基于图象差的平面大范围视觉伺服控制

Image-error-based Planar Global Visual Servoing

为解决大范围偏差的控制问题 ,将期望图象按给定的角度间隔旋转 ,离线生成一系列子期望图象。比较实时采集图象与期望子图象间的差异程度可获取目标绕重心的旋转运动参数。结合图象求重心方法给出的平动参数 ,实现了在大范围偏差时迅速将摄象机调整到期望位姿。在期望位姿附近结合直接图象反馈方式 ,实现了基于图象差的平面大范围视觉伺服控制。

A scheme for large scale deviation control in a vision guided grasping task is presented. The desired image is rotated with a specific interval to generate a series of desired sub images off line. The rotational control parameter could be obtained by comparing the desired sub images with the real time sampled image. Combining the rotational parameters with the translational parameters obtained from image centroid, the camera could be rapidly regulated to the desired pose. Round the desired pose, the method is combined with the direct image feedback control scheme to realize the image error based planar global visual servoing.