Design and Implementation of an Active Camera Motion Controller for Multiple Objects Recognition and Tracking
碩士 === 國立彰化師範大學 === 電機工程學系 === 95 === This thesis uses image processing approaches to design methods for a static camera that recognizes multiple moving targets. Active camera tracking methods for identifying a specified target from multiple moving targets are studied. The adopted image processing a...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2007
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/14536188006762708054 |
| Summary: | 碩士 === 國立彰化師範大學 === 電機工程學系 === 95 === This thesis uses image processing approaches to design methods for a static camera that recognizes multiple moving targets. Active camera tracking methods for identifying a specified target from multiple moving targets are studied. The adopted image processing approach includes a noise filter, image differential, binary imaging and morphological operations for following and labeling multiple target segmentation. The geometrical and texture features of the segmented targets are determined and compared with a template to identify the specified target. The template matching is implemented using a new three-step search (N3SS) procedure that has the ability to reduce the search time. When the specified target is recognized, its location difference from the image center is used to control the camera pan and tilt motor drive to aim at the specified target. The experimental results from a static camera recognizing moving multiple targets compared against an active camera tracking a specified target isolated from multiple moving targets are used to illustrate the proposed methods’ performance.
|
|---|