Research And Application Of Image-based Cable Geometric Parameters Detection Algorithm

With the rapid development of China’s economy,electrical energy has played an increasingly important role in the industrial and civil fields.As a carrier for transmitting electrical energy,the demand for electrical cables is increasing.Due to manufacturing technology,market competition,and other reasons,cables often have quality problems such as sub-standard ac withstand voltage,poor mechanical elongation,and irrational structural dimensions.The use of inferior cables will not only bring economic losses to power enterprises and users,but also cause huge security risks to the operation of the power grid.Therefore,it is of great significance to perform fast and reliable detection of cable quality.The main factors that determine cable quality are electrical performance,mechanical performance,thermal performance and so on.Among them,the geometrical parameters of the cable are one of the important indicators that determine the quality of the cable.At present,the detection of cable geometric parameters mainly relies on manual detection,which is time-consuming and labor-intensive.However,the conventional image-based geometric detection methods are not suitable for cable parameters detection with complex structures.Therefore,design accurate robust cable geometry parameter detection algorithm for cable quality automatic detection has very important practical significance and application value.This paper mainly researches the detection of cable geometric parameters from the following aspects:(1)Cable area segmentation.Based on the in-depth analysis of the cable structure composition,the preliminary segmentation of the cable insulation layer is realized by super pixel segmentation,k-means clustering,and bottom-up hierarchical clustering.Aiming at the problem that it is difficult to determine the hyperparameters in the precise segmentation of the insulation layer,a region segmentation method based on the turning point threshold boundary is proposed.The precise segmentation of the insulation layer is completed by calculating adaptive parameters,and then the conductor region and the sheath region are segmented.(2)Cable contour extraction.Aiming at the problem that the edge extraction algorithm cannot extract the pixel-level contours of the target area well,a precise contour extraction method based on candidate point merge search is proposed,the pixel-level contours are extracted by filtering and combining contour candidate points.(3)Development of cable geometric parameter detection software.Designed the image-based cable geometric parameter detection software.The software has the functions of system management,code management,manufacturer information management,cable information management,manufacturer’s cable information management,cable inspection and query statistics.