基于ABC优化的集成分类手势识别方法研究  

Research on Integrated Classification Gesture Recognition Method Based on ABC Optimization

作　　者：缑新科[1] 王润 GOU Xin-ke;WANG Run(Lanzhou university of technology,Lanzhou 730050 China;Gansu Provincial Key Laboratory of Industrial Process Control,Lanzhou 730050 China)

机构地区：[1]兰州理工大学,甘肃兰州730050 [2]甘肃省工业过程先进控制重点实验室,甘肃兰州730050

出　　处：《自动化技术与应用》2022年第6期112-116,共5页Techniques of Automation and Applications

摘　　要：本文首先提取方向梯度直方图(Histogram of Oriented Gradients,HOG)的特征,再结合主成分分析算法(Principal Component Analysis,PCA)寻找最佳的特征维度。在此基础上提出了一种集成分类器,即以SVM、K临近算法(K-Nearest Neighbor,KNN)以及极端梯度提升(Extreme Gradient Boosting,XGBoost)算法为基分类器,通过人工蜂群(Artificial Bee Colony algorithm,ABC)算法为这三种分类器进行优化,从而形成一个性能更优的集成分类器。实验表明与单一分类器相比,在数据集Hand Postures中该集成分类器可以获取更好的手势识别效果。The paper firstly extracts the features of Oriented Gradients(HOG), and then combines Principal Component Analysis(PCA) to find the best characteristic dimension. Based on this, an integrated classifier is proposed, which takes SVM, K-nearest Neighbor(KNN) and Extreme Gradient Boosting(XGBoost) as the base classifiers, and the Artificial Bee Colony Algorithm(ABC) algorithm is used to optimize the three classifiers, so as to form an integrated classifier with better performance. Experiments show that the integrated classifier can obtain better gesture recognition effect in Hand Postures in data set compared with single classifier.

关 键 词：HOG PCA 人工蜂群算法 集成分类器 手势识别 

分 类 号：TP391.44[自动化与计算机技术—计算机应用技术]