| 摘要: |
| 提出一种基于差值局部方向模式的人脸特征表示方法(difference local directional pattern,简称DLDP):首先,通过Kirsch掩模卷积运算,为每个像素计算8个方向的边缘响应值;然后,计算8个相邻边缘响应值的强度差,前k个最突出的强度差对应的方向编码为1,其他方向编码为0,形成一个8位二进制数表示对应的DLDP模式;此外,针对高分辨率的Kirsch掩模单纯考虑方向性而没有考虑像素位置权重的问题,提出相应的掩模权值设计方法;最后,把每幅图像划分成多个不重叠的局部图像块,通过统计图像块上不同DLDP模式个数生成相应的子直方图,所有子直方图被串联起来表示一幅人脸图像.实验结果表明,该方法在光照、表情、姿态和遮挡方面获得了较好的结果,尤其针对遮挡情况,表现更为突出. |
| 关键词: 差值局部方向模式 Kirsch掩模 特征表示 人脸识别 |
| DOI：10.13328/j.cnki.jos.004896 |
| 分类号: |
| 基金项目:国家自然科学基金(61170185,61303016); 辽宁省教育庁科学研究一般项目(L2015403); 珠海市重点实验室科技攻关项目(2012D0501990026); 沈阳航空航天大学校博士启动金(15YB05) |
|
| Face Feature Representation Based on Difference Local Directional Pattern |
|
LI Zhao-Kui1, DING Li-Xin2, WANG Yan1, HE Jin-Rong3, DING Guo-Hui1
|
|
1.School of Computer, Shenyang Aerospace University, Shenyang 110136, China;2.State Key Laboratory of Software Engineering (Wuhan University), Wuhan 430072, China;3.College of Information Engineering, Northwest A & F University, Yangling 712100, China
|
| Abstract: |
| A face feature representation method based on difference local directional pattern (DLDP) is proposed. Firstly, each pixel of every facial image sub-block gains eight edge response values by convolving the local neighborhood with eight Kirsch masks. Then, the difference of each pair of neighboring edge response values is calculated to form eight new difference directions. The top k difference response values are selected and the corresponding directional bits are set to 1. The remaining (8-k) bits are set to 0, thus forming the binary expression of a difference local direction pattern. In addition, high-resolution Kirsch masks only consider directions but ignore the weight values of each pixel location. DLDP proposes a design method of weight values. Finally, the sub-histogram is calculated by accumulating the number of different DLDP of image blocks. All sub-histograms of an image are concatenated into a new face descriptor. Experimental results show that DLDP achieves state-of-the-art performance for difficult problems such as expression, illumination and occlusion-robust face recognition in most cases. Especially, DLDP gets better results for occlusion problem. |
| Key words: difference local directional pattern (DLDP) Kirsch mask feature representation face recognition |
