摘要
为了更好地分析潜土逆转旋耕刀的抛土性能,用两台高速摄像机组成双目立体视觉系统对潜土逆转旋耕机的抛土过程进行同步记录,然后跟踪图像序列中同一土粒的运动轨迹。提出了用BP神经网络建立双目立体视觉模型的方法,利用此模型进行标定,找出物体的图像坐标与世界坐标之间的映射关系,解决了土粒三维运动轨迹的提取问题。
In order to analyze the performance of submerged reverse-rotary blade, processing of the motion of clods of submerged reverse-rotary tiller was recorded by the binocular vision system consisted of two high speed imaging systems ; function relation between the image coordinates and the world coordinates of the object with BP neural network were found, then 3-D moving trace was restored by using the trained network.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2006年第6期98-101,共4页
Transactions of the Chinese Society for Agricultural Machinery
基金
原机械工业部重点学科专项资助项目(项目编号:95054)
关键词
潜土逆转旋耕
神经网络
运动轨迹
标定
Submerged reverse-rotary tillage, Neural network, Moving trace, Calibration
