摘要
低对比度手指静脉图像的特征提取,在图像分析方面影响较大。对低对比度手指静脉图像进行特征提取时,需对图像的信号和像素的灰度值进行计算,确定图像像素点的纹线,完成特征提取。传统方法主要根据图像特征空间中高维数据进行特征提取,但忽略了对像素点文线的确定,导致出现特征提取准确率低的问题,提出基于方向检测的低对比度手指静脉特征提取方法。根据低对比度手指静脉图像的原始信号和离散信号,对图像进行小波变换,并利用小波函数的低通、高通滤波器将小波变换的图像进行重构,分层计算图像的信号、量化图像阈值,对低对比度手指静脉图像进行去噪。利用图像的灰度值和方向数计算图像的方向图,确定图像像素点的纹线,利用检测点计算图像方向图的特征点,实现低对比度手指静脉图像的特征提取。仿真结果表明,提出方法能够有效的完成对图像的特征提取,并且具有较高的准确率。
In the traditional method, the feature extraction mainly depends on high-dimensional data in the image feature space, but the feature extraction accuracy is low. This article presents a method to extract low-contrast finger vein feature based on direction detection. Firstly, based on original signal and discrete signal of low contrast finger vein image, wavelet transformation was performed on the image. Meanwhile, the low-pass filter and the high-pass filter of wavelet function were used to reconstruct the wavelet transform image. Moreover, the image signal was hierarchically calculated and the image threshold was quantized. Then, the noise of low-contrast finger vein image was removed. On this basis, the gray value and the direction number of image were used to calculate the directional diagram and determine the streak line of pixel point. Finally, feature points of directional diagram were calculated by detection points. Thus, the feature extraction of low-contrast finger vein image was achieved. Simulation results show that the proposed method can effectively extract the feature of image, which has high accuracy.
引文
[1] 于来行,等. 自适应融合目标和背景的图像特征提取方法[J]. 计算机辅助设计与图形学学报, 2016,28(8):1250-1259.
[2] 黄坚,刘桂雄,林镇秋. 基于多角点结合的机箱标准件图像特征提取方法[J]. 中国测试, 2017,43(9):123-127.
[3] 王仕民,等. 基于加权多尺度张量子空间的人脸图像特征提取方法[J]. 数据采集与处理, 2016,31(4):791-798.
[4] 贾旭,等. 具有普适性的改进非负矩阵分解图像特征提取方法[J]. 计算机应用, 2018,38(1):233-237.
[5] 裴晓芳,王洁,宋林. 基于FPGA的快速图像纹理特征提取方法的研究[J]. 电子测量与仪器学报, 2017,31(7):1067-1073.
[6] 孙继平,杨坤. 一种煤岩图像特征提取与识别方法[J]. 工矿自动化, 2017,43(5):1-5.
[7] 刘兴旺,王江晴,徐科. 一种融合AutoEncoder与CNN的混合算法用于图像特征提取[J]. 计算机应用研究, 2017,34(12):3839-3843.
[8] 余伶俐,等. 基于仿生视觉的图像RST不变属性特征提取方法[J]. 仪器仪表学报, 2017,38(4):985-995.
[9] 刘时城,等. 基于面向对象特征提取的植物叶片面积测量方法[J]. 西北农林科技大学学报(自然科学版), 2017,45(5):161-167.
[10] 贺炜. 多维彩色图像特征快速抽取方法仿真研究[J]. 计算机仿真, 2017,(2):389-392.