摘要
为实现视觉目标可靠快速三维定位,设计了一种新型的基于DSP的嵌入式主动双目红外视觉系统。与普通的红外相机不同,该系统采用常规CCD相机加红外滤光片的新方法采集红外图像,通过红外滤光后的低噪声图像可以转化为低噪声二值化图像,以此提高目标定位的准确度。提出了离散化遍历视觉目标搜索算法,通过离散控制两个相机的旋转和倾斜运动,实现视觉目标快速主动遍历搜索。实验证明,该系统目标三维定位有良好的鲁棒性。
In order to realize three-dimensional object localization, an active binocular vision system with infrared filters was proposed. Different from custom infrared cameras, ordinary CCD cameras with infrared filters are used instead in the system for image acquisition. IR filtered images containing less noise could be transferred to binary images before object localization to improve accuracy. A global dis- crete-traversal search algorithm is proposed for active localization by controlling the pan-tilt movement of two cameras discretely. Experi- ments show that the system can localize objects robustly.
出处
《单片机与嵌入式系统应用》
2013年第3期7-10,共4页
Microcontrollers & Embedded Systems
基金
教育部博士点基金新教师基金资助项目(编号:1099110)
