Research On Inspection System For Product's Surface Quality Based On Image Recognition

Industrial vision inspection of industrial automation in modern industry has a broad application prospects. The core technology of machine vision inspection is the image-recognition technology, this paper make an in-depth study of some theories and key technologies of product surface quality testing software system, based on image recognition. An image recognition-detection software system model and related algorithms are proposed, which are based on framework storage, digital image processing, BP neural network and PCA classifier.Firstly, the background knowledge and existing related technologies to industrial vision inspection are introduced, the software composition of vision inspection system is studied, and a detection system software design, which is composed of the image pre-processing module, image recognition classifier module, base memory modules, is presented. Secondly, the study realized variety of image processing algorithms and these algorithms are used to the technical input and output pre-processing as the detection system. The detection system classifier based on BP neural network and PCA classifier is proposed, as well the identification and training methods of the classifier. As the detection system memory modules, the image data storage platform is researched and constructed based on framework and OCI technology. This platform abstracts the core components of the business systems, defines the statue of the collaboration between the two components, and achieves a systematic framework. Finally, the training and testing to the detection system is demonstrated by actual data.The results show that the proposed program is an effective and workable system. The product surface quality inspection system based on image recognition has unique advantages, in terms of speed, accuracy and scalability.