摘要
针对汽车驾驶环境中的手势交互需求,提出了一个基于改进的自生成神经网络的手势检测和识别方法。该方法主要分为手势分割、特征提取和手势识别三部分。首先采用基于粒子群优化的自生成神经网络聚类算法检测并分割图像中的手势区域,然后提取手势的特征信息并构造特征向量,最后通过训练自生成神经网络生成分类神经树识别手势类型。该方法对手势检测与识别的各个阶段进行了优化,实验结果表明,该方法能达到较高的识别精度,是一个可行高效的手势检测与识别方法。
To improve the accuracy and efficiency of gesture recognition in vehicular environments,we put forward a gesture recognition algorithm based on improved self-generating neural networks.This algorithm consists of three main stages:gesture segmentation,feature extraction and gesture recognition.First,a self-generating neural network clustering algorithm based on particle swarm optimization is used to detect and segment the gesture region in the image.Then the feature information is extracted and the feature vectors are constructed.Finally,the neural network is trained to generate the gesture types.The algorithm has optimized the stages for gesture recognition,and can identify gestures quickly and accurately.Experimental results show that the proposed algorithm can achieve higher recognition accuracy,and is a feasible,efficient and accurate gesture recognition method.
作者
雷蕾
赵涓涓
史曜华
LEI Lei ZHAO Juanjuan SHI Yaohua(College of Computer Science and Technology, Taiyuan University of Technology, Taiyuan 030024, China College of Communication Engineering ,Xidian University, Xi' an 710126 ,China)
出处
《太原理工大学学报》
CAS
北大核心
2016年第6期793-798,共6页
Journal of Taiyuan University of Technology
基金
国家自然科学基金资助项目(61540007
61373100)
虚拟现实技术与系统国家重点实验室项目(BUAA-VR-15KF02)
关键词
人机交互
手势识别
自生成神经网络
粒子群算法
特征提取
human-computer interaction
hand gesture recognition
self-generating neural networks
particle swarm optimization
feature extraction