Research On Detection Of Tile Surface Defects Based On Digital Image Processing

Due to the low efficiency,instability,and high costs of manual inspection of ceramic tiles,and it is difficult to separate because the gray values of the defect is similar to the gray values of the pattern background in the automatic detections.Based on digital image processing,defect detection of tile surface is studied.After a large number of experiments,three detection algorithms,a defect detection method based on threshold segmentation,a self-adaptive detection method,and a defect detection method fused with morphological and wavelet transforms,are summarized and designed in this paper.In the threshold segmentation defect detection,images are preprocessed and filtered.After separating the big different background and pattern information,the pixels with similar gray values are excluded according to the regional features.Considering that the threshold segmentation method relies on gray values and high defect contrast ratio,an adaptive morphological template is designed to preprocess the brick surface,and different structural elements are used to the original image for adaptive morphological parallel and serial processing.Contrast-conspicuous images are detected by multi-scale morphology and contours are finally detected.Considering the disadvantages of adaptive morphology to change contours,with a comparative experiment,a fusion method is designed.After preprocessing with an adaptive morphological template,images are subjected to a secondary wavelet transform,and the high and low frequency components are used differently.For low-frequency information containing a large number of defects and backgrounds,the extra backgrounds are removed by morphometry.For high-frequency horizontal,vertical,and diagonal information containing a small number of edge contours and mutations,the three-component low-frequency information is removed by an improved sixth-order Butterworth high-pass homomorphic filter,and the defect profile is obtained after reconstruction.Compare the accuracy,ease of detection,and innovation of the three methods,and the advantages and disadvantages of the three methods are summarized.The corresponding detection system is designed based on the algorithm.The system is mainly divided into three parts: acquisition,control,and processing.According to the need to select the corresponding system hardware,the relationship between hardware is determined.Combined hardware is placed on the packaging line,and the system's inspection process is conceived.In the image display module,after the driver is installed,the selected CCD camera is controlled,the corresponding library file is linked in VC++,the path for reading,saving,and storing are set,and the library function is called on the camera screen in real time display.In the control module,the upper computer is communicated with the lower computer,and the camera is located directly above the center axis of the conveyor belt.The motor connected to the screw is controlled by the lower computer and moves in a four-phase,eight-beat manner.In the processing module,processing interface is designed.Finally,the research is summarized,and the deficiencies in the research are introduced and prospected.