Monocular vision-based real-time measurement on overhead crane payload swing angle
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
万方数据
针对目前基于视觉的桥式起重机负载摆角测量方法实时性较低,且没有考虑由运动模糊导致的测量精度下降问题,将一个对旋转和倾斜不敏感的球形标识物安装在负载上,设计一种计算灰度概率分布特征余弦相似度的方法,结合卡尔曼滤波器实时跟踪标识物.提出一种基于背景灰度值的局部自适应分割算法,使模糊边缘变清晰.从感兴趣区域(region of interest,ROI)的中心向周围8个方向寻找灰度梯度最大的点作为模糊图像的边缘点,对这些点采用最小二乘法进行拟合确定标识物位置.通过几何关系模型计算负载摆角.实验结果表明,本文方法在图像模糊情况下的测量精度和实时性能更优.
To address the problems that the current vision-based overhead crane payload swing angle measurement methods have low real-time performance and the decrease of measurement accuracy due to motion blur is not considered,a spherical marker insensitive to rotation and tilt is mounted on the payload,a method is designed to calculate the cosine similarity of grayscale probability distribution characteristics,and a Kalman filter is combined to track the marker in real time.A local adaptive segmentation algorithm based on the background gray value is proposed to clarify the blurred edges.The points with the largest gray gradient in eight directions around the center of the region of interest(ROI)are found as edge points of blurred images,and these points are fitted by the least-squares method to determine the location of the marker.The swing angle of the payload is calculated by the geometric relationship model.The experimental results show that the measurement accuracy and real-time performance of the method are better in the case of image blur.
NETL
NSTL
万方数据
