Font Size: a A A

Research On Edge Detection Of Wear Particle By Combining Lifting Wavelet And Grey Relational Analysis

Posted on:2014-06-27Degree:MasterType:Thesis
Country:ChinaCandidate:F M ChenFull Text:PDF
GTID:2268330422452943Subject:Mechanical design and theory
Abstract/Summary:PDF Full Text Request
With the increasing demand for modern production equipment’s management, failuredetection and diagnosis techniques for the running status of mechanical equipment has become afocus of engineering. Ferrography is a kind of important technique to judge and forecast therunning state of mechanical facilities by wear particles analysis which can improve the ability offailure detection and diagnosis and reduce the probability of the machine failure. Ferrographyimage segmentation is the precondition of wear particles analysis, it has decisive effect onferrography. The edge of wear particle is an important character for ferrography imagesegmentation, it provides important characteristic parameters for wear particles analysis.Firstly, the integration of image smoothing, median filtering and brightness regulation is usedto remove the noises. It is important for the edge detection of wear particles if separating wearparticles from the background accurately, the paper proposed a new method named H&DC toseparate wear particles from the background by histogram and image depth conversion andcontrast analysis on H&DC and otsu.This paper focuses on how to edge detection of accurate continuous and single pixel edge.Firstly, The basic knowledge of wavelet transform is studied in this paper, D4lift Wavelettransform based multi-scale edge detection is used to edge detection of wear particles and thedetection results are analyzed. Secondly, grey relational analysis is studied for edge detection ofwear particles and the detection results are analyzed. Finally, the paper proposes a new algorithmof wear particles detection, named L&G algorithm, which is a kind of integration algorithms onbasis of lift wavelet transform and grey relational analysis according to their advantages anddisadvantages.Compared with the Soble algorithm, the Laplacian algorithm, the Canny algorithmand multi-scale edge detection algorithm, L&G algorithm is more efficient at reducing noise,accurate continuous and single pixel edge image is obtained by L&G algorithm. it is an efficientway to the edge detection of wear particles.
Keywords/Search Tags:Ferrography Image, Edge Detection, Lifting Wavelet, Grey Relational Analysis
PDF Full Text Request
Related items