计算机应用 ›› 2016, Vol. 36 ›› Issue (12): 3389-3393.DOI: 10.11772/j.issn.1001-9081.2016.12.3389

• 虚拟现实与数字媒体 • 上一篇    下一篇

用于纹理特征提取的改进的成对旋转不变共生局部二值模式算法

于亚风, 刘光帅, 马子恒, 高攀   

  1. 西南交通大学 机械工程学院, 成都 610031
  • 收稿日期:2016-06-06 修回日期:2016-07-26 出版日期:2016-12-10 发布日期:2016-12-08
  • 通讯作者: 刘光帅
  • 作者简介:于亚风(1990-),男,河南周口人,硕士研究生,主要研究方向:图像纹理识别、机器视觉;刘光帅(1978-),男,贵州天柱人,副教授,博士,主要研究方向:逆向工程、图形图像处理;马子恒(1990-),男,河南商丘人,硕士研究生,主要研究方向:图像处理、机器视觉;高攀(1993-),男,四川眉山人,硕士研究生,主要研究方向:图像分割、机器视觉。
  • 基金资助:
    国家自然科学基金资助项目(5127543);四川省科技支撑计划项目(2015GZ0200)。

Improved pairwise rotation invariant co-occurrence local binary pattern algorithm used for texture feature extraction

YU Yafeng, LIU Guangshuai, MA Ziheng, GAO Pan   

  1. School of Mechanical Engineering, Southwest Jiaotong University, Chengdu Sichuan 610031, China
  • Received:2016-06-06 Revised:2016-07-26 Online:2016-12-10 Published:2016-12-08
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (5127543), the Science and Technology Support Project of Sichuan Province (2015GZ0200).

摘要: 针对用于纹理特征提取的成对旋转不变共生局部二值模式(PRICoLBP)算法计算特征维度大、旋转不变性较差、对光照变化敏感的问题,提出一种融合局部纹理信息的改进PRICoLBP算法。首先,分别最大化和最小化图像像素点的二值序列,得到两个邻域像素点的坐标,由中心像素点坐标和得到的邻域像素点坐标计算出共生点对的坐标;其次,利用完备二值模式(CLBP)算法提取图像的每个像素点的纹理信息。在相同分类器下,对Brodatz、Outex(TC10,TC12)、Outex(TC14)、CUReT和KTH_TIPS数据库的分类实验中,所提算法的识别率比PRICoLBP算法分别提高了0.17、0.24、2.65、2.39和2.04个百分点。实验结果表明,所提算法在处理纹理旋转变化、光照条件多样的图像时具有较好的识别效果。

关键词: 特征提取, 局部二值模式, 成对旋转不变共生局部二值模式, 旋转不变性, 光照鲁棒性

Abstract: The texture feature extraction algorithm of Pairwise Rotation Invariant Co-occurrence Local Binary Pattern (PRICoLBP) has characteristics of high computing feature dimension, poor rotation invariance and sensitivity to illumination change. In order to solve the issues, an improved PRICoLBP algorithm was proposed. Firstly, the coordinates of two neighboring pixels were obtained by respectively maximizing and minimizing the binary sequence of image pixels. Then, the position coordinates of co-occurred pixel points were calculated via the position coordinates of the center pixel and the two neighboring pixels. Secondly, the texture information of every image pixel was extracted through utilizing the Completed Local Binary Pattern (CLBP) algorithm. Compared with PRICoLBP, the recognition rate of the proposed method was improved respectively by the percentage points of 0.17, 0.24, 2.65, 2.39 and 2.04, on the image libraries of Brodatz, Outex(TC10, TC12), Outex(TC14), CUReT and KTH_TIPS under the same classifier. The experimental results show that the proposed algorithm has better recognition effect for the images with texture rotation variation and illumination change.

Key words: feature extraction, Local Binary Pattern (LBP), Pairwise Rotation Invariant Co-occurrence Local Binary Pattern (PRICoLBP), rotation invariance, illumination robustness

中图分类号: