摘要
提出了基于深度学习的指静脉识别算法。在指静脉图像采集过程中,由于受光照强度的影响,手指轮廓存在一定的模糊。为了获得良好的静脉区域图像,采用形态学算法对指静脉原始图像进行感兴趣的区域提取,并进一步使用高斯高通滤波器来增强图像。指静脉采集过程中,手指存在不同程度的旋转,为了消除该影响,使用角度修正算法对指静脉图像进行矫正。由于深度学习在图像分类上表现优异,尤其是AlexNet在ImageNet大赛中的杰出表现,因此采用基于AlexNet的深度神经网络对指静脉图像进行分类。为了加快训练速度,在AlexNet深度神经网络的基础上提出改进方案,主要包括改变卷积核大小和卷积层的构造,从而减少网络参数,降低网络复杂度,加速网络的训练。实验结果表明,利用深度学习对指静脉图像进行分类具有较好的效果。
In this paper,we put forward a finger vein recognition algorithm based on deep learning. In the collection of finger vein image,sometimes the finger contour is blur due to the influence of illumination intensity. In order to obtain a clear vein image,morphological algorithm is used to extract the region of interest for the original image,and the Gaussian high pass filter is adopted to enhance it. An angle correction algorithm is for eliminating the influence caused by the different degrees of finger rotation in the collection of finger vein image. Moreover,deep learning is performed well in image classification,especially the AlexNet in the ImageNet contest. Therefore,the neural networkbased on AlexNet is adopted to classify the finger vein image. In order to speed up the training,we modify the structure of the AlexNet network,including the size of the convolution kernel and the structure of the convolution layer,which lowers the network parameters,reducesnetwork complexity,and accelerates the training of the network. The experiments show that using deep learning for classification of fingervein image has better effect.
出处
《计算机技术与发展》
2018年第2期200-204,共5页
Computer Technology and Development
基金
国家自然科学青年基金项目(61401227)
关键词
指静脉识别
形态学算法
角度修正算法
深度学习
finger vein recognition
morphological algorithm
angle correction algorithm
deep learning
