基于最小化图像技术的带式输送机载物识别研究

Research on Belt Conveyors Loading Based on Minimized Image Recognition Technology

针对带式输送机载运物体CPU处理资源冗余问题,提出一种控制摄像机触发信号方法,利用机器视觉处理技术在带式输送机运动过程中识别物体的位置和旋转数据,并利用这些数据由机器人进行拣选和放置操作,进而实现最小化图像处理资源和重复数据删除以识别带式输送机上的物体。为了实现这些目标,并考虑到摄像机的识别范围、物体的大小和带式输送机的速度,尽量减少成像处理的CPU资源,同时为一个物体在摄像机的多个帧中被识别为独立物体的问题提供了解决方案。根据带式输送机速度和物体大小等系统条件,获得最适合图像处理的fps值,通过试验及仿真证明,在使用触发信号前0.4545 m/s的带式输送机速度和使用触发信号后0.450 m/s的带式输送机速度下,CPU资源的比较表明,使用机器视觉技术时,资源的使用量分别减少了31%和14%,充分证明了所提理论和方法的正确性。

Aiming at the problem of redundant processing resources of the CPU of the belt conveyor carrying objects,a method of controlling the camera trigger signal was proposed.The machine vision processing technology was used to identify the position and rotation data of objects during the movement of the belt conveyor,and these data were used by the robot to pick and place operations,thus realizing the minimization of image processing resources and data deduplication to identify objects on the belt conveyor.In order to achieve these goals,taking the recognition range of the camera,the size of the object,and the speed of the belt conveyor into account,the CPU resources for imaging processing were minimized as much as possible.At the same time,a solution was provided for the problem of an object being recognized as an independent object in multiple frames of the camera.Based on system conditions such as belt conveyor speed and object size,the most suitable fps value for image processing was obtained.Through experiments and simulations,it was proved that under the belt conveyor speed of 0.4545 m/s before using the trigger signal and 0.450 m/s after using the trigger signal,the comparison of CPU resources showed that using machine vision technology reduced respectively resource usage by 31%and 14%.The correctness of the proposed theory and method has been fully demonstrated.