Research On 3D Reconstruction Of Aero Engine Blade Defects Based On Edge Fast Stereo Matching

With the development of our country's civil aviation career,the number of civil aircraft increases rapidly and steadily.Engine maintenance expense accounted for 41% of the whole aircraft maintenance expense.Because aircraft engine blade always works in high temperature high pressure and high speed environment,the blades are the heart of aero-engine and are easy to be damaged.In order to saving cost and maintain stability of engine performance the damaged blades need to be repaired and reused.At present our damaged blades must be sent abroad.This channel increases maintenance costs and prolong maintenance cycle.In order to break the distress situation,develop aircraft engine blade automatic welding repair equipment with independent intellectual property right is our most urgent things.In the process of research making 3D reconstruction for blade defect part for welding and grinding is important step.This paper makes use of stereo vision system to get defect parts' 3d appearance information.In the premise of meeting accuracy requirement,this method could use as little information as possible to make 3d reconstruction and the speed of reconstruction is also promoted.In view of the blade tip's damage,this article adopts trinocular stereo vision system to collect information and then make use of edge character stereo matching to make 3d reconstruction for defect parts.This paper proposes a kind method of basing same vertical plane ternary stereoscopic vision for defect blades' information collection.First make use of the calibration toolbox in Matlab software for stereo calibration through three cameras and concluding conversion relationship between three cameras and conversion relationship between pixel coordinate system and world coordinate system.Second use improved Canny operator to extract image's edge data.Through NURBS curve to fit and expand edge curve getting defect parts' complete edge in the two-dimensional image.Finally use edge character for stereo matching.By matching point's Spatial coordinates to make curve fitting.After fitting we could get defect parts' 3d shape information.Through experiment the defect parts' average error under 0.2mm.The data could provide theoretical basis for subsequent welding and polishing.