Research On Online Detection Algorithms Of Drip Irrigation Pipe Drilling Position Based On Machine Vision

In view of the unstable perforation quality in the production of drip irrigation pipe with inner labyrinth patches and the current situation of manual testing,this paper applies machine vision technology to the production of drip irrigation pipe.The position deviation of the drilling position is identified and measured through a series of image analysis and processing by industrial camera.Then the position deviation is fed back to the controller to control the drilling machine to adjust,so that the system can form a closed loop and realize online detection and feedback adjustment of the drilling position.In this paper,after the overall structure and workflow design of the detection system,the hardware selection and installation of the machine vision module are carried out.Then the online detection algorithms of drip irrigation pipe drilling position based on machine vision is studied emphatically.Based on the actual situation of visual identification of drip irrigation pipe,the identification process is divided into two major steps:rough identification and fine identification.The main contents are as follows:Firstly,in rough identification,a hole location identification algorithm based on LBP and CNN is proposed.In the framework of TensorFlow,the algorithm combines LBP texture feature extraction with CNN deep learning algorithm,and builds a mathematical model for rough identification of drip irrigation pipe drilling position.It can identify normal drilling position and three other abnormal drilling position by rough identification,and feedback the identification results to the controller to achieve rough adjustment of drilling position.Through comparative experiments,the effectiveness of hole location identification is verified.Secondly,to meet the requirement of accurate identification of drip irrigation pipe drilling position,an optimization algorithm for extracting ROI of drip irrigation pipe based on multiple depth extraction of illuminated white area and edge enhancement was proposed.The algorithm includes image preprocessing,preliminary extraction of ROI of drip irrigation pipe,extraction of ROI of illuminated white area and image edge enhancement,which provides a basis for accurate hole location identification.Thirdly,the precise detection algorithm of drip irrigation pipe drilling position is studied.Based on the Hough circle transformation algorithm,the identification of the center of drilling position is realized.The effectiveness of the proposed edge enhancement optimization algorithm for the identification of the center of drilling position is verified by comparative experiments.The identification of the front boundary of the drip head is completed by depth extraction ROI,image histogram equalization and Hough line transformation algorithm.Then the precise measurement from the center of the drip hole to the front boundary of the drip head has been completed.Finally,through a large number of comparative experiments,it is proved that the detection algorithm of drip irrigation pipe drilling position proposed in this paper has higher detection accuracy than the general detection algorithm.The average relative error of the hole position measurement is only 4.96%,which fully meets the requirements of closed loop control of punch.