Study On Micro-particle Size Measurement Technology Based On Machine Vision

China is currently a major country in industrial manufacturing,and there is a huge demand for micro-parts in industrial production.With the industrial transformation and upgrading,more stringent requirements have been put forward for the parameter indicators of micro-workpieces.The size of unqualified workpieces may cause great harm to the safe operation of the entire equipment.In view of the cumbersome and inefficient operation of small particle size measurement under the current nuclear power background,a set of small size measurement device based on machine vision has been researched and built.The main tasks completed in this thesis are as follows:?According to the characteristics of the measured target in this thesis,designed a suitable visual measurement scheme for the size of each layer of particles based on the metallographic method.Researched and built the hardware system of the micro-size measurement device,designed the structure of the entire measurement device platform.Selected and designed the metallurgical microscope,imaging device,lighting system,vertical adjustment device and two-dimensional translation stage in the hardware system,then seted up a measuring device.?Studied the size measurement plan of each layer of the micro particle profile.According to the characteristics of the particle profile,it is determined that the main steps of image processing are noise filtering,adaptive contour detection,and multi-criterion contour extraction.In the case of poor results,adaptive threshold segmentation is used to extract contours of each layer combined with gradient feature information.The scanning recognition and positioning methods of the particles on the sample include image denoising,image enhancement,and core center positioning.In the process of scanning recognition and measurement,the damaged particles on the sample can be automatically counted and the damage rate can be obtained.Steps such as denoising,dark pixel statistics and straight line detection are adopted for the device origin location.The dynamic calibration of the measuring device realizes the accurate measurement of dimensions.? Designed the control method and program of the two-dimensional translation stage,adopted custom instructions to ensure accurate data transmission,and selected the appropriate image acquisition and scanning method to speed up the scanning speed.Designed and developed human-computer interaction software,designed login modules,measurement modules,and data management modules to improve the automation of the device.Finally,repeated experiments that were performed on the entire measurement device.The analysis of experimental data showed that the measurement accuracy of the device was no more than 3?m,and the average time for scanning,identifying and positioning particles on a measurement sample was 30 minutes.The entire measurement process is simple to operate,which greatly improves the automated measurement level of particles.