Image Splicing And Setrieval Method Of Cable Circumferential Surface

Cable circumferential image splicing and retrieval refers to splicing the collected cable images,presenting a complete cable surface expansion map,and performing model retrieval in the splicing gallery according to the characteristics of the cable to be tested.At present,most manufacturers use manual inspection for the inspection of cylindrical surfaces,and use the inspector’s experience as the basis for judgment.This inspection method is not only time-consuming and labor-intensive,but also prone to false inspections.The improvement of detection technology has become an urgent problem for manufacturers.Cable circumferential image splicing and retrieval technology mainly includes five parts:cable image collection,cable area positioning and edge fitting,tilt cable correction,cable image splicing,template area generation,and area retrieval.The main content of the subject research:Firstly,according to the actual characteristics of the cable,select the appropriate camera,lens,light source and other hardware equipment,through repeated adjustments and tests,select the appropriate exposure time and build a machine vision imaging system to image the cable Real-time collection.Use the Halcon image algorithm library to perform image processing on the collected cable image.By observing the peaks and valleys of the gray histogram,the image is thresholded to obtain the cable area.Because the cable edge may be interfered by impurities such as fluff,the cable edge The least squares method is fitted to remove the bumps caused by impurities at the edge of the cable.Perform tilt correction on the slanted cable area,mainly using affine transformation and topology-based tilt correction for comparative testing.After correction,the fuzzy area caused by the telecentric lens at the edge of the cable is removed,and two stitching methods are used for the pre-processed cable image.Image splicing: The first method is to extract the feature points of the image and build a transformation matrix based on the feature points of the associated image to splice the cable image;the second method is to randomly select an area near the center of the cable as the retrieval template Region,through the NCC-based template matching principle,the matching region is searched in the repeated region on the associated image,and the image is spliced according to the center coordinates.The retrieval function of the cable model is realized on the basis of the image splicing function.It is mainly to build the spliced cable image into a cable gallery,present different characteristic information according to different cable diameters,and divide the cable into a thick cable gallery and a thin cable gallery.According to different characteristics,different retrieval methods are used to retrieve the cable type,and different optimization methods are used to optimize the algorithm.Finally,on the Visual Studio 2012 development platform,a good software architecture is designed using C++language to ensure that functions such as image processing and motion control can work normally.Tested on the test bench,the topology-based tilt correction method can avoid the deformation of the cable area after the correction.Based on the template matching splicing method,the splicing result has no repeated areas and the splicing speed is fast,and the image will not be rotated and deformed.For different projects Cables of all specifications will produce a good splicing effect.The splicing result of the current gallery meets the requirements with a compliance rate of 100%,and the splicing time is about 4s.After optimizing the cable model retrieval algorithm,the retrieval speed has been improved,and the retrieval time is about 7s.