Human identification and detection through visual processing
To identify a person through visual processing, an image capturing device is used to capture a static image of the person before transmitting it to the visual processing system. The system then matches the input against a database of identified images. Facial recognition is a two-dimensional problem...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
2009
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/17123 |
| _version_ | 1826120360224882688 |
|---|---|
| author | Chia, Kian An. |
| author2 | Ng Geok See |
| author_facet | Ng Geok See Chia, Kian An. |
| author_sort | Chia, Kian An. |
| collection | NTU |
| description | To identify a person through visual processing, an image capturing device is used to capture a static image of the person before transmitting it to the visual processing system. The system then matches the input against a database of identified images. Facial recognition is a two-dimensional problem whereby its limitations of the processed images are to be taken in frontal view and in fixed lighting conditions. An approach to the problem was to develop a near-real-time face recognition system which extracts the subject’s head before recognizing the person by comparing the characteristics of the face to that of the known individuals. The face images are projected into the feature space “face space”, defined by Eigenfaces, consisting of the eigenvectors of the face, not necessarily corresponding to the isolated features such as eyes, ears, and noses. The framework provides the ability to learn to recognize new faces in an unsupervised manner. Faces are recognized by using the Euclidean distance between the Eigenvectors of the captured image and the known images in the database.
For evaluation, experiments are conducted to test the accuracy of the facial recognition algorithm under various threshold values. Similar experiments were conducted again to analyze the effectiveness of preprocessing techniques on the accuracy of the system. Results show that the system has a lower false acceptance rate using low threshold value and lower false rejection rate using high threshold value; and not all preprocessing techniques used yield better results. |
| first_indexed | 2024-10-01T05:15:31Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/17123 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T05:15:31Z |
| publishDate | 2009 |
| record_format | dspace |
| spelling | ntu-10356/171232023-03-03T20:52:50Z Human identification and detection through visual processing Chia, Kian An. Ng Geok See Quek Hiok Chai School of Computer Engineering Centre for Computational Intelligence DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision To identify a person through visual processing, an image capturing device is used to capture a static image of the person before transmitting it to the visual processing system. The system then matches the input against a database of identified images. Facial recognition is a two-dimensional problem whereby its limitations of the processed images are to be taken in frontal view and in fixed lighting conditions. An approach to the problem was to develop a near-real-time face recognition system which extracts the subject’s head before recognizing the person by comparing the characteristics of the face to that of the known individuals. The face images are projected into the feature space “face space”, defined by Eigenfaces, consisting of the eigenvectors of the face, not necessarily corresponding to the isolated features such as eyes, ears, and noses. The framework provides the ability to learn to recognize new faces in an unsupervised manner. Faces are recognized by using the Euclidean distance between the Eigenvectors of the captured image and the known images in the database. For evaluation, experiments are conducted to test the accuracy of the facial recognition algorithm under various threshold values. Similar experiments were conducted again to analyze the effectiveness of preprocessing techniques on the accuracy of the system. Results show that the system has a lower false acceptance rate using low threshold value and lower false rejection rate using high threshold value; and not all preprocessing techniques used yield better results. Bachelor of Engineering (Computer Science) 2009-06-01T01:08:29Z 2009-06-01T01:08:29Z 2009 2009 Final Year Project (FYP) http://hdl.handle.net/10356/17123 en Nanyang Technological University 58 65 p. application/pdf |
| spellingShingle | DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision Chia, Kian An. Human identification and detection through visual processing |
| title | Human identification and detection through visual processing |
| title_full | Human identification and detection through visual processing |
| title_fullStr | Human identification and detection through visual processing |
| title_full_unstemmed | Human identification and detection through visual processing |
| title_short | Human identification and detection through visual processing |
| title_sort | human identification and detection through visual processing |
| topic | DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision |
| url | http://hdl.handle.net/10356/17123 |
| work_keys_str_mv | AT chiakianan humanidentificationanddetectionthroughvisualprocessing |