The Wear Recognition On Guide Surface Based On The Feature Of Radar Graph

In order to solve the wear recognition problem of the machine tool guide surface, anew machine tool guide surface recognition method based on the radar chart ba rycentrefeature is presented in this dissertation. The main contents include: the construction ofwear acquisition platform for the guide surface image, image pretreatment as denoisingenhanced, feature extraction, multi-dimensional data visualization based on radar chart,barycentre feature extraction, et al.1. Experiment platform. A series of Microvision machine vision products, including:camera, LED light source, the light source controller, image acquisition card and otherhardware are selected to build the image acquisition wear platform; MVIPS software isapplied to develop the software platform for image acquisition.2.Image preprocessing. After comparative analysis, the wiener filtering(neighborhood average filtering and median filtering, wiener filtering) is chose as thedenoising method for the guide surface image. Histogram equalization (piecewise lineartransformation, histogram equalization) is adopted as the method of contrastenhancement for image, Using Canny edge detection algorithm to ge t the clear image ofwear edge.3. Characteristic extraction. To extract the geometry and gray features of the wearimages，and the next step is data processing, Radar chart is selected for the data analysisin this dissertation, extracting the gravity of adjacent two-dimensional data and thecentre as a new feature. A hybrid particle swarm algorithm introduced crossover andmutation operations of the genetic algorithm is presented to get the best gravity featurevalue by optimal order the multidimensional data.4. Classification experiment. SVM classification model is chosen for a contrastexperiment of three sets of data, that are the original multi-dimensional feature data, theextracted gravity feature data after randomly sort, the gravity feature data obtained fromthe optimal ordering by hybrid algorithm.The results of image processing show that the method which is proposed and basedon the radar chart barycentre feature in this dissertation can detect the guide surfaceeffectively, and can determine the degree of wear to some extent.