Template Matching in Multiple View Using Homography
碩士 === 華梵大學 === 資訊管理學系碩士班 === 103 === Research and application in human-machine interaction(HMI)，hand recognition and tracking is one of the very important issue. Because of the tech of hand tracking and recognition, more and more product can be controlled with hand through image devices. And based...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2015
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/28528490705300064586 |
| Summary: | 碩士 === 華梵大學 === 資訊管理學系碩士班 === 103 === Research and application in human-machine interaction(HMI)，hand recognition and tracking is one of the very important issue. Because of the tech of hand tracking and recognition, more and more product can be controlled with hand through image devices. And based on the trend that price and size of image device keep falling and being smaller increasingly, the chance using multiple views to solve the problems of image processing will raise.
In this paper we take advantages of multiple views to process template matching. It’s easier to obtain the geometry between the camera and the target than using single view for template matching. Our experiment use homographic geometry to alter hand template for each cameras which have different angle of view in order to perform template matching in multiple views.
But this approach for altering template is easily influenced by corresponding point. So we find the corresponding point with epipolar line.
|
|---|