Defect Detection In Casting Surface Based On Machine Vision

Castings are widely used in various fields of manufacturing industry.The surface quality of castings,as a key part of product quality,has attracted wide attention.Due to the problems of imperfect process flow and incomplete controllable processing environment in the production process of castings,some defects inevitably exist on the surface of castings.These defects will affect the performance of the product.At present,the detection of casting surface defects in domestic enterprises mainly relies on manual completion,and the detection accuracy and efficiency are difficult to meet the actual production requirements.Machine vision-based inspection technology can effectively avoid manual detection of existing problems.It has many advantages such as noncontact,robustness and high efficiency.At present,the technology of casting surface defect detection based on machine vision is still developing.This research takes the brake disc as the object and studies the detection technology of casting surface defects with the method of machine vision.The main research contents are as follows:(1)Design of surface defect detection system for castings.Firstly,the features of casting surface defects and the inspection requirements are analyzed,and the inspection technical index requirements are clarified.Secondly,according to the requirements of indicators,key components such as cameras,lenses,light sources are selected,and an experimental platform for surface defect detection is built.(2)Research on image processing methods of casting defects.Firstly,by analyzing and comparing various filtering algorithms,the filter with the best performance is selected to complete image denoising.Secondly,a template matching algorithm based on similarity computing termination strategy and image pyramid strategy is proposed,which effectively realizes the segmentation of brake disc surface area.Thirdly,a method combining maximum entropy threshold segmentation with morphological open operation is proposed.The method is used to segment the suspected defect area on the surface of the brake disc.Experiments show that the algorithm can segment the suspected defect region.(3)Research on feature extraction and classification methods of surface defects of castings.Firstly,the characteristics of various surface defects in images are analyzed.The geometric features and uniform LBP are extracted.And the dimensionality reduction of fusion high-dimensional features is realized by using PCA.Secondly,the relevant theory of SVM is studied,and the appropriate kernel function and related parameters are selected according to the experiment,and the required classification results are obtained.(4)Test and analysis of casting surface defects.Firstly,according to the actual needs,the functions of the software system are defined and the related functions are tested.Secondly,the software was tested with the experimental platform and brake disc samples.The feasibility and effectiveness of the proposed algorithm are verified.The experimental results meet the technical specifications of defect detection.