基于Mask RCNN的目标识别与空间定位

Target Recognition and Spatial Location Based on Mask RCNN

在机械臂的自主抓取系统研究中,为了自动获取目标物体的空间位置,采用Kinect深度传感器采集RGB图像,利用改进的深度学习算法Mask RCNN对RGB图像上的目标进行识别与分割,并通过Kinect深度传感器模型,将二维图像坐标转换成三维空间坐标,对目标物体进行三维建模,达到空间定位的目的;通过大量数据训练的Mask RCNN算法,可以同时识别多种特征差异很大的目标物体,具有广泛的应用空间;经过实验表明,获得的目标物体的三维空间坐标较为准确,且受环境影响较小,对机械臂抓取系统的研究具有较为重要的意义。

In the research of autonomous grasping system of manipulator,in order to acquire the spatial coordinates of the object automatically,Kinect depth sensor is used to collect RGB image,improved depth learning algorithm Mask RCNN is used to recognize and segment the target on RGB image,and through the Kinect depth sensor model,the two-dimensional image coordinates are transformed into three dimensional space coordinates,and the object is modeled in three-dimensional to achieve the purpose of spatial positioning.Mask RCNN algorithm trained by a large amount of data can recognize many objects with different features at the same time,so it has wide application space.Experiments show that the three-dimensional coordinates of the target object are more accurate and less affected by the environment,it is of great significance to the research of the manipulator grasping system.