A Defects Detection Algorithm Of Ceramic Tile Surface With Complex Background Suppression

Ceramic tiles are important building and decoration materials.Their surface quality directly affects their appearance,performance,and service life.Accurately and efficiently detecting their surface defects is crucial for product quality control.However,in most of the ceramic tile enterprises,the quality determination still relies on manual inspection,which is inefficient and subjective,making it difficult to meet the production demands.Therefore,an automatic detection of the defects is essential.Most of the existing methods for surface defects detection rely on machine vision.However,the detection accuracy of ceramic tile surface is limited due to the influence of dust,dirt,uneven lighting,and textures.False detection is easily caused by the interference of complex background factors.To solve this problem,this paper proposes an accurate and efficient algorithm for the detection of ceramic surface defects in the case of complex background.(1)The image acquisition method for the surface of ceramic tile is studied based on the plane array camera and the linear array camera,respectively.After the image acquisition scheme is proposed according to the detection requirements,the illumination model adapted to the highly reflective characteristics of the ceramic tile surface is studied.The image acquisition effects of the two image acquisition methods are experimentally compared by establishing the image quality evaluation function,and the applications of the two methods to the image acquisition of ceramic tile surfaces are summarized.(2)The algorithm of complex background suppression for the ceramic tile surface is studied.Considering that light,dust,dirt,and texture background affect the accuracy of the defect detection,a comprehensive background suppression algorithm for the ceramic tile surface is proposed.This algorithm is based on the research of single background suppression,which improves the defect detection accuracy by 8.5% compared to the detection without background suppression.(3)The saliency detection algorithm for ceramic tile surface defects is studied.By developing the saliency detection of multi-layer image fusion,while taking into consideration the color and structural characteristics of the defects,the ceramic tile surface saliency map is calculated.The salient region is sized to segment the image sub-block as the suspected defect area for defect recognition,which can reduce the calculation range of the feature vector.(4)The defect recognition algorithm for image sub-blocks of the ceramic tile surface is studied.A cost sensitive random forest model is developed by comprehensively considering the uneven number and variety of the image sub-blocks.The recognition effects of the model and existing machine learning classification models on the defects of image sub-blocks are compared through experiments.The obtained results demonstrate that the proposed model has good generalization ability for different types of ceramic tiles,and the recognition accuracy for the overall image sub-blocks is 97.4%.(5)The defect detection algorithm of the ceramic tile surface under complex background suppression is integrated,and the feasibility and effectiveness of the algorithm are verified by experiments.By building the experimental system,the influence of the background suppression algorithm on the defect detection accuracy is experimentally analyzed on monochrome and textured tiles.An experiment is conducted to compare the defects detection effects of the proposed algorithm with the existing related detection algorithms on the ceramic tile surface.The experimental results show that the detection accuracy of the proposed algorithm on ceramic tile samples of unknown categories is 96.08%,which is 9.00%higher than that of the Re Net-50 model which has the highest detection accuracy on samples of known categories.In summary,the research on defects detection of the ceramic tile surface with complex background suppression can reduce the interference of background information on defect feature descriptions,improve the robustness of the detection algorithm to different types of ceramic tiles,and provide a new solution for promoting the automatic defect detection of ceramic tile surfaces.