基于机器视觉的轴承缺陷检测技术研究.pdfVIP

浙江大学硕?l：学位论文摘要

Abstract

defectswillaffectthe

surfaceseriousIy

Bearing

istosubstandard

andoftheautomaticallyidentify

veryimportant

safetyproduct．It

industrialsurfacedefects

inacnJalanenvironment，bearing

productsproduction．In

randomandlackoffixeddetection

thecharacteristicsof

havechangeshapes．The

methodsbasedontT．aditional

ofartinciaIandvisualimage

accuracyrecognition

dataandlabeldatafbr

notthemassjVedeep

ishigh，andimagerequired

processing

ideaof

It’shardtoit．Basedonthe

areinthisscenegetdeepIeaming，this

leaming

surfacedefectsastheobtains

thesistakesquicklytraining

bearingstaningpoint，and

neuralnetworks

labelsusesconVolutional

anddata，and

imagesbysynthesizingimage

mainresearchcontentof

onsurfacedefects．The

tooutdetectionresearch

carrvbearing

thisthesisincludes．thedirections：

following

def色ctdetectionisf}omscrat