Abstract: 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.

Key words: Hilbert-Huang Transform（HHT）, Empirical Mode Decomposition（EMD）, neural network, blood cell recognition, feature extraction

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。

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