| 基于改进LPB特征算子的行人检测算法研究 |
| |
| 引用本文： | 王根岭,徐洋洋,盛实旺.基于改进LPB特征算子的行人检测算法研究[J].工业控制计算机,2014(3):43-45. |
| |
| 作者姓名： | 王根岭 徐洋洋 盛实旺 |
| |
| 作者单位： | 浙江理工大学信息学院,浙江杭州310018 |
| |
| 摘 要： | LBP特征算子具有光照不变性、旋转不变性且计算简单等优点,是行人检测中描述人体常用的特征算子,但在场景光照与行人对比度较大时LBP算子对行人特征描述不太准确,容易造成跟踪失败。针对以上情况提出了改进的LBP算子,在计算LBP二进制码时不仅考虑中心点与邻域点像素值的大小,还将二者之间的对比度信息加入到LBP算子中。通过实验证明,改进的LBP特征算子对图像噪声不太敏感,且在光照不足的环境中对行人具有更强的表示能力和判别能力。
|
| 关 键 词： | 行人检测 LBP SVM 对比度因子 |
New Algorithm of Pedestrian Detection Based on Improved LBP Operator |
| |
| Abstract: | LBP characteristic operator has the advantages of illumination invariance,rotation invariance and simple calculation.It is the common operators that used in describing human features in pedestrian detection.LBP operator can not accurately de- scribe the characteristics of pedestrians when the contrast between the scene lighting and pedestrian is large,it＇s easy to cause track failure.Be challenged by this problem,this paper brings up an improved LBP operator.When calculating the LBP binary codes not only take the the pixels of center and neighborhood nodes into account,but also add the contrast infor- mation between them to the LBP ooerator. |
| |
| Keywords: | oedestrian detection LBP SVM contrast factor |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
