基于特征区分度的静脉图像质量评价算法

Vein Image Quality Assessment Algorithm Based on Feature Discrimination

针对目前采集静脉图像的质量评价标准主要依赖主观判断的缺陷,提出了一种基于特征区分度的静脉图像质量评价算法。该算法首先基于近红外静脉图像的特点建立了质量评价模型,而后依据训练样本特征的区分度确定质量评价模型参数,最后基于该模型优化采集到的静脉图像能够提取出优质的静脉特征。实验结果表明,基于该方法获取的特征具有较好的类间区分性,进而提高识别效果。

For the defects that vein image quality assessment criterion was mainly relied on subjective judgment currently, a vein image quality assessment algorithm based on feature discrimination was proposed. Firstly, a quality assessment model was proposed based on vein image features, then model parameters were fixed according to the discrimination of training samples feature, and finally high quality features of acquired vein images can be extracted based on model optimization. Experiments show that the extracted features have good dispersion in different classes, and the recognition effect is improved.