Key Technology Research Of Ceramic Surface Defect Detection Based On Computer Vision

Ceramic ware is a common daily necessity in people’s lives,and its quality and safety are widely concerned.During the production process,defects in ceramic ware may affect its quality and aesthetics.Therefore,defect detection has become an indispensable part of ceramic manufacturers.Traditional ceramic defect detection mainly relies on manual visual inspection or magnifying glass observation,which has problems such as low detection efficiency and subjective results.With the continuous development of artificial intelligence technology,the defect detection method based on computer vision technology has gradually become a research hotspot.Computer vision technology can extract the characteristics of defects on the ceramic surface to achieve the detection and classification of defects.It can greatly improve the efficiency and accuracy of defect detection.In view of this,this paper is aimed at ceramic cup defect detection,the main content is as follows.First,a ceramic surface image acquisition system was built in order to acquire ceramic surface images in all directions.The requirements and difficulties of defect detection were analyzed,suitable light sources,lenses and cameras were selected for ceramic cups,and the components and roles of the hardware system were introduced.The ceramic surface defect dataset was created by acquiring the original images and labeling them using the image acquisition system,amplifying the images by data amplification techniques,and dividing the data into training set,validation set,and test set.Secondly,a watershed constraint-based and edge-linked image segmentation method was proposed and applied to the detection of cracked defects on ceramic surfaces.Firstly,the initial edges were obtained by the edge detection algorithm after acceleration,then the hyper-segmented edges were obtained by the marker-based watershed algorithm,and finally several complete edge contours were obtained by the edge linking strategy.The method was applied to ceramic surface cracking defect detection,and experiments show that the method achieves 88.81% accuracy on the cracking dataset.Again,a deep learning-based defect detection method was proposed.The Yolov5network-based model was used and improved to address the problems of ceramic surface defect diversity and difficulty in detecting small target defects.The model was trained and validated on the ceramic surface defect dataset and achieves 93.2% m AP on the test set.Finally,a vision software system for ceramic surface defect detection was designed and implemented to realize the functions of model training,remote transmission of images and local images for model inference,and to verify that the camera acquires ceramic images and transmits them remotely,and the software receives and performs defect detection in real time.In summary,this paper investigated the key technology of ceramic surface defect detection based on computer vision technology,which improved the efficiency and quality of detection compared with manual detection,and at the same time could meet the real-time detection requirements of ceramic surface defect detection,which was of greater significance to the production efficiency of the ceramic industry.