A Novel Face Detection Technique Using Depth Information and Active Shape Models
碩士 === 開南大學 === 資訊管理學系 === 101 === In computer vision, many methods employ color images to detect human faces. But it is easy affected by surrounding factors and brightness. Then the object of face positions may errors. Recently, Microsoft has presented an interaction sensor which called Kinect. It...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2013
|
Online Access: | http://ndltd.ncl.edu.tw/handle/85507086155236096616 |
id |
ndltd-TW-101KNU00396010 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-101KNU003960102015-10-13T22:18:46Z http://ndltd.ncl.edu.tw/handle/85507086155236096616 A Novel Face Detection Technique Using Depth Information and Active Shape Models 使用深度資訊與主動式形狀模型之人臉定位技術 Ting-Wen Chen 陳鼎文 碩士 開南大學 資訊管理學系 101 In computer vision, many methods employ color images to detect human faces. But it is easy affected by surrounding factors and brightness. Then the object of face positions may errors. Recently, Microsoft has presented an interaction sensor which called Kinect. It can output the range between object and sensor from infrared ray controller to produce the depth image. The depth image against most of surrounding brightness, and be used to increase the accuracy in object recognition. In this paper, we employ Kinect to capture color and depth image. After then we detect the face blocks by ASM. When faces were detected, we compare the face object size between detection and table we count distance in depth image. Finally, remove the error blocks to increase the detection accuracy from ASM. Experiment result shows our method not only removes the error blocks successfully, but also makes ASM detection higher accuracy. And we proposed the table that count distance will be used in future work. Kuo-Ming, Hung 洪國銘 2013 學位論文 ; thesis 38 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 開南大學 === 資訊管理學系 === 101 === In computer vision, many methods employ color images to detect human faces. But it is easy affected by surrounding factors and brightness. Then the object of face positions may errors. Recently, Microsoft has presented an interaction sensor which called Kinect. It can output the range between object and sensor from infrared ray controller to produce the depth image. The depth image against most of surrounding brightness, and be used to increase the accuracy in object recognition.
In this paper, we employ Kinect to capture color and depth image. After then we detect the face blocks by ASM. When faces were detected, we compare the face object size between detection and table we count distance in depth image. Finally, remove the error blocks to increase the detection accuracy from ASM.
Experiment result shows our method not only removes the error blocks successfully, but also makes ASM detection higher accuracy. And we proposed the table that count distance will be used in future work.
|
author2 |
Kuo-Ming, Hung |
author_facet |
Kuo-Ming, Hung Ting-Wen Chen 陳鼎文 |
author |
Ting-Wen Chen 陳鼎文 |
spellingShingle |
Ting-Wen Chen 陳鼎文 A Novel Face Detection Technique Using Depth Information and Active Shape Models |
author_sort |
Ting-Wen Chen |
title |
A Novel Face Detection Technique Using Depth Information and Active Shape Models |
title_short |
A Novel Face Detection Technique Using Depth Information and Active Shape Models |
title_full |
A Novel Face Detection Technique Using Depth Information and Active Shape Models |
title_fullStr |
A Novel Face Detection Technique Using Depth Information and Active Shape Models |
title_full_unstemmed |
A Novel Face Detection Technique Using Depth Information and Active Shape Models |
title_sort |
novel face detection technique using depth information and active shape models |
publishDate |
2013 |
url |
http://ndltd.ncl.edu.tw/handle/85507086155236096616 |
work_keys_str_mv |
AT tingwenchen anovelfacedetectiontechniqueusingdepthinformationandactiveshapemodels AT chéndǐngwén anovelfacedetectiontechniqueusingdepthinformationandactiveshapemodels AT tingwenchen shǐyòngshēndùzīxùnyǔzhǔdòngshìxíngzhuàngmóxíngzhīrénliǎndìngwèijìshù AT chéndǐngwén shǐyòngshēndùzīxùnyǔzhǔdòngshìxíngzhuàngmóxíngzhīrénliǎndìngwèijìshù AT tingwenchen novelfacedetectiontechniqueusingdepthinformationandactiveshapemodels AT chéndǐngwén novelfacedetectiontechniqueusingdepthinformationandactiveshapemodels |
_version_ |
1718075674247299072 |