计算机与现代化 ›› 2025, Vol. 0 ›› Issue (10): 32-36.doi: 10.3969/j.issn.1006-2475.2025.10.006

• 图像处理 • 上一篇    下一篇

基于改进的DeepLabv3+模型的自然环境下舌象分割方法

  


  1. (1.湖南中医药大学信息与工程学院,湖南 长沙 410208; 2.泰芯半导体有限公司长沙分公司,湖南 长沙 410024)
  • 出版日期:2025-10-27 发布日期:2025-10-28
  • 作者简介: 作者简介:刘嵘澂(2002—),女,浙江金华人,学士,研究方向:医学图像处理,E-mail: 2395345893@qq.com; 通信作者:辛国江(1979—),男,辽宁大连人,副教授,博士,研究方向:医学图像处理,E-mail: lovesin_guojiang@126.com; 张杨(1999—),男,湖北枣阳人,硕士,研究方向:医学图像处理,E-mail: 1335841431@qq.com; 朱磊(1997—),男,湖南双峰人,硕士,研究方向:医学图像处理,E-mail: 15717351085@163.com。
  • 基金资助:
     基金项目:国家级大学生创新训练项目(2022批次); 湖南省一流本科课程(2021-896); 湖南省教改课题(HNJG-2021-0584)
       

Natural Environment Tongue Image Segmentation Method Based on Improved Labv3+ Model


  1. (1. School of Informatics, Hunan University of Chinese Medicine, Changsha 410208, China;
    2. Changsha Branch, Taixin Semiconductor Co., Ltd., Changsha 410024, China)
  • Online:2025-10-27 Published:2025-10-28

摘要:
摘要:在自然环境下采集的舌象图像,受光照、背景等因素的影响,其精确分割具有很高的难度。基于改进的DeepLabv3+算法,本文提出一种DeepLabv3-MAC模型,分割自然环境下采集的舌象。首先,将DeepLabv3+模型的主干网络替换为MobileNetv2网络来降低模型复杂度;其次,采用非对称卷积模块增强卷积神经网络的卷积核骨架,提高卷积信息的利用率;最后,引入CBAM注意力机制,不仅能关注空间和通道上参数的重要程度,同时也能提升模型分割精度。实验结果表明,相较于经典的舌象分割算法,本文提出的DeepLabv3-MAC模型具有较好的分割性能,同时,模型的参数量大大减少,可以更快地对自然环境下的舌象进行分割,有利于后期在移动端部署。


关键词: 关键词:舌象分割, DeepLabv3+, DeepLabv3-MAC, 非对称卷积模块, CBAM注意力机制

Abstract:
Abstract: Tongue image segmentation in natural environments poses great challenges due to factors such as lighting and background interference. This paper proposes a DeepLabv3-MAC model based on improved DeepLabv3+ algorithm for segmenting tongue images captured in natural settings. Firstly, the backbone network of the DeepLabv3+ model is replaced with a MobileNetv2 network to reduce model complexity. Secondly, an asymmetric convolutional module is adopted to enhance the convolutional kernel skeleton of convolutional neural network, thereby improving the utilization of convolutional information. Lastly, the introduction of the CBAM attention mechanism not only focuses on the importance of parameters in space and channels, but also enhances the segmentation accuracy of the model. Experimental results demonstrate that, compared to classical tongue image segmentation algorithms, the proposed DeepLabv3-MAC model exhibits superior segmentation performance. Additionally, the model significantly reduces the number of parameters, enabling faster segmentation of tongue images in natural environments and facilitating future deployment on mobile devices.

Key words: Key words: tongue image segmentation, DeepLabv3+, DeepLabv3-MAC, asymmetric convolution module, CBAM attention mechanism

中图分类号: