基于感兴趣区域的改进型LBP手指静脉识别

Improved LBP finger vein recognition based on RoI

为进一步提升手指静脉识别算法的识别率,在图像预处理阶段提出一种快速感兴趣区域(RoI)提取方法,简化候选区域提取的计算过程,缩短手指区域提取时间。识别特征则是在局部二值模式(LBP)的基础上,利用邻域像素的平均值代替中心值,通过邻域像素的关系引入,提升了图像的纹理表达效果。在SDUMLA数据库与天津市智能实验室采集指静脉图像数据库上,分别取得了99.53%,99.74%的识别率,表明了算法优良的识别性能与泛化能力。

In order to further improve the recognition rate of finger vein recognition algorithm,a fast region of interest(RoI)extraction method in image preprocessing stage is proposed,which simplifies the calculation process of the candidate region extraction and shortens the extraction time of the finger region.The recognition feature is based on the local binary pattern(LBP),using the average value of the neighboring pixels instead of the center value,and through introducing relationship of neighboring pixels,which improves the texture expression effect of the image.On the SDUMLA database and the finger vein image database collected by the Tianjin intelligent laboratory,the recognition rates of 99.53%and 99.74%are achieved,respectively,which proves the excellent recognition performance and generalization ability of the algorithm.