中国邮电高校学报(英文) ›› 2011, Vol. 18 ›› Issue (3): 110-119.doi: 10.1016/S1005-8885(10)60072-6

• Others • 上一篇    下一篇

Urban scene recognition by graphical model and 3D geometry

任柯燕1,孙汉旭2,贾庆轩3,Weiyu Zhang4   

  1. 1. 北京邮电大学自动化学院
    2.
    3. Automation school,bupt
    4. University of Pennsylvania
  • 收稿日期:2010-10-11 修回日期:2011-02-25 出版日期:2011-06-30 发布日期:2011-06-13
  • 通讯作者: 任柯燕 E-mail:aizi625@qq.com
  • 基金资助:

    This work was supported by the National Natural Science Foundation of China (60803103), Research Found For Doctoral Program of Higher Education of China (200800131026), Fundamental Research Funds for the Central Universities (2009RC0603, 2009RC0601).

Urban scene recognition by graphical model and 3D geometry

  • Received:2010-10-11 Revised:2011-02-25 Online:2011-06-30 Published:2011-06-13
  • Supported by:

    This work was supported by the National Natural Science Foundation of China (60803103), Research Found For Doctoral Program of Higher Education of China (200800131026), Fundamental Research Funds for the Central Universities (2009RC0603, 2009RC0601).

摘要:

This paper proposes a simple and discriminative framework, using graphical model and 3D geometry to understand the diversity of urban scenes with varying viewpoints. Our algorithm constructs a conditional random field (CRF) network using over-segmented superpixels and learns the appearance model from different set of features for specific classes of our interest. Also, we introduce a training algorithm to learn a model for edge potential among these superpixel areas based on their feature difference. The proposed algorithm gives competitive and visually pleasing results for urban scene segmentation. We show the inference from our trained network improves the class labeling performance compared to the result when using the appearance model solely.

关键词:

scene recognition, CRF, graphical model, 3D geometry

Abstract:

This paper proposes a simple and discriminative framework, using graphical model and 3D geometry to understand the diversity of urban scenes with varying viewpoints. Our algorithm constructs a conditional random field (CRF) network using over-segmented superpixels and learns the appearance model from different set of features for specific classes of our interest. Also, we introduce a training algorithm to learn a model for edge potential among these superpixel areas based on their feature difference. The proposed algorithm gives competitive and visually pleasing results for urban scene segmentation. We show the inference from our trained network improves the class labeling performance compared to the result when using the appearance model solely.

Key words:

scene recognition, CRF, graphical model, 3D geometry