Research On Recognition Method Of Particles On Cable Based On Image Processing

It puts high demands on China's power supply and development in consideration of its large population,as well as huge volume and rich types of manufacturing and service industry.As the wire and cable products is indispensable in power supply,the quality of cable definitely affects the prosperity and level of Chinese power development.In the cable production process,conductor needs to be coated in order to keep insulated.The insulation process is the first step after the conductor processing,which has a profound impact on the electrical conductivity of the product and the final quality of the product.At present,the quality of the coating process of cable is controlled by the distribution of particles on sprayed cable surface,and the detection of particles on cable is carried out by combining manual detection with equipment detection.However,manual detection often means various errors and low detection efficiency.The detection equipment is mainly controlled by foreign companies and updated slowly,which increases the cost of cable production and reduces both efficiency and quality of cable production.With the rapid development of software and hardware in the computer field in recent years,the storage and computing of digital images have developed greatly.Correspondingly the digital image processing technology has also made significant progress and been applied in various scenarios and fields.Therefore,in this paper,we use image processing technology and machine learning to determine the distribution and size of particles on cable by detecting the digital image of sprayed cable.The main content of this paper is as follows:(1)The characteristics of cable particles digital images is analyzed and briefly described.The image preprocessing method is determined by comparing the effects of several image noise reduction methods on the cable particle image which has noise,clutter background,brightness distribution and other problems.According to the image characteristics of the cable particle image with obvious edges and lack of texture and color features,the edge detection operator is used to extract the feature s of the cable particle image.(2)A method is proposed for the detection and location of cable particles.Aiming at the irregularity distribution and numerous number of particles in cable particle image,this paper constructs the detection model of the cable s by the sliding window method and the support vector machine.Considering that the proportion of backgrounds in the image is considerable,and there are many unrelated information such as noise and “cloud” in the background,the watershed algorithm is used to extract the image foreground before the detection process.Use the non-maximum suppression algorithm to remove extra points after the detection and obtain the final detection result.(3)A method is proposed for detecting the particle size of cable particles.The markers-based watershed algorithm is used to segment the image to obtain a simple closed edge description of the particles,which uses the detection results,the locations,of cable particle as seed points.Then,Hough transform is used to detect the cable particles size.Because the detection result of the particle diameter expressed as the number of pixels,we use the scale information in the image to obtain the actual particle size value.