Abstract：Leukocyte morphology with mixed attributes is a complex system with attribute multi-hierarchical relationship. The six-classification technology of leukocyte morphology can be achieved effectively by sufficient expression and variable level use of hierarchical attribute. A novel method of leukocyte morphology classification based on attribute multi-hierarchical relationship was proposed. The decision relationships and associated relationships of classification were obtained based on analysis of leukocyte morphology features attribute multi- hierarchical relationship. The classifier was established according to the relationships and reached an average classification accuracy of 95.98% significantly higher than the other 3 kinds of classical algorithm in a contrast experiment to a dataset of 952 hospital actual leukocyte images. Experimental results show that the method has a better classification performance, and also prove that attribute multi-hierarchical relationship of complex system has brilliant perspective on pattern recognition.
［1］魏宁康，杜元元. 注重外周血细胞形态学检查降低血液病漏诊和误诊［J］. 检验医学与临床,2013,10(9):1166-1167.
［2］Khashman A. Investigation of different neural models for blood cell type identification［J］. Neural Computing & Applications, 2012(21):1-7.
［3］Fu KS, Albus JE, Anderson RH. Syntactic pattern recognition and application ［M］. New Jersey: PrenticeHall, 1982:152-164.
［4］傅京孙. 模式识别及其应用［M］. 北京：北京大学出版社,1990:20-28.
［5］Jaume G, Ernest V, Horst B. Feature selection on node statistics based embedding of graphs ［J］. Pattern Recognition Letters, 2012, 33(15):1980-1990.
［6］Anand R. Revisioning the unification of syntax, semantics and statistics in shape analysis ［J］. Pattern Recognition Letters, 2013,42(12)： 2301-2314.
［7］Frank TA, Jason M,Kinser, et al., A neural bridge from syntactic to statistical pattern recognition ［J］. Neural Networks, 1999, 12(3):519-526.
［8］Impoco G, Tuminello L, Fuca N. Segmentation of structural features in cheese micrographs using pixel statistics ［J］. Computers and Electronics in Agriculture, 2011, 79(2):199-206.
［9］Slavakis K, Giannakis G, Mateos G. Modeling and optimization for big data analytics: (statistical) learning tools for our era of data deluge［J］. IEEE Signal Process Mag, 2014, 31: 18-31.
［10］Hong Wenxue,Yu Jianping,Cai Fu, et al. A new method of attribute reduction for decision formal context［J］. ICIC Express Letters Part B:Applications,2012,3(5): 1061-1068.
［11］郝连旺，洪文学，李婷. 基于属性层次关系的白细胞图像类间特异特征选取方法研究［J］. 生物医学工程学杂志,2014,31(6):1202-1206.
［12］郝连旺，洪文学. 基于多颜色空间特征融合的彩色白细胞图像识别［J］. 生物医学工程学杂志,2013,30(5):909-913.
［13］Ganter B,Wille R. Formal Concept Analysis:Mathematical Foundantions［M］.New York: SpringerVerlag, 1999:42-65.
［14］Poelmans J, Elzinga P,Viaene S, et al. Formal concept analysis in knowledge discovery: a survey［J］.Conceptual Structures: from Information to Intelligence,2010:139-153.
［15］Hong Wenxue, Li Shaoxiong, Yu Jianping, et al. A new approach of generation of structural partialordered attributes diagram ［J］. ICIC Express Letters, 2012, 3(4):2035-2042.
［16］张勇. 彩色白细胞显微图像分析与识别［D］. 西安:西安交通大学,1999.
［17］汤学民，林学訚，何林. 白细胞图像自动识别系统的研究［J］. 生物医学工程学杂志,2007,24(6):1250-1255.
［18］周颖颖. 彩色白细胞图像的特征选择与分类识别［D］. 南京:东南大学,2006.
