计算机工程与应用 ›› 2016, Vol. 52 ›› Issue (20): 158-162.

• 模式识别与人工智能 • 上一篇    下一篇

基于LBP和稀疏表示的天际线检测算法研究

涂  兵1,2,3,潘建武1,2,3,张国云1,2,3,李朝辉1,3,李孝春1,3   

  1. 1.湖南理工学院 信息与通信工程学院,湖南 岳阳 414006
    2.湖南理工学院 复杂系统优化与控制湖南省普通高等学校重点实验室,湖南 岳阳 414006
    3.湖南理工学院 IIP创新实验室,湖南 岳阳 414006
  • 出版日期:2016-10-15 发布日期:2016-10-14

Research on skyline detection algorithm based on LBP and sparse representation

TU Bing1,2,3, PAN Jianwu1,2,3, ZHANG Guoyun1,2,3, LI Chaohui1,3, LI Xiaochun1,3   

  1. 1.School of Information and Communication Engineering, Hunan Institute of Science and Technology, Yueyang, Hunan 414006, China
    2.Key Laboratory of Optimization and Control for Complex Systems, Hunan Institute of Science and Technology, Yueyang, Hunan 414006, China
    3.Laboratory of Intelligent-Image Information Processing, Hunan Institute of Science and Technology, Yueyang, Hunan 414006, China
  • Online:2016-10-15 Published:2016-10-14

摘要: 天际线的检测在视觉导航、地理位置标注中具有重要的作用。提出一种基于LBP(Local Binary Pattern)和稀疏表示融合算法用来检测输入图像中的天际线。首先对图像作灰度化处理,接着对训练样本图像天际线相邻像素点坐标建立3×3的特征提取区域,根据LBP特征统计直方图建立天际线LBP特征向量,得到字典;然后在测试样本上依次搜索3×3区域中的像素点,通过稀疏分解系数计算重构误差,根据重构误差阈值的设定判断此区域为天际线区域,从而得到此区域中的天际线坐标。提出的算法在内华达大学机器视觉实验室Web Set数据集上进行了测试,实验结果表明:提出的算法能有效地检测出输入图像中的天际线像素点坐标,具有较好的有效性和时效性。

关键词: 视觉导航, 局部二值模式(LBP), 稀疏分解, 重构误差

Abstract: The skyline detection plays an important role in visual navigation and geographical location annotation. In this paper, an algorithm based on LBP and sparse representation is proposed to detect the skyline in the input images. Firstly, the image is processed with gray scale. And then, the 3×3 feature extraction region is established for the adjacent element coordinates of the skyline in the training sample images. According to the LBP statistical histogram, the LBP feature vector of skyline is established to get the dictionary. Finally, it searches the pixel points of the 3×3 region in the training samples. The reconstruction error is calculated through spare decomposition coefficient. And the skyline region is recognized by setting the threshold of reconstruction error. According to the coordinates of skyline in the training samples, the corresponding skyline coordinates are obtained in this skyline region. The proposed algorithm is tested on the Web Set in University of Nevada’s Machine Vision Laboratory. And the experimental results show that the algorithm can detect the skyline pixel coordinates effectively in the input images, which has good validity and timeliness.

Key words: visual navigation, Local Binary Pattern(LBP), sparse decomposition, reconstruction error