Active Eye Detection for Autostereoscopic Display System

摘要：

Eye detection is a necessary component in an autostereoscopic display system. This paper validates the conditions and influencing factors of red-eye effect through experiment, and takes this phenomenon into real-time eye detection. In the paper, we analysis the influence of distance between infrared light source and camera optical axial, as well as the distance between the camera and the viewer. To deal with the various scales of candidate regions caused by different distances of viewers, invariant features, which are independent of candidate region size, are extracted to characterize the eye target. Support vector machine (SVM) is employed as the final classifier. The result shows that the approach works well on various distances eye detection.

关键词： Autostereoscopic Eye Detection Feature Extraction SVM

作者: Chunbo Chen Yuanqing Wang

作者单位: Department of Electronics Science & Engineering,Nanjing University, Nanjing 210093, China

会议类型: 国际会议

会议名称: 2007亚洲显示国际会议（AD07）

会议地点: 上海

会议语种:英文

页码: 2034-2038

在线出版日期: 2007-03-12（万方平台首次上网日期，不代表论文的发表时间）