Research On Surface Damage Detection Of Mechanical Parts Based On Machine Vision

Surface damage is an important problem for mechanical parts in service,which can affect the operation of the mechanical equipment.Dealing with this problem by remanufacturing damaged mechanical part is a better way in economical and time aspects than replacing with a new part.Correctly detecting damaged surface region is one of the most important parts of remanufacturing process.After detecting damaged region,3D model needed to be prepared with damaged region on it for remanufacturing process.The purpose of this thesis is to by using a 2D camera images of damaged mechanical part to detect damaged region and prepare damaged 3D model for further remanufacturing process.Firstly,2D image of damaged part is obtained and matched,by the novel method presented in this work,with a most suitable 2D view obtained from 3D model of damaged mechanical part.Image matching is done by finding the difference in distance between selected points.The smaller difference in distance the closer points to each other,the most suitable match is one with the smallest distance differences.Secondly,on obtained image of damaged part damaged region is needed to be detected.The damaged region is detected by looking in difference in light level.The damaged part has different light level than undamaged region on the same surface.By selecting threshold value and transform image to binary damaged regions can be detected.Boundary of detecting region and its coordinates are defined.And finally,damaged region need to be defined on the 3D model.The most suitable matched image is show position of the 3D model and helps put 3D model in corresponding position to the coordinates of the boundary of the damaged region.After this by interpolating boundary coordinates and 3D model coordinates,we subtract damaged region from 3D model and obtain damaged 3D model.A set of experiments is conducted to analyze the accuracy and effectiveness of the image-based detection method of surface damage.The proposed detection method can be applied to find different kinds of damage,such as abrasion,erosion,cavitation and fracture on damaged parts surface.