Human posture and action recognition
Digital image processing is a very interesting subject. Although it is a relatively new field in technology, digital image processing has impacted the everyday lives of people all around the world. From Adobe® Photoshop© to the use of webcams, digital image processing has been used by people from al...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
2009
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/16942 |
| _version_ | 1826121309164142592 |
|---|---|
| author | Lau, Melvin Eng Heok. |
| author2 | Maylor Karhang Leung |
| author_facet | Maylor Karhang Leung Lau, Melvin Eng Heok. |
| author_sort | Lau, Melvin Eng Heok. |
| collection | NTU |
| description | Digital image processing is a very interesting subject. Although it is a relatively new field in technology, digital image processing has impacted the everyday lives of people all around the world. From Adobe® Photoshop© to the use of webcams, digital image processing has been used by people from all walks of life, no matter if these people are professionals or not.
However, there still exist many unexplored territories in digital image processing, especially with the techniques of pattern recognition. Therefore, it was the decision to base this project off MATLAB© 7.4, an engineering software which is commonly used for digital image processing applications. Learning to use MATLAB© was a very difficult task, but it was proven to be a good choice of software as it provided many powerful tools for image processing and had a very big and established online community to provide support.
In this project, a couple of pattern recognition techniques was studied and slightly altered as a part of the exploration of a new pattern recognition technique, which was also used for human posture recognition. In this new technique, the author attempted to take an image pose and break it up into 4 different parts. Features, such as Zernike moments, were then extracted from these parts to be used for the recognition process.
Tests were carried out in different test phases and in a modular manner. These test phases were mainly designed to ensure that the various components of the system worked individually, and also worked when integrated as a whole system. They were also designed to test the feasibility of this new technique, whether it was feasible or not to implement such a technique. Current results were satisfactory, but various problems, such as high processor utilisation, were identified when using this technique as well. The author feels that more improvements can be applied onto the project for better results. |
| first_indexed | 2024-10-01T05:30:32Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/16942 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T05:30:32Z |
| publishDate | 2009 |
| record_format | dspace |
| spelling | ntu-10356/169422023-03-03T20:54:17Z Human posture and action recognition Lau, Melvin Eng Heok. Maylor Karhang Leung School of Computer Engineering Forensics and Security Lab DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision Digital image processing is a very interesting subject. Although it is a relatively new field in technology, digital image processing has impacted the everyday lives of people all around the world. From Adobe® Photoshop© to the use of webcams, digital image processing has been used by people from all walks of life, no matter if these people are professionals or not. However, there still exist many unexplored territories in digital image processing, especially with the techniques of pattern recognition. Therefore, it was the decision to base this project off MATLAB© 7.4, an engineering software which is commonly used for digital image processing applications. Learning to use MATLAB© was a very difficult task, but it was proven to be a good choice of software as it provided many powerful tools for image processing and had a very big and established online community to provide support. In this project, a couple of pattern recognition techniques was studied and slightly altered as a part of the exploration of a new pattern recognition technique, which was also used for human posture recognition. In this new technique, the author attempted to take an image pose and break it up into 4 different parts. Features, such as Zernike moments, were then extracted from these parts to be used for the recognition process. Tests were carried out in different test phases and in a modular manner. These test phases were mainly designed to ensure that the various components of the system worked individually, and also worked when integrated as a whole system. They were also designed to test the feasibility of this new technique, whether it was feasible or not to implement such a technique. Current results were satisfactory, but various problems, such as high processor utilisation, were identified when using this technique as well. The author feels that more improvements can be applied onto the project for better results. Bachelor of Engineering (Computer Engineering) 2009-05-29T02:17:13Z 2009-05-29T02:17:13Z 2009 2009 Final Year Project (FYP) http://hdl.handle.net/10356/16942 en Nanyang Technological University 68 p. application/pdf |
| spellingShingle | DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision Lau, Melvin Eng Heok. Human posture and action recognition |
| title | Human posture and action recognition |
| title_full | Human posture and action recognition |
| title_fullStr | Human posture and action recognition |
| title_full_unstemmed | Human posture and action recognition |
| title_short | Human posture and action recognition |
| title_sort | human posture and action recognition |
| topic | DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision |
| url | http://hdl.handle.net/10356/16942 |
| work_keys_str_mv | AT laumelvinengheok humanpostureandactionrecognition |