Research On Surface Quality Inspection Techniques Based On Surface Registration

With the development of industrial ability,we would like the products have more complex surface.Therefore,manufactures with complicated surface have been more widely used in product design.Surface inspection techniques based on surface registration is the most common way to measure the quality of those manufactures.However,there is obscuration during the scanning process.In addition,different manufactures have totally distinctive surfaces,which would make it largely difficult to inspect parts by one method.In this paper,we work on issues about parts inspection and surface registratio n,including registration of raw scans,improvement on artificial bee colony algorithm and applications in registration field.First,we build the model of the techniques of parts inspection via surface registratio n.Then a quantitative analysis is conducted.With the help of linear transformation,we convert the rotation and translation into numerical optimization problem in 6-degrees space.2D and 3D models are chosen to show the concavity and convexity of the target function by drawing the functional images.Second,we propose a new local surface descriptor for image registration.A local reference frame can be easily constructed by singular value decomposition.Then information entropy is introduced to define the local descriptor,i.e.,spherical dissection signature.Dozens of experiments show that our new descriptor can register surfaces with a success rates near 100% if models have enough geometric features.Third,in order to enhance the robustness of artificial bee colony algorithm,we make some improvements about ABC and apply them into surface registration.The selfoptimizing strategy of particle search optimization algorithm is introduced to the employed bees and the onlooker bees of ABC,thus,an improved ABC algorithm-PABC is proposed.Experiments have been performed to test the performance of the new algorithm on surface registration.Results show that the accuracy and speed of the new algorithm are tremendously increased compared to the standard ABC,which can meet the requirement of industrial applications.Finally,we summary the whole paper and look into the future of this field.