Feature matching driven background generalization neural networks for surface defect segmentation

In this paper, we address the challenge of background generalization in surface defect segmentation for surfacemounted device chips, particularly focusing on template-sample comparison algorithms. These algorithms often struggle with background features in templates and samples that exhibit spatial variations, including translation and rotation. The inherent spatial equivariance in CNN -based algorithms complicates the elimination of noise attributed to these spatial variations. To address this issue, we developed the Background Generalization Networks (BGNet). BGNet effectively reduces spatial variation noise by subtracting background features of samples and templates based on their matching relationships. It starts by extracting dense features rich in global and interactive information via a Siamese network and then applies self- and cross -attention mechanisms from Transformers. The matching score is calculated based on feature similarity, with matching relations established using the Mutual Nearest Neighbour (MNN) algorithm. These relations enable us to mitigate the noise caused by spatial variations and implement a multiscale fusion of detailed and semantic information, leading to more accurate segmentation results. Our experiments on OCDs and PCBs have shown that BGNet surpasses existing state -of -the -art methods in terms of performance. Furthermore, the code for this work is available on GitHub: https://github.com/Max-Chenb/BG-Net.