光学精密工程2025，Vol.33Issue(22)：3536-3548,13.DOI:10.37188/OPE.20253322.3536
基于多尺度特征融合与分块注意力的齿轮表面缺陷分割算法
Gear surface defect segmentation algorithm based on multi-scale feature fusion and block-wise attention
摘要
Abstract
To address the limitations of traditional segmentation models in handling complex background interference and subtle defect regions of gears—particularly their insufficient feature-representation capabili-ty and poor robustness—this paper proposed a novel segmentation network based on multi-scale feature fu-sion and block-wise attention,aiming to enhance the representation of gear visual features and improve the detection performance of fine defects.First,a multi-scale feature-enhancement module replaced the stan-dard downsampling blocks in the UNet encoder;it leveraged a parallel multi-branch convolutional struc-ture to collaboratively extract multi-scale and multi-directional features,thereby enhancing the model's perception of both local details and global context.Second,a block-wise feature-focusing module was intro-duced after downsampling;it employed a block-wise multi-head attention mechanism to independently ana-lyze local regions,significantly improving sensitivity to minute defects and local texture variations.Final-ly,a weighted hybrid loss function was designed by combining Dice loss,binary cross-entropy(BCE)loss,and a gradient-difference constraint,effectively mitigating the class-imbalance issue and optimizing the quality of segmentation boundaries.Experimental results on both a self-constructed and public gear de-fect dataset demonstrate that the proposed method outperforms UNet and other state-of-the-art models in various gear defect segmentation tasks,achieving accuracy rates of 91.27%and 85.88%,respectively.The results validate the effectiveness and superiority of the proposed approach for precise detection and segmentation of surface defects in gears.关键词
表面缺陷检测/图像分割/UNet/分块注意力/齿轮Key words
surface defect detection/image segmentation/UNet/block-wise attention/gear分类
信息技术与安全科学引用本文复制引用
ZHAO Lin,MA Siqi,FANG Yiming,LUO Kai,ZHANG Guoyun,SHI Zhaoyao..基于多尺度特征融合与分块注意力的齿轮表面缺陷分割算法[J].光学精密工程,2025,33(22):3536-3548,13.基金项目
国家自然科学基金资助项目(No.62473148) （No.62473148）
湖南省研究生科研创新项目(No.CX20240095,No.CX20240968) （No.CX20240095,No.CX20240968）
