HHT和神经网络结合的血细胞识别算法  

Blood cell recognition on combination of HHT and neural network

作　　者：龚莹[1] 龙伟[1] 李蒙[1] 张晓[1] 

机构地区：[1]南昌大学信息工程学院,南昌330031

出　　处：《计算机工程与应用》2014年第10期266-270,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.61261911)

摘　　要：多形态血细胞信号影响细胞分类与计数。提出了一种希尔伯特黄变换(Hilbert-Huang Transform,HHT)和神经网络相结合的血细胞识别算法。利用经验模态分解(Empirical Mode Decomposition,EMD)和Hilbert变换提取血细胞信号能量特征,与时域特征一起构成特征向量;建立神经网络模型进行训练与仿真,以实现对多形态血细胞信号的识别。仿真结果表明,该算法识别准确率高,具有良好的识别效果。Polymorphous blood cell signal affects cell classification and counting. This paper proposes an algorithm composed with Hilbert-Huang Transform （HHT）and neural network to recognize blood cell. The energy features of blood cell signal, extracted with empirical mode decomposition and Hilbert transform are combined with time domain features to constitute eigenvector. A model on neural network is built and trained and simulated to recognize blood cells. Simulation results show that the algorithm has high accuracy and good effect on recognition.

关 键 词：希尔伯特黄变换(HHT) 经验模态分解(EMD) 神经网络 血细胞识别 特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]