计算机应用 ›› 2020, Vol. 40 ›› Issue (9): 2781-2788.DOI: 10.11772/j.issn.1001-9081.2019122131

• 应用前沿、交叉与综合 • 上一篇    

基于DeepLab v3的西藏地区降雨云团分割方法

张永宏1,2, 刘昊2, 田伟3, 王剑庚4   

  1. 1. 南京信息工程大学 气象灾害预报预警与评估协同创新中心, 南京 210044;
    2. 南京信息工程大学 自动化学院, 南京 210044;
    3. 南京信息工程大学 计算机与软件学院, 南京 210044;
    4. 南京信息工程大学 大气物理学院, 南京 210044
  • 收稿日期:2019-12-19 修回日期:2020-02-27 出版日期:2020-09-10 发布日期:2020-05-13
  • 通讯作者: 刘昊
  • 作者简介:张永宏(1974-),男,山东临沂人,教授,博士,主要研究方向:大气遥感检测、图像处理分析;刘昊(1996-),男,江苏镇江人,硕士研究生,主要研究方向:遥感图像处理、大数据分析;田伟(1980-),男,江苏涟水人,副教授,博士,主要研究方向:计算机软件、人工智能降雨估计、大数据处理;王剑庚(1982-),男,甘肃定西人,副教授,博士,主要研究方向:积雪遥感、"地-气"相互作用。
  • 基金资助:
    国家自然科学基金资助项目(41875027)。

Rainfall cloud segmentation method in Tibet based on DeepLab v3

ZHANG Yonghong1,2, LIU Hao2, TIAN Wei3, WANG Jiangeng4   

  1. 1. Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science and Technology, Nanjing Jiangsu 210044, China;
    2. School of Automation, Nanjing University of Information Science and Technology, Nanjing Jiangsu 210044, China;
    3. School of Computer and Software, Nanjing University of Information Science and Technology, Nanjing Jiangsu 210044, China;
    4. School of Atmospheric Physics, Nanjing University of Information Science and Technology, Nanjing Jiangsu 210044, China
  • Received:2019-12-19 Revised:2020-02-27 Online:2020-09-10 Published:2020-05-13
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (41875027).

摘要: 针对高原地区数值预测法建模复杂,雷达回波外推法易产生累积误差且模型参数难以设置的问题,提出了一种基于改进DeepLab v3网络模型的西藏地区降雨云团的分割方法。首先,通过编码网络中的卷积层和残差模块进行下采样;然后,利用空洞卷积构建多尺度采样模块,并且加入注意力机制模块提取深层高维特征;最后,通过解码网络利用反卷积恢复特征图分辨率。将所提方法与谷歌语义分割网络DeepLab v3等模型在验证集上进行比较,实验结果表明所提方法具有更好的分割性能与泛化能力,其降雨云团分割结果更为准确,平均交并比(Miou)达到0.95,与原始DeepLab v3相比提高了15.54个百分点。在小目标上和非平衡数据集上,该方法可以更准确地分割出降雨云团,为降雨云团监测预警提供参考。

关键词: 降雨云团分割, 多尺度采样, 注意力机制, DeepLab v3, 遥感图像处理

Abstract: Concerning the problem that the numerical prediction method is complex in modeling, the radar echo extrapolation method is easy to generate cumulative error and the model parameters are difficult to set in plateau area, a method for segmenting rainfall clouds in Tibet was proposed based on the improved DeepLab v3. Firstly, the convolutional layers and residual modules in the coding network were used for down-sampling. Then, the multi-scale sampling module was constructed by using the dilated convolution, and the attention mechanism module was added to extract deep high-dimensional features. Finally, the deonvolutional layers in the decoding network were used to restore the feature map resolution. The proposed method was compared with Google semantic segmentation network DeepLab v3 and other models on the validation set. The experimental results show that the method has better segmentation performance and generalization ability, has the rainfall cloud segmented more accurately, and the Mean intersection over union (Miou) reached 0.95, which is 15.54 percentage points higher than that of the original DeepLab v3. On small targets and unbalanced datasets, rainfall clouds can be segmented more accurately by this method, so that the proposed method can provide a reference for the rain cloud monitoring and early warning.

Key words: rainfall cloud segmentation, multi-scale sampling, attention mechanism, DeepLab v3, remote sensing image processing

中图分类号: