基于特征值加权融合的手指静脉图像质量评估

Finger Vein Image Quality Evaluation Based on Eigenvalues Weighting and Fusing

对手指静脉图像进行处理:首先提取了空间域梯度、对比度、图像的二维熵、位置偏移度、信噪比等5个特征值,然后对其进行加权融合进而建立质量评估模型,最后利用模型进行分类实验和识别对比实验。结果表明:将提取的特征值加权融合后建立的模型不仅能够很好地对高低质量的手指静脉图像进行分类,也能在一定程度上提高识别系统的识别性能。

According to processing the finger vein image,firstly,this article extracted five characteristics,which including the gradient in the spatial domain,contrast,the two-dimensional entropy of image,position deviation and signal-to-noise ratio. Secondly,the five eigenvalues were weighted and fused to set up a quality evaluation model. Finally,the classification and identification-contrast experiments were done using the model. The results show that using the model to set up after characteristics weighting and fusing,not only the high and low quality finger vein images can be well classified,but also the recognition performance of recognition system can be improved to some extent.