仿真假体视觉下基于深度图像的手势识别研究  

Research on gesture recognition based on depth image under simulated prosthetic vision

作　　者：赵瑛[1] 王冬晖 李琦[1] 于爱萍 谷宇[1,2] ZHAO Ying;WANG Donghui;LI Qi;YU Aiping;GU Yu(Inner Mongolia Key Laboratory of Pattern Recognition and Intelligent Image Processing,School of Information Engineering,Inner Mongolia University of Scienceand Technology,Baotou 014010,China;School of Computer Engineering and Science,Shanghai University,Shanghai 200444,China)

机构地区：[1]内蒙古科技大学信息工程学院内蒙古自治区模式识别与智能图像处理重点实验室,内蒙古包头014010 [2]上海大学计算机工程与科学学院,上海200444

出　　处：《现代电子技术》2019年第16期131-135,139,共6页Modern Electronics Technique

基　　金：国家自然科学基金(81460279);国家自然科学基金(61841204);内蒙古自治区自然科学基金(2018LH08066);内蒙古自治区自然科学基金(2015MS0604);内蒙古自治区高等学校科学研究项目(NJZY145);包头市科技计划项目(2015C2006-14;2017C1002)~~

摘　　要：针对仿真假体视觉下彩色图像和深度图像对于手势识别的不同效果,研究使用Kinect获取彩色图像以及深度图像进行手势识别。通过Kinect提取的骨骼信息与提取的深度图像结合,将人体与背景图像分离,对OpenCV库分离后的图像进行降噪,并进行像素化处理。在不同分辨率(32×32,48×48,64×64)下进行彩色图像和深度图像的手势识别实验。实验结果表明,随着分辨率的增加,手势识别的准确率也不断增加。同一分辨率下,深度图像下的手势识别率总体高于彩色图像下的手势识别率,且在32×32分辨率下,二者差异显著。In allusion to the different effects of gesture recognition for color image and depth image under simulated pros-thetic vision,Kinect is used to acquire color image and depth image to conduct the gesture recognition.In combination with the skeleton information and depth image obtained by Kinect,the human body is separated from background image.The separated image is denoised and pixelated by using OpenCV library.The experiments of gesture recognition of color and depth images at different resolutions(32×32,48×48,64×64)were carried out.The experimental results show that the accuracy of gesture recog-nition is increasing with the increase of resolution,at the same resolution,the gesture recognition rate of the depth image is higher than that of color image,and their difference is significant at the resolution of 32×32.

关 键 词：视觉假体 手势识别 深度图像 骨骼信息 图像降噪 像素化处理 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP301.6[电子电信—信息与通信工程]