| 一种改进的LBP特征实现铁路扣件识别 |
| |
| 引用本文： | 王强,李柏林,侯云,范宏.一种改进的LBP特征实现铁路扣件识别[J].西南交通大学学报,2018,53(5):893-899. |
| |
| 作者姓名： | 王强 李柏林 侯云 范宏 |
| |
| 摘 要： | 为了提高铁路扣件基于视觉的自动化检测精度，提出了一种改进的LBP （local binary pattern）编码算法. 该方法为了避免基本LBP对噪声敏感问题，根据不同邻域的不同噪声强度，结合测量误差服从高斯分布的原则，计算邻域内像素均值和偏差；根据偏差大小，自动设置阈值，实现自适应噪声抑制. 为了避免基本LBP表达邻域差分关系不完整的缺陷，提出了利用邻域内随机采样的方式得到采样点对，通过比较随机点对的差分关系得到LBP编码. 对在晴天、阴天、雨天等不同天气条件下的铁路扣件图像上进行实验，并与原始以及其他改进LBP进行比较. 结果表明，本文的算法具有更高的检测准确率，晴天提高了3.32%，阴天提高了3.27%，雨天提高了4.10%，能够满足铁路扣件自动化检测的需要.
|
| 关 键 词： | LBP 抗噪声 扣件 图像识别 图像检测 |
| 收稿时间： | 2017-08-08 |
An Improved LBP Feature for Rail Fastener Identification |
| |
| WANG Qiang,LI Bailin,HOU Yun,FAN Hong.An Improved LBP Feature for Rail Fastener Identification[J].Journal of Southwest Jiaotong University,2018,53(5):893-899. |
| |
| Authors: | WANG Qiang LI Bailin HOU Yun FAN Hong |
| |
| Abstract: | An improved LBP (local binary pattern) algorithm is proposed to raise the auto-detection accuracy on fasteners. The original LBP is sensitive to noise. To solve this problem, the pixel means value and deviation are calculated, according to the different noise in different neighborhood and the measuring error is following the Gaussian distribution. Then the threshold is set to realize adaptive noise suppression, according to the deviation values. The original LBP cannot completely express the neighboring difference relationship. To avoid this defect, the proposed method gets the sampled point pairs by random sampling in neighborhood. Then LBP coding is generated through comparing the difference relationship of random point pairs. Tests are carried on fasteners images on clear, cloudy and rainy days with original and other improved LBP algorithms. The comparing results show the proposed method owns more detection accuracy. The detection accuracies increase by 3.32%, 3.27% and 4.10% independently on clear, cloudy and rainy days, which shows the proposed method can meet auto-detection on railway fasteners. |
| |
| Keywords: | |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《西南交通大学学报》浏览原始摘要信息 |
|
点击此处可从《西南交通大学学报》下载全文 |
|
