面向机械动力学运动的机器人视觉去雾系统研究

Research on Robot Vision Dehazing System for Mechanical Dynamic Motion

从机器人系统动力学运动的角度来看,视觉感知在机器人和环境之间的智能互动中发挥着越来越重要的作用。由此,提出一种面向机械动力学运动的机器人视觉去雾系统。首先介绍了视觉反馈控制对于机器人机械动力学运动的重要性,然后基于大气散射模型,采用图像分割方法在天空区域中估计全局大气光值,并建立雾图像和透射率之间的映射关系,设计了多尺度卷积神经网络(Multi Scale Convolutional Neural Network,MS-CNN)优化场景透射率。实验结果表明,面向机械动力学运动的机器人视觉去雾方法的结果更接近于清晰图像,该方法的均方误差(Mean-Square Error,MSE)值低,表明所提出算法的失真小,图像内容更接近原始图像。

From the perspective of robot system dynamics,visual perception plays an increasingly important role in the intelligent interaction between robots and the environment.Firstly,the importance of visual feedback control for robot mechanical dynamics is introduced.The image segmentation method is used to estimate the global atmospheric light value in the sky area,and the mapping relationship between the fog image and the transmittance is established.The Multi Scale Convolutional Neural Network(MS-CNN)network is designed to optimize the scene transmittance.The experimental results and analysis show that the results of the robot vision dehazing method in this paper are closer to the clear image.The Mean-Square Error(MSE)value of the dehazing method in this paper is low,which indicates that the distortion of the proposed algorithm is small and the image content is closer to the original image.