基于OpenCV的静态手势识别研究  

Research on static gesture recognition based on OpenCV

作　　者：赵家玮 董玉华 ZHAO Jiawei;DONG Yuhua(School of Information and Communication Engineering,Dalian Minzu University,Dalian 116605,Liaoning,China)

机构地区：[1]大连民族大学信息与通信工程学院,辽宁大连116605

出　　处：《智能计算机与应用》2024年第12期190-194,共5页Intelligent Computer and Applications

摘　　要：搭载手势识别的高科技产品在人们日常生活中发挥着重要作用。针对手势识别过程中背景繁杂等问题,本文设计了基于OpenCV的手势识别系统。该系统采用基于YCrCb空间Cr分量的OTSU法选取阈值对手势进行分割及二值化,去除手势的背景信息;在使用Canny边缘检测算子的基础上,使用傅里叶描述子作为手势的特征向量,把手势轮廓数字化;将提取的手势特征通过SVM分类器构建手势模型,从而识别手势动作所代表的数字信息。实验结果表明这一手势识别系统能够有效提取手势轮廓并且具有较高的准确率,有助于实现人机交互的自然化和人性化。High-tech products equipped with gesture recognition play an important role in people′s daily life.A gesture recognition system based on OpenCV is designed to solve the problem of complicated background in gesture recognition.The OTSU method based on the Cr component of YCr Cb space is used to select the threshold to segment and binarize the gesture,and remove the background information of the gesture.On the basis of using Canny edge detection operator,the Fourier descriptor is used as the feature vector of gesture,and the gesture contour is digitized.The extracted gesture features are used to construct a gesture model by SVM classifier,so as to identify the digital information represented by gesture actions.The experimental results show that this gesture recognition system can effectively extract gesture contours and has a high accuracy,which is helpful to realize the naturalization and humanization of human-computer interaction.

关 键 词：OPENCV 图像处理 特征提取 SVM 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]