Mosaic ceramic surface defect detection enhancement algorithm based on guided filter and multi-scale fusion

Abstract This study proposes an enhanced algorithm based on guided filter and multi-scale fusion that detects surface defects of mosaic ceramics, allowing to tackle the small and difficult-to-detect surface reflection problem. This method first combines the gathered multi-angle light source images to obtain the highlight removal image. The guided filter then separates the image into base and detail images. Afterwards, the contrast limited adaptive histogram equalization (CLAHE) algorithm is used to enhance the detailed image. Taplacian contrast weights, saliency weights, and saturation weights are then introduced for the multi-scale fusion of the base and detail images. Finally, defects are extracted by non-linear enhancement, threshold segmentation, morphological processing, and the Canny operator. The proposed algorithm can effectively remove the mosaic ceramic surface highlights, enhance the details of the defect edges, remove the noise interference, improve the overall image quality, and perform rapid detection of defects on the surface of mosaic ceramics. Experiments are then conducted to verify the efficiency of the proposed algorithm. The results show that it has high performance, and it reaches a mosaic ceramic defect detection accuracy of 97.9 % with low leakage and false detection rates of 2.1 % and 1 %.