Face detection using skin color and eigenface techniques for human-robot interaction
The main objective of the research undertaken here is to develop an automated face detection system to be implemented on a mobile robot for humanrobot interaction (HRI) applications. HRI has gained increasing attention in recent years due to escalating numbers of assimilation of robots into human li...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2005
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/4327/1/WongHweeLingMFKE2005.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The main objective of the research undertaken here is to develop an automated face detection system to be implemented on a mobile robot for humanrobot interaction (HRI) applications. HRI has gained increasing attention in recent years due to escalating numbers of assimilation of robots into human lives. Face detection is a vital aspect to sense human presence in a robotic environment. Despite the fact that research on face detection has long been carried out, development and implementation of face detection on mobile robot remains a great challenge due to the complexity of the robot itself. A HRI model is presented to represent a humanrobot relationship in the shared robotic environment. The work focuses in building a natural form one human to one robot HRI based on face detection. The main concern of the work is to develop face detection using multiple features to increase robustness of the system. The face detection comprises of a skin color detection based on various chrominance information of images using multiple color spaces. Later, the skin region is verified using crisp rules face verification. Different rules are applied at different robot behavior that accommodates the HRI. Where applicable, eigenface method is implemented for further verification. Later, the robot makes decision and reacts based on the face detection result using heuristic search on the decision tree. An adaptive weighting technique for the branches in the decision tree is presented to solve problem for various robot behaviors and noise in images. Graphical user interface (GUI) and text-to-voice ability are employed on the robot to support HRI. Through experiments, it appears that the robot can successfully detect human face and react meaningfully towards the face that has been detected. |
|---|