ᮝಽᄡ⪹⧕᧝⦽݅. ⮲႒ᩢᔢᄡ⪹ᨱᔍᬊࡽŖ᜾ᮡᩍ్

aḡaᯩᮝӹ, ᅙᩑǍᨱᕽ۵ᬑญӹ௝⢽ᵡႊ᜾ᯙNTSCႊ᜾ᮥ

ᱢᬊ⦹ᩍ, Y(໦ࠥ)= 0.299R + 0.587G + 0.114B ᮹ᄡ⪹Ŗ᜾ᮥ

ᔍᬊ⦹ᩡ݅.

ᯕḥ⪵ᨱ۵Ⓧíࢱaḡႊ᜾ᯕ᳕ᰍ⦹۵ߑ, ⦹ӹ۵ᱥᩎᱢ

ᯕḥ⪵(Global thresholding)᪡, ḡᩎᱢᯕḥ⪵(Local thresholding) ಽ ӹ٩݅. ᱥᩎᱢᯕḥ⪵۵ᩢᔢ᮹ᱥᩢᩎᨱ݉ᯝᯕḥ⪵ʑᵡ

sᮥᱢᬊ⦹ᩍᄡ⪹⦹۵ႊჶᯕ݅. ᱥᩎᱢᯕḥ⪵۵݉ᯝĞĥs

อᮥ ⦥᫵ಽ ⦹ʑ ভྙᨱ ᩑᔑᗮࠥa ዁෕݅۵ ᰆᱱᯕ ᯩᮝӹ,

ᩢᔢᨱɁᯝ⦹ḡᦫᮡ໦ࠥ᮹ᄡ⪵a᳕ᰍ⦹Ñӹ, ᳑໦, əฝᯱ᪡

zᮡⓑᩢᩎᮥ₉ḡ⦹۵໦ࠥ₉a᳕ᰍ⦹۵Ğᬑ, ᯕḥ⪵aᬱ⪽⯩

ᯕ൉ᨕḡḡᦫᦥⓑ᪅₉ෝၽᔾ⦹íࡽ݅. ə౑ߑ⎹Ⓧญ✙⢽໕ᱩ ᔎᰆእ᮹᯲ᨦ✚ᖒᔢᝅ᫙᯲ᨦᯕݡᇡᇥᮥ₉ḡ⦹໑ᰆእᯱℕ

ੱ۵ᵝᄡᰆᧁྜྷಽᯙ⦹ᩍᱩᔎ໕ᨱəฝᯱaၹऽ᜽ၽᔾ⦹í

ࡽ݅. ᯕ Ğᬑ ݉ᯝ ʑᵡ sᮥ ᱢᬊ⦹۵ ᱥᩎᱢ ᯕḥ⪵۵ ๅᬑ

׳ᮡ᪅₉ෝaḩᙹၷᨱᨧ݅. ঑௝ᕽḡᩎᱢᯕḥ⪵᦭Łญ᷹ᮥ

᯦ࠥ⦹ᩍ ᯕ్⦽ ྙᱽॅᮥ ⧕đ⦹Łᯱ ⦹ᩡ݅.

ḡᩎᱢᯕḥ⪵᦭Łญ᷹ᯕ௡, ᯕḥ⪵ᨱ⦥᫵⦽Ğĥsᮥᩢᩎᨱ

঑௝݅෕íᱢᬊ⦹۵᦭Łญ᷹ᯕ݅. ✚ᱶⓍʑ᮹ษᜅⓍ, ᷪ, ⦞ᖡ

ᮥᱢᱶⓍʑಽə൚⦲(Grouping)ᮥ⦹ᩍᩢᔢᮥᇥ⧁⦹Łᇥ⧁ࡽ

ᩢᩎԕᨱᕽ᮹ᯕḥ⪵ෝᙹ⧪⦽݅. ᯕḥ⪵᦭Łญ᷹ᮡ⩥ᰍ60~70 aḡaᯩᮝ໑, ⎹Ⓧญ✙ᱩᔎᩢᔢᮥ⩥ᰍჵᬊᱢᮝಽᔍᬊࡹ۵

᦭Łญ᷹(⠪Ɂ, ↽ኩs, Otsu, ᅕeჶ॒)ᮥᱢᬊ⦹ᩍ↽ᱢ᮹ᯕḥ

⪵ ᦭Łญ᷹ᮥ ₟ᦥᅕᦹ݅(Oh ॒, 2003, Hryciw. ॒, 2006, Gonzalez. ॒, 2002). ᅙ ᩑǍᨱᕽ۵ ḡᩎᱢ ᯕḥ⪵ෝ ᭥⦹ᩍ

Ⓧʑ16×16 ษᜅⓍෝᔾᖒ⦹ᩡ݅. ᅙᩑǍᨱᕽᔍᬊ⦹۵օ✙ᬭⓍ

⋕ີ௝۵640×480᮹⧕ᔢࠥಽ↍ᩢ⦹အಽ⦽⥥౩ᯥᨱ40×30,

ᷪ1200}᮹ษᜅⓍෝᔾᖒ⦹íࡽ݅. ᩢᔢ᮹༉ु⦞ᖡᮥᜅ⋵⦹໑

bb᮹ษᜅⓍᨱᗮ⦹۵⦞ᖡᱶᅕෝᯕᬊ⦹ᩍOtsu ᦭Łญ᷹ᮥ

ᱢᬊ⦹ᩡ݅(Otsu, 1975).

Otsu ᦭Łญ᷹ᮡFig. 6ŝzᯕᵝಽ⯩ᜅ☁əఉᯕ✚ᱶ໦ࠥෝ

ʑᵡᮝಽࢱ}᮹⯩ᜅ☁əఉᇥ⡍ෝaḡ۵ᩢᔢᨱᕽ↽ݡᇥᔑ⊹

ෝaḡ۵໦ࠥෝĥᔑ⦹ᩍࢱ⯩ᜅ☁əఉᔍᯕ᮹ĞĥsᮥǍ⦹۵ ߑᥑᯙ݅. ᱩᔎ໕ᮡ݅ᯕᦥ།ऽ⍅░᮹✚ᖒᔢ, ᱩᔎ໕ŝእᱩᔎ໕ ᮝಽӹڹíࡽ݅. ࢱ໕᮹໦ࠥ₉ᯕᨱᕽၽᔾ⦹۵⯩ᜅ☁əఉ᮹

ᇥ⡍aࢱaḡಽᇥඹࡹအಽOtsu ᦭Łญ᷹ᮥᱢᬊ⦹ʑᨱᱢ⧊⦽

ᩢᔢ✚ᖒᮥ aḥ݅.

4.3.3 ڋଲல୪Ջ

ᯕḥ⪵aᯕ൉ᨕḥᩢᔢᮡ⎹Ⓧญ✙᮹ŉᰍ, əฝᯱ, ᯵ᩍᇥḥ, ၵ⒕ᯱǎ(Skid marks), ԏᮡݡ᳑ᩢᔢ(Low-contrast image) ॒ᨱ

᮹⦽יᯕᷩaၽᔾ⦹íࡽ݅. ᯕ్⦽יᯕᷩ۵⣩ḩšญෝ᭥⦹ᩍ

ᱽÑࡹᨕ᧝ ⦽݅. יᯕᷩ ᱽÑ ႊჶᮡ Gaussian, Soften, Blur, Erode ॒ᯕᯩ݅. ↽ᱢ᮹יᯕᷩᱽÑႊჶᮥ᭥⦹ᩍᅙᩑǍᨱᕽ᮹

ᱩᔎᩢᔢᮥᱢᬊ⦹ᩍəđŝෝá☁⦹ᩡ݅. ▭ᜅ✙đŝᯝၹᱢᮝ ಽฯᯕᥑᯕ۵Blur, Soften ॒᮹᦭Łญ᷹ᮡᵝಽྜྷℕෝ⮱ภ⦹í

อऽ۵ ᩑᔑᮥ ᙹ⧪⦹အಽ ⬉ŝa ਉᨕḡ۵ äᮝಽ ӹ┡ԍ݅.

Erode(⋉᜾)ᯕᇥḥ॒ᮝಽᯙ⦽᯲ᮡיᯕᷩෝ⬉ŝᱢᮝಽᱽÑ⦹

۵᦭Łญ᷹ᮝಽᅝᙹᯩᨩ݅. ⎹Ⓧญ✙ᱩᔎ໕᮹✚ᖒᮥᱽݡಽ

ၹᩢ⦹ᩍ ⬉ŝᱢᯙ יᯕᷩ ᱽÑෝ ᙹ⧪⦹۵ ᦭Łญ᷹ᮡ Erode ( ⋉᜾) ⬥Dilate(⪶ᰆ)ෝᱢᬊ⦽ᩕฝ(Opening) ᩑᔑᯕaᰆ⬉ŝ ᱢᯕ݅. ঑௝ᕽ⃹ญŝᱶᯕ݉ᙽ⦹Ł዁ෙ3*3 ษᜅⓍෝᔾᖒ⦹ᩍ

⋉᜾ᩑᔑᮥḥ⧪⦹ᩡ݅. ⋉᜾ᩑᔑᯕ௡ษᜅⓍԕ᮹໦ࠥ0ᯙ⦞ᖡᯕ

᳕ᰍ⦹۵ษᜅⓍ۵ᵲᦺ⦞ᖡᮥ0ᮝಽᄡ⪹⦹Ł, ၹݡಽษᜅⓍԕ ᮹ ໦ࠥa ༉ࢱ 255ᯙ Ğᬑᨱ۵ ᵲᦺ ⦞ᖡᮥ 255ಽ ᄡ⪹⦹۵

᦭Łญ᷹ᯕ݅.

ᯕ᦭Łญ᷹ᮡྜྷℕ᮹↽᫙b⦞ᖡᮥ႑Ğŝ࠺⪵᜽┅۵ᩑᔑᯕ အಽᇥḥ॒ŝzᮡⓍʑa᯲ᮡיᯕᷩ۵႑Ğᨱᔍ௝ḡíࡽ݅.

᯲ᮡ᯦ᯱॅŝzᯕᱩᔎ໕ੱ⦽zᯕᙹ⇶⦹íࡹᨕ᪅₉ෝၽᔾ⦹

íࡽ݅. ᯕᨱݡ⦽᪅₉ෝᅕᱶ⦹ʑ᭥⦽⪶ᰆᩑᔑᮥᙹ⧪⦹í

ࡽ݅. ⪶ᰆᩑᔑᯕ௡⋉᜾ᩑᔑŝ۵ၹݡ᮹}ֱᮝಽ៉, ྜྷℕ᮹⦞ᖡ

ᮥ ႑Ğ᮹ ⦞ᖡᨱ ⪶ᰆ⧉ᮝಽ៉ ྜྷℕa ໕ᱢᯕ ⪶ᰆ⦹í ࡽ݅.

ษᜅⓍԕ᮹⦞ᖡᯕ༉ࢱ0ᯙĞᬑᵲᦺ⦞ᖡᮥ0ᮝಽᄡ⪹⦹Ł

əᯕ᫙᮹Ğᬑ۵255ಽᄡ⪹⦽݅. ᩢᔢᨱᕽיᯕᷩ۵ᗭ໙ࡹᨕ

ᔍ௝ḥᔢ┽ᨱᕽᵥᨕुᱩᔎ໕ᮥ݅᜽⪶ᰆ⦹ᩍᬱᩢᔢ᮹ᱩᔎ໕ ŝzᮡ໕ᱢᮥaḡ۵ᩢᔢᮝಽᄡ⪹⦹íࡽ݅. ᯕಽ៉ᱩᔎ໕ᮡ

᪉ᱥ⯩ ᅕᱥ⦹໕ᕽ יᯕᷩ۵ ᱽÑ⦹í ࡽ݅(Fig. 7).

4.4 ઽঃंজէրౢߚ

ᯕၙḡ⥥ಽᖙᝒᯕ᪥ഭaࡹ໕đŝᩢᔢ᮹ᱥℕ⦞ᖡᵲᱩᔎ໕

ᮥӹ┡ԕ۵໦ࠥ255᮹⦞ᖡॅ᮹ ⟝ᖝ✙ෝᔑ⇽⦹íࡽ݅. ᅙ

ᩑǍᨱᕽ ᱶ᮹⦽ ⥥ಽ☁⎽ᮡ ❭ᯝ໦, ᜽e ᱶᅕ᪡ ᱩᔎ ໕ᱢŝ

ᱩᔎᩍᇡ(OK, Not Goodᮝಽ⢽⩥)᮹ԕᬊॅᮥ⏅ษ(Comma)ෝ

Ǎᄥᯱಽࢵ⩶┽ᯕ໑Fig. 8ᨱᕽᅕ۵ၵ᪡zᯕ▮ᜅ✙❭ᯝಽ

(7)

(a) After thresholding (b) Erode (c) Dilate Fig. 7. Applying Noise Removal Algorithm

File Edit Object View Help

Fig. 8. Results of Image Processing

⇽ಆ⦹ࠥಾ ⥥ಽəఉᮥǍᖒ⦹ᩡŁᯕ❭ᯝᮡ☖⧊ᗭ⥥✙ᭉᨕᨱ ᕽᝅ᜽eᮝಽಽऽ⦹ᩍᩢᔢ⃹ญ↽᳦đŝJPEG ❭ᯝŝ⧉̹

ॵᜅ⥭౩ᯕࡽ݅. ᩢᔢ❭ᯝbbᨱݡ⦽đŝෝ▮ᜅ✙❭ᯝ⦹ӹᨱ

ݕŁ ᯩᨕ ᱩᔎ ŝᱶ ᱥℕᱢᯙ đŝෝ ⦽ ჩᨱ ❭ᦦ⦹ʑ ⯹ु

ᱱᮥqᦩ, ᯕၙḡ⥥ಽᖙᝒ⥥ಽəఉᱶḡ⬥༉ुᱶᅕෝ⡍⧉☁ಾ

⦹۵ ಽə(log) ❭ᯝ ❭ᯝᮥ ᔾᖒ⦽݅.

4.5 ਏਆഗധ෍

4.5.1 ਏਆഗ૶ઽࢺ࣑

ᦿᕽᨙɪ⧩ॐᯕ, ☖⧊ᗭ⥥✙ᭉᨕ⥥ಽəఉᮡ᯲ᨦᯱ᮹᯲ᨦ

⠙᮹ᖒᮥ᭥⦹ᩍ༉ुᱶᅕෝᱽŖ⦹ࠥಾ}ၽࡹᨕᯩ݅. əญ⦹ᩍ

Fig. 9 ᪡zᯕນᝁእᱥ᜽ᜅ▽ᨱᕽ᮹⦹ऽᭉᨕ᪡RTK GPS ၰ

ྕᖁ ᘂᙹᝁ ༉ऩᨱᕽ ॅᨕ᪅۵ ߑᯕ░ෝ ☖⧊ ᗭ⥥✙ᭉᨕ ၰ

ນᝁእᱥᗭ⥥✙ᭉᨕᨱᕽᇥᕾ⦹ᩡ݅. ੱ⦽⥥ಽəఉᔍᯕᨱᕽ᮹

࠺ʑ⪵⥥ಽᖙᜅෝÑℱđŝᱢᮝಽ☖⧊ ᗭ⥥✙ᭉᨕᨱᕽ༉ु

ߑᯕ░ॅᮥॵᜅ⥭౩ᯕ⧁ᙹᯩࠥಾ⦹ᩡᮝ໑šಉߑᯕ░ᄁᯕᜅ

ෝ ᄥࠥಽ ᱡᰆ⦹۵ ᜽ᜅ▽ᮥ Ǎᖒ⦹ᩡ݅.

ᩢᔢ☖ᝁ༉ऩᯕ༉ࢱᖅ⊹aࡽᔢ┽ᨱᕽ, ⎹✙೅ᖝ░ᨱᕽ᮹

ԕᇡᱢᯙ⪹Ğᖅᱶᯕ⦥᫵⦹݅. օ✙ᬭⓍ⋕ີ௝ಽIP ᱲᗮᮥ☖⧕

ᩢᔢ☖ᝁ⪹ĞᮥǍ⇶⦽݅ᮭ, ᩢᔢᱥᘂ✙ญÑᝁ⪙᪡⧉̹ᱽᨕᰆ

⊹᮹FTP ᕽქෝ☖⦹ᩍJPEGᮝಽᦶ⇶ࡽᩢᔢ❭ᯝᮥᱥᘂၰ

ᱡᰆ⦽݅. ᱡᰆࡽᩢᔢᮥᯕၙḡ⥥ಽᖙᝒ⥥ಽəఉŝᩑĥ⦹ᩍ

↽᳦ đŝෝ ᔑ⇽⦽݅.

4.5.2 ധ෍ীඹൈଆઘࢫܛ׆ฃ

࠺ʑ⪵⥥ಽᖙᜅ۵⣩ḩšญᨱᕽᔾᖒ⦹۵đŝ▮ᜅ✙❭ᯝᮥ

ᯕᬊ⦹۵ߑ, GPS ᜽eᱶᅕ᪡đŝᩢᔢ᮹᜽eᱶᅕෝ⢽⩥⦹۵

đŝ❭ᯝ໦(օ✙ᬭⓍ⋕ີ௝᪡⍕⥉░᪡᮹ᕽქ᜽e)ᯕᯝ⊹⦹໕

đŝ ▮ᜅ✙ ❭ᯝᮥ ᩕࠥಾ ᖅĥ⦹ᩡ݅. ▮ᜅ✙ ❭ᯝᨱ ʑಾࡽ

ᱶᅕෝ☁ݡಽᩢᔢ⃹ญđŝ❭ᯝᮥ☖⧊ᗭ⥥✙ᭉᨕᨱ⢽᜽⦹Ł

᯲ᨦ ᖒŖ ᩍᇡෝ ⢽⩥⦹ࠥಾ }ၽ⦹ᩡ݅.

ੱ⦽ ☖⧊ ᗭ⥥✙ᭉᨕ ᵲᦺᨱ ᭥⊹⦽ ⢽໕ᱩᔎ ᯲ᨦ᮹ Ğಽ

ĥ⫮ᔢᨱ RTK GPSಽᇡ░ ᙹᝁࡹ۵ ⩥ᰍ ᰆእ᮹ ᭥⊹a ᬱ᮹

⩶┽ಽ⢽⩥ࡹ໑əᬱ᮹ᔪᮥᯕᬊ⦹ᩍ⣩ḩᱽᨕđŝaOKᯙ

ĞᬑⅩಾᔪ, ə౨ḡᦥܩ⦽Ğᬑᱢᔪᮝಽӹ┡ӹíࡽ݅. ᰆእ

᳦᳑ᯱ۵ᯕෝ ᵝ᜽⦹ᩍ Ğಽᔢᨱᱢᔪ Ǎeᯕ ၽᔾ⦽Ğᬑ ᰍ

ᱩᔎ໦ಚᮥྕᖁ᳦᳑ᰆ⊹ෝᯕᬊ⦹ᩍԕตᙹᯩᮝ໑, RTK GPS ۵ ᰆእ᮹ ᗮࠥ᪡ Ğಽᨱᕽ ჸᨕӽ Ñญෝ ⊂ᱶ⦹ᩍ ☖⧊

ᗭ⥥✙ᭉᨕᨱ ӹ┡ԕᵝᨕ ᳦᳑ᯱ᮹ ᮹ᔍđᱶᮥ ḡᬱ⦽݅. ᬑ⊂

⦹݉ᨱ۵ᯕၙḡ⥥ಽᖙᝒ᮹đŝ, ᷪᩢᔢ⃹ญđŝᯕၙḡ, ᱩᔎ ශ, ᱩᔎᩍᇡ, ᰍᱩᔎ⦥᫵Ǎe᮹}ᙹෝ٥ᱢ⦹ᩍᅝᙹᯩࠥಾ

ᯙ░⟹ᯕᜅෝ Ǎᖒ⦹ᩡ݅(Fig. 10).

(8)

Fig. 10. Display Screen with Quality Management Function

Table 2. Result of Machine Vision Test for Grinded Concrete Surface

Daylight Shadow

Average 16.71 16.21

Standard Deviation 4.07 2.24

Fiducial Value(%) 16.46

Accuracy 96.6% (350 sample were tested)

Fig. 11. Final Experiment

5. ⩥ᰆᱢᬊᝅ⨹đŝ

⣩ḩšญ᮹ ↽᳦ đŝෝ ᭥⦽ ᱩᔎᩍᇡ ʑᵡ ⟝ᖝ✙ ᔑᱶᮥ

᭥⦹ᩍᅙᩑǍᨱᕽᔍᬊࡽə௝ᯙ޵᮹⢽໕⩶┽᪡ᱩᔎ➉▕, ᝅ⨹ᮝಽᇡ░ӹ᪉ᯕၙḡ⥥ಽᖙᝒđŝෝእƱ⦹ᩍ↽ᱢ᮹ʑᵡ

ᮥ ᔑ⇽⦹ᩡ݅. ᝅ⨹ ᩢᔢᮡ ᱩᔎᯕ ᯹ ḥ⧪ࡹᨩ݅Ł ❱݉ࡹ۵

Ğᬑ(100% ᱩᔎ) ᯝၹᵝŲᔢ┽100 ⥥౩ᯥ, əฝᯱᔢ┽100

⥥౩ᯥᮥᖁᱶ⦹ᩍ⠪Ɂᮥĥᔑ⦹ᩡ݅. ੱ⦽ᅙᩑǍᨱᕽ350}

ᔹ⥭ᮥ᳑ᔍ⦹ᩍᱽᦩ⦽⣩ḩᱽᨕ᜽ᜅ▽᮹ ᱶ⪶ࠥෝ᦭ᦥᅕʑ

᭥⦹ᩍ100% ᱩᔎࡽǍeᨱݡ⦹ᩍᅙ᜽ᜅ▽ᮥᱢᬊ⦹Ł⣩ḩᱽ ᨕđŝaNGಽ⢽᜽ࡹ۵Ǎeᮥ᳑ᔍ⦹ᩡ݅. ⅾ12}᮹NGǍe,

ᷪ⣩ḩᱽᨕ᮹᪅₉aၽᔾ⦹ᩡᮝ໑ᯕෝ☁ݡಽᱶ⪶ࠥෝᇥᕾ⦽

đŝ96.6%᮹đŝෝ᨜íࡹᨩ݅. ᯕđŝ۵᜽ᜅ▽ᖅĥaᱶᨱᕽ ᮹ 95%ෝ չ۵ อ᳒ᜅ్ᬕ đŝಽ ᅝ ᙹ ᯩ݅.

⥥౩ᯥ ⦹ӹᨱ ᗭ᫵ࡹ۵ ⥥ಽᖙᜅ ᜽eᮡ ⠪Ɂ 1.50Ⅹᯕ໑, ᯕ۵4Ⅹݚ1⥥౩ᯥᮥ↍ᩢ⦹ࠥಾǍᖒ⦽CCD ⋕ີ௝᮹FPMᅕ

᯲݅ᮡᙹ⊹ᯕအಽ⥥ಽᖙᜅŝᱶᨱᕽॽ౩ᯕaᨧᨕ⥥ಽəఉe ᮹ ࠺ʑ⪵a ᯕ൉ᨕḡḡᦫ۵ ྙᱽ۵ ၽᔾ⦹ḡ ᦫᦹ݅.

əญŁFig. 11ŝzᯕ᜽ᜅ▽᮹ᬕᩢᖒŝǍ࠺ᝁ഑ᖒᮥ⠪a⦹

ʑ᭥⦹ᩍ}ၽࡽນᝁእᱥ᜽ᜅ▽ၰ☖⧊ᗭ⥥✙ᭉᨕෝ☁ݡಽ

⩥ᰆ▭ᜅ✙ෝᙹ⧪⦹ᩡᮝ໑ᩢᔢ⃹ญ᦭Łญ᷹ᮥ ☖⧕ᄡ⪹ࡽ

↽᳦ ᱩᔎ໕ ᯕၙḡa ☖⧊ ᗭ⥥✙ᭉᨕ ᬑ⊂ ⦹݉ᨱ ӹ┡ӹí

ࡹ໑OK, NGᨱݡ⦽⣩ḩšญđŝsᯕĞಽᔢᨱ݅ෙᔪᮝಽ

⢽⩥ᯕࡹ໑⟝ᖝ✙đŝsࠥ᪍ၵ෕íᅕᩍḡ۵äᮥ⪶ᯙ⦹ᩡ݅

(Fig. 11).

6. đು

ᅙᩑǍᨱᕽ۵⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᨱᕽ᮹⣩ḩᱽᨕෝ᭥⦽

ນᝁእᱥ᜽ᜅ▽ŝᬱĊ᳦᳑ᮥḡᬱ⦹ʑ᭥⦽ᩢᔢ⃹ญ᦭Łญ᷹

ŝ⥥ಽəఉᮥᱽᦩ⦹ᩡ݅. ⎹Ⓧญ✙⢽໕ᱩᔎ໕᮹ນᝁእᱥ᦭Ł ญ᷹ᮡᯕḥ⪵s᮹ᮁ⬉ᖒᮥ׳ᯕʑ᭥⦽⯩ᜅ☁əఉᜅ✙౩⋎, 16×16 mask ᔾᖒᮥ☖⦽ḡᩎᱢOtsu ᦭Łญ᷹, ᯵ᩍᇥḥၰ

ŉᰍ॒ᮝಽᯙ⦽יᯕᷩෝᱽÑ⦹ʑ᭥⦽ᩕฝᩑᔑᮝಽᯕ൉ᨕḡ ໑20ᩍ}᮹ᵝŲၰəฝᯱaᯩ۵᳑Õᨱᕽᝅ⨹ᮥᝅ᜽⦹ᩍ

ນᝁእᱥ᦭Łญ᷹᮹⣩ḩĞĥsᮥࠥ⇽⦹ᩡ݅. đŝsᮡ༉ࢱ

▮ᜅ✙❭ᯝಽ☖⧊ᗭ⥥✙ᭉᨕ᪡࠺ʑ⪵ࡹࠥಾᖅĥ⦹ᩡᮝ໑

(9)

▮ᜅ✙❭ᯝᨱ۵⣩ḩšญ᮹༉ुᱶᅕॅᮥ⡍⧉☁ಾǍᖒ⦹ᩡ݅.

ᱽᦩࡽ⣩ḩᱽᨕ ᜽ᜅ▽ᮡ96.6%᮹ ᱶ⪶ࠥෝw⇵ᨩᮝ໑ᯕ۵

ᖅĥ ʑᵡ⊹ 95%ෝ ᔢ⫭⦹۵ อ᳒ᜅ్ᬕ đŝෝ ᅕᯙ݅Ł ⧁

ᙹ ᯩ݅. ☖⧊ ⥥ಽəఉᮥ ☖⧕ᰆእ᮹ ↽ᱢ ᯕ࠺Ğಽ᪡ ႊ⨆, ᗮࠥ, ᪅₉, ⣩ḩšญđŝs॒ᮥ⦹ӹ᮹ॵᜅ⥭౩ᯕᨱᕽ⪶ᯙ

⧁ ᙹ ᯩᮝ໑ Ğಽ ᯕ࠺᜽ ᰆእ᮹ ⨩ᬊ᪅₉ 2cm ᯕԕಽ ᰆእ

ᬕᬊᯕࢉᮥ᦭ᙹᯩᨩ݅. ᯕ۵ᰆእᬕᬊᨱᯩᨕ᳦᳑ᯱ᮹⧊ญᱢᯙ

❱݉ᮥ ࠥᬙ ᙹ ᯩ۵ ∊ᇥ⦽ ᱶၡࠥෝ w⇵ᨩ݅Ł ᅝ ᙹ ᯩ݅.

ᅙםྙᨱᕽᱽᦩ⦽᦭Łญ᷹ᮥၵ┶ᮝಽ⦽⣩ḩᱽᨕ᜽ᜅ▽ᯕ

Ǎ⇶ࢉᮝಽᕽᅙᩑǍ᮹Ǣɚᱢ༊⢽ᯙ☖⧊ᗭ⥥✙ᭉᨕ༉ऩᮥ

w⇵íࡹᨩ݅. ⨆⬥ᯕ్⦽Õᖅᯱ࠺⪵ᇥ᧝᮹ᙹ᫵۵Йᵡ⯩

᷾a⧁äᮝಽʑݡࡹ໑ᅙםྙᨱᕽᩑǍ⦹ᩡ޹⣩ḩšญ᜽ᜅ▽

ŝ☖⧊ᗭ⥥✙ᭉᨕ۵əᙹ᫵ᨱ⦥ᙹᇩađ⦽᳕ᰍᯕʑᨱ݅᧲⦽

ᩑǍॅᯕḥ⧪ࢁäᯕ௝ᩩᔢࡽ݅. ᅙᩑǍෝ☖⧕݅᧲⦽⩶┽᮹

Õᖅᰆእᨱᱢᬊࢁᙹᯩ۵☖⧊⣩ḩšญℕĥෝᱽŖ⧁ᙹᯩᮥ

äᮝಽ ʑݡࡽ݅.

⨆⬥ᅙםྙᨱᕽᱽᦩ⦽ԕᬊᮥၵ┶ᮝಽ݅᧲⦽ॵḡ▙ᩢᔢ⃹

ญႊჶ᮹ᱢᬊᮥ☖⦽ນᝁእᱥ᦭Łญ᷹}ᖁႊᦩᨱݡ⦽ḡᗮᱢ ᯙ ᩑǍa ᫵Ǎࡹ໑ ☖⧊ ᗭ⥥✙ᭉᨕ᮹ ᩎ⧁ᨱ ᯩᨕ ນᝁእᱥ

⥥ಽəఉŝ ᄥࠥಽ ᖅĥࡹᨕ ▮ᜅ✙ ❭ᯝಽ ࠺ʑ⪵ ࡽ ᇡᇥᮥ

ນᝁእᱥ༉ऩಽᕽ☖⧊ᗭ⥥✙ᭉᨕᨱ⥭్əᯙ⩶┽ಽ᜽ᜅ▽

ᖅĥaࡽ݅໕᪥ᄞ⦽☖⧊ᗭ⥥✙ᭉᨕಽǍᖒࢁᙹᯩ݅Ł❱݉ࡹ

໑ ᯕෝ ☖⧕ ⨆⬥ ⥥ಽəఉe ☖⧊ ྙᱽෝ ޵ᬒ ᬱ⪽⯩ ⧕đ

⧁ ᙹ ᯩᮥ äᯕ݅.

References

Gonzalez, C. R. and Woods, R. E. (2002). Digital image processing, Pearson Education. pp. 595-611.

Haran, G. J., Dillenburg, J. and Nelson, P. (2006). “Real-time image processing algorithms for the detection of road and environ- mental conditions.” Proc. Of 9

^th

International Conference on App- lications of Advanced Technology in Transportation, ASCE, Illinois, Chicago, pp. 55-60.

Hryciw, D. R., Shin, S. and Jung, Y. (2006). “Soil image processing -single grains to particle assemblies.” GeoCongress 2006: Geo- Technical Engineering in the Information Technology Age, pp. 1-6.

Lee, S., Chang, L. and Skibniewski, M. (2006). “Automated recog- nition of surface defects using digital color image processing.”

Automation in Construction, Vol. 15, pp. 540-549.

Lee, W., Seo. J., Moon, S. and Lim, J. (2007). “Development of tele-operated equipment for concrete surface grinding.” KSCE Journal of Civil Engineering, Vol. 27, No. 6, pp. 741-748 (in Korean).

Leu, S. and Chang, S. (2005). “Digital image processing based ap- proach for tunnel excavation faces.” Automation in Construction, Vol. 14, pp. 750-765.

Nobuyuki Otsu (1975). “A threshold selection method from gray level histograms.” IEEE Trans. Sys. Man., Cyber, Vol. 9, No. 1, pp. 62-66.

Oh, J. T. and John, D. L. I. (2003). “Vehicle detection using video image processing system : Evaluation of PEEK Video Trak.”

Technical notes, Journal of Transportation Engineering ASCE, Vol. 129, pp. 462-465.

Yu, S., Jang, J. and Han, C. (2007). “Auto inspection system using a mobile robot for detecting concrete cracks in a tunnel.”

Machine Vision based Quality Management System for Tele-operated Concrete Surface Grinding Machine

Received April 24 2013, Revised May 14 2013, Accepted May 21 2013

Copyright ⵑ 2013 by the Korean Society of Civil Engineers

׌୨ฅ ȵ඿਎૴ ȵছஂ଀

Kim, Jeonghwan*, Phi, Seung Woo**, Seo, Jongwon***

Machine Vision based Quality Management System for Tele-operated Concrete Surface Grinding Machine

ABSTRACT

Key words : Machine vision, Concrete grinding, Image processing, GPS

Ⅹಾ

⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᮡ⡍ᰆ໕᮹י⪵ੱ۵❭ᗱᮝಽᯙ⦽ᅕᙹ᯲ᨦŝə൉ኺ(Grooving) ᜽Ŗᮥ☖⦽⡍ᰆ໕᮹႑ᙹ܆ಆᮥv⪵⦹Ñӹ⠪

┥ᖒᮥ⪶ᅕෝ᭥⦹ᩍᯱᵝᱢᬊࡹ۵Ŗჶᯕ݅. ə్ӹə᯲ᨦ✚ᖒᯕי࠺Ḳ᧞ᱢᯕŁᇥḥ, ᜍ్ḡ, ᗭᮭ॒ᮝಽᯙ⦽ᮁ⧕⦽᯲ᨦ⪹Ğᮥᅕᮁ

⦹Łᯩᮝ໑ᰆእෝ݅൉۵ʑ܆Ŗ᮹ᙺಉࠥᨱ঑௝ᔾᔑᖒၰᱩᔎ⣩ḩ᮹⠙₉aⓑĞ⨆ᯕᯩ݅. ঑௝ᕽᰆእ᳦᳑ᯱab᳦᭥⨹ᨱי⇽ࡹḡ

ᨦ᮹ḥ⃺॒ࠥᮥ᳦⧊ᱢᮝಽᔑ⇽⦹ᩍᬭⓍᜅ▭ᯕᖹᨱ⢽᜽⧉ŝ࠺᜽ᨱນᝁእᱥ᜽ᜅ▽ᨱ᮹⦽᯲ᨦ⣩ḩᱶᅕ᪡᮹☖⧊ᮥ᭥⦽⥥ಽəఉᮥ

}ၽ⦹ᩡᮝ໑, ⩥ᰆᱢᬊ▭ᜅ✙ෝ☖⧕ᅙʑᚁᮥá᷾⦹ᩡ݅.

áᔪᨕ ນᝁእᱥ, ⎹Ⓧญ✙ᱩᔎ, ᩢᔢ⃹ญ, GPS

  ֨ėěν

Fig. 1. Integrated Operation Program Interface

1. ᕽು

⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᮡ⡍ᰆ໕᮹י⪵ੱ۵❭ᗱᮝಽᯙ⦽

ᅕᙹ᯲ᨦŝə൉ኺ(Grooving) ᜽Ŗᮥ☖⦽⡍ᰆ໕᮹႑ᙹ܆ಆᮥ

v⪵⦹Ñӹ⠪┥ᖒᮥ⪶ᅕෝ᭥⦹ᩍᯱᵝᱢᬊࡹ۵Ŗჶᯕ݅. ə్

ӹʑ᳕᮹⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᮡᱩᔎ᯲ᨦ᜽ၽᔾࡹ۵ᇥḥᯕ ӹᜍ్ḡಽᯙ⦹ᩍๅᬑᩕᦦ⦽᯲ᨦ⪹Ğᨱᕽᯕ൉ᨕḡŁᯩ۵

⩥ᝅᯕ݅. ޵ᇩᨕ᯲ᨦᯱ᮹ᙺಉᱶࠥᨱ঑௝᯲ᨦ⣩ḩ᮹⠙₉a

⍅⣩ḩšญᨱᨕಅᬡᯕ঑෕໑᜖᜾ᱩᔎ᮹Ğᬑၽᔾࡹ۵ᜍ్ḡ ᮹⃹ญᨱ᮹⦽⪹Ğᱢྙᱽၰእ⬉ᮉᱢᯙ᯲ᨦᮝಽᯙ⦽eᱲᱢᯙ

ᗱ⧕ၽᔾ᮹ᬑಅaᯩ݅. ᯕ్⦽ᮁ⧕⦽᯲ᨦ⪹Ğᨱᕽၽᔾ⦹۵

݅᧲⦽⦝⧕ෝ↽ᗭ⪵⦹ʑ᭥⦹ᩍ᯲ᨦᯱෝ⩥ᰆᨱᕽມญਉᨕḥ

Ŕᨱ ᭥⊹᜽┅໕ᕽᱩᔎ᯲ᨦᮥ ᪅₉ ᨧᯕᙹ⧪⧁ ᙹ ᯩᮥᐱอ

ᦥܩ௝ᰆእ᳦᳑ᯱ᮹᮹ᔍđᱶᨱݡ⦽ḡᬱʑ܆ᮥw⇹ᬱĊ᳦᳑

⎹Ⓧญ✙ ⢽໕ᱩᔎ ʑĥ᮹ }ၽᯕ ⦥᫵⦹݅.

ᯕ్⦽ ᰆእ᮹ }ၽᮥ ᭥⧕ᕽ۵ ⢽໕ᱩᔎ ᯲ᨦ᮹ ↽ᱢ Ğಽ

ĥ⫮, ᜽Ŗᱶᅕၰᰆእ᮹ᔢ┽ᱶᅕᐱอᦥܩ௝ᬱĊᨱᕽᱩᔎ

⣩ḩᮥ⊂ᱶ· ❱݉⧁ᙹᯩ۵᜽ᜅ▽ŝᬱĊ᳦᳑ᮥ᭥⦽᯲ᨦᱶᅕ

☖⧊ ᜽ᜅ▽ Ǎ⇶ᯕ ⦥ᙹᱢᯕ݅. ᅙ ᩑǍᨱᕽ۵ ᯕ్⦽ ༊ᱢŝ

⦥᫵ᖒᮝಽນᝁእᱥᮥ⪽ᬊ⦽⣩ḩᱽᨕ᪡ᰆእᬕᬊᨱᯩᨕ᫵Ǎ

ࡹ۵༉ुᱶᅕ᪡ᯕෝၵ┶ᮝಽ᳦᳑ᯱ᮹᮹ᔍđᱶ, ᷪᰆእ᳦᳑

໦ಚᮥԕตᙹᯩ۵᯲ᨦᱶᅕ☖⧊ᬕᬊ᜽ᜅ▽ᮥᱽ᜽⦹ᩡ݅.

ᯕෝ᭥⧕ນᝁእᱥ᜽ᜅ▽ᮥ⪽ᬊ⦽ᔍಡᩑǍෝᇥᕾ⦹ᩍ↽ᱢ᮹

ᩢᔢ⃹ญ᦭Łญ᷹ᮥ}ၽ⧉ŝ࠺᜽ᨱ, Łᱶၡ᮹᭥⊹⊂ᱶʑᚁᯙ

RTK(Real-Time Kinematic) GPS ᪡᮹ᩑĥෝ☖⧕Ğಽၰ᯲ᨦ

⣩ḩđŝෝ᳦⧊ᱢᮝಽ⢽⩥⧁ᙹᯩ۵⦹ऽᭉᨕᖁᱶၰ⥥ಽəఉ

}ၽᮥᙹ⧪⦹ᩡ݅. ੱ⦽᜽ᜅ▽☖⧊݉ĥᨱᕽ⥥ಽəఉe࠺ʑ⪵

ྙᱽෝ⧕đ⦹ʑ᭥⦹ᩍ ⥥ಽ☁⎽ᮥᱶ᮹⦹ᩡᮝ໑⩥ᰆᝅ⨹ᮥ

☖⧕ ᖒ܆ᮥ á᷾⦹ᩡ݅.

2. ᬱĊ᳦᳑⎹Ⓧญ✙⢽໕ᱩᔎᰆእ}ၽ

2.1 ෇݁ଆઘऀࢂ

⎹Ⓧญ✙⢽໕ᱩᔎᰆእ᮹༉ℕ۵⡎110cm, ʙᯕ۵210cm,

׳ᯕ۵115cm, ⅾᵲప1tonᮝಽǍᖒ⦹ᩍᗭ⩶✙౎ᨱࠥᱢᰍa

a܆☁ಾ⦹ᩡᮝ໑, ॵᲅᨵḥ᮹⇽ಆᮥᯕᬊ⦹ᩍ⎹Ⓧญ✙⢽໕ᮥ

ᱩᔎ⦹۵ 65cm ⡎᮹ Õ᜾ ݅ᯕᦥ།ऽ ə௝ᯙ޵ෝ ⫭ᱥ᜽┅໑, 24V ႑░ญಽǍ࠺ࡹ۵2}᮹⬥ල, ᳑⨆ᮥ᭥⦽ᱥලᰆ⊹, ᰆእ

׳ԏᯕ᳑ᱩᮥ᭥⦽ᅖ࠺᜾ᮁᦶᝅฑ޵ෝ⡍⧉⦽݅. ੱ⦽2}᮹

GPS ᙹᝁʑෝᯕᬊ⦽ႊ⨆❭ᦦ(GPS ӹ⋉ၹ), b᳦ྕᖁᘂᙹᝁ

༉ऩ, ⋕ີ௝༉ऩ, ⎹Ⓧญ✙ᇥḥᮥ⯂᯦⦹ʑ᭥⦽Ḳḥᰆ⊹a

⇵a ᰆ₊ࡽ݅(Lee ॒, 2006).

2.2 ীඹൈଆઘऀࢂ

⎹Ⓧญ✙⢽໕ᱩᔎ᯲ᨦᮥᬱĊᨱᕽᙹ⧪⦹ʑ᭥⦹ᩍ2.1ᱩᨱᕽ

ᗭ}⦽⦹ऽᭉᨕෝᱽᨕ⦹Łᬕᬊ⧁ᙹᯩ۵ᗭ⥥✙ᭉᨕ᮹}ၽᯕ

⦥ᙹᱢᯕ݅. ᯕᨱ ᅙ ᩑǍᨱᕽ۵ MFC (Microsoft Foundation Class)ෝᯕᬊ⦹ᩍ⥥ಽəఉᮥ}ၽ⦹ᩡᮝ໑, ⥥ಽəఉ᮹⧖ᝍᱢ ᯙ ʑ܆ŝ ə ༊⢽۵ ᯲ᨦ ᙹ⧪ᯱಽ ⦹ᩍɩ ༉ु ᯲ᨦŝ ᯲ᨦ

ᱶᅕa ☖⧊ ᗭ⥥✙ᭉᨕᨱᕽ ᙹ⧪ ၰ ⢽⩥ࡹࠥಾ ⦹ᩡ݅. ੱ⦽

ᷪ, Ğಽĥ⫮, ⣩ḩšญ, ᰆእᱽᨕ᪡šಉࡽ༉ु᯲ᨦᱶᅕa⦽

⪵໕ᨱ⢽⩥ࡽ݅. Fig. 1ᨱᕽⴗᮡ᯲ᨦĞಽ᪡⩥ᰍʑĥᰆእ᮹

᭥⊹ෝ, ⴘ۵ ḥ⧪ ႊ᭥bᮥ, ⴙᮡ ⩥ᰍ ᗮࠥ᪡ ĥ⫮ Ğಽᨱᕽ

ჸᨕӽÑญ॒ᮥ, ⴚ۵ນᝁእᱥʑၹ⣩ḩᱽᨕđŝෝ⢽⩥⦹໑, ⴛ۵᜽ᜅ▽ᖅᱶᮥ⧁ᙹᯩíࡹᨕᯩ݅. ☖⧊ᗭ⥥✙ᭉᨕᨱᕽ

⣩ḩšญᨱݡ⦽ᱶᅕෝ⢽⩥⦹۵ⴚ᮹ʑ܆ᮡᱩᔎ᯲ᨦđŝෝ

Ğಽᨱ঑௝ʑᵡ⊹ᯕᔢ/ᯕ⦹ᩍᇡෝॵᜅ⥭౩ᯕ⧁ᙹᯩ۵ʑ܆ᮥ

w⇵ᨕ᧝⦽݅. ݅ᮭᰆᨱᕽ۵ᱩᔎ⢽໕᮹⊂ᱶႊ᜾ᨱݡ⦽á☁ෝ

☖⧕ນᝁ እᱥ ʑᚁᮥ⣩ḩšญ ႊ᜾ᨱ ᖁᱶ⦽ᯕᮁᨱ š⦹ᩍ

ʑᚁ⦹ࠥಾ ⦹ā݅.

3. ⎹Ⓧญ✙ᱩᔎ⢽໕᮹⣩ḩ⊂ᱶႊჶ

ᬱĊ᳦᳑⎹Ⓧญ✙⢽໕ᱩᔎᰆእ᮹ĞᬑᰆእᬕᱥᯱaᬱÑญ ᨱᕽᰆእෝ ᱽᨕ⦹ʑ ভྙᨱᱩᔎࡽ ᯲ᨦ᮹ ⣩ḩᮥ⊂ᱶ⧁ ᙹ

ᨧᮝအಽᬱÑญᨱᕽ᯲ᨦࡽᱩᔎ⣩ḩᮥ⠪a⧁ᙹᯩ۵⣩ḩšญ

᜽ᜅ▽ᯕ᫵Ǎࡽ݅. ᬱÑญᨱᕽᱩᔎ໕ᨱݡ⦽⣩ḩᮥ⊂ᱶ⦹ʑ

᭥⧕ᕽ ݅ᮭ Table 1ŝ zᯕ ᱢᬊ a܆⦽ ⣩ḩ ⊂ᱶ ʑᚁॅᮥ

á☁⦹ᩡ݅.

bႊ᜾ᄥᰆ݉ᱱၰᝅԕᖒ܆ᝅ⨹ᇥᕾđŝ᯦ℕᩢᔢ᜽ᜅ▽ᮥ

ᔍᬊ⦹۵äᯕaᰆ୑ᨕӽᖒ܆ᮥၽ⭹⦹ᩡᮝӹ⩥ᰍᔢᬊ⪵ࡹᨕ

ᯩ۵᯦ℕᩢᔢ᜽ᜅ▽ᮡࠥಽᗭᖒᄡ⩶⊂ᱶᮥ᭥⧕ᝅ᫙ᔍᬊᨱ

Table 1. Quality Control Method

Method Advantage Disadvantage

3D Camera Easy Implementation GRequires a Target/Pattern Laser

Scanner High Precision GRequires Some Time to Measure

׌୨ฅȵ඿਎૴ȵছஂ଀

Kim, Jeonghwan, Phi, Seung Woo, Seo, Jongwon

áᔪᨕ ນᝁእᱥ, ⎹Ⓧญ✙ᱩᔎ, ᩢᔢ⃹ญ, GPS

֨ėěν