正在加载图片...
第14卷第2期 智能系统学报 Vol.14 No.2 2019年3月 CAAI Transactions on Intelligent Systems Mar.2019 D0:10.11992/tis.201804028 网络出版地址:http:/kns.cnki.net/kcms/detail/23.1538.TP.20180516.1009.002.html 基于质心分水岭算法的静态手势分割算法模型 董旭德,许源平',舒红平',张朝龙2,卢丽,黄健 (1.成都信息工程大学软件工程学院,四川成都610225,2.英国哈德斯菲尔德大学计算与工程学院,西约克 郡哈德斯菲尔德HD13DH) 摘要:为了解决在类肤色背景下难以从图像中高效地分割出完整静态手势的问题,提出了基于质心分水岭算 法(improved centroid watershed algorithm,ICWA)的静态手势分割模型。该ICWA算法可以有效地减少图像梯度 对手势分割的影响并完整地提取出肤色区域。此外,本文设计了一种将PCA(principal component analysis)降维 和凸性检测算法相结合的方法,可以根据对凸点准确提取手腕的割线。同时,利用卷积神经网络(convolutional neural networks,.CNN)在标准数据库上进行了初步的手势自动识别实验。实验结果表明:该分割模型对于9种 静态手势的平均识别率达到了97.85%。 关键词:类肤色背景;静态手势分割;ICWA算法:手腕分割:手势识别;凸性检测;PCA降维:深度学习 中图分类号:TP18文献标志码:A文章编号:1673-4785(2019)02-0346-09 中文引用格式:董旭德,许源平,舒红平,等.基于质心分水岭算法的静态手势分割算法模型J.智能系统学报,2019,14(2): 346-354. 英文引用格式:DONG Xude,,XU Yuanping,SHU Hongping,.etal.Static gesture segmentation algorithm model based on centroid watershed algorithmJ].CAAI transactions on intelligent systems,2019,14(2):346-354. Static gesture segmentation algorithm model based on centroid watershed algorithm DONG Xude',XU Yuanping'SHU Hongping ZHANG Chaolong,LU Li',HUANG Jian (1.School of Software Engineering,Chengdu University of Information Technology,Chengdu 610225,China;2.School of Comput- ing and Engineering,University of Huddersfield,Huddersfield HD1 3DH,UK) Abstract:Considering the difficulty in effectively achieving complete static gesture segmentations from skin-like back- ground regions,this paper proposes an integrated static gesture segmentation model based on an improved centroid wa- tershed algorithm(ICWA).The ICWA algorithm significantly reduces the interference of image gradient on gesture seg- mentations such that it can completely extract skin regions from images.Moreover,a novel method is designed and im- plemented by integrating principal component analysis(PCA)dimension reduction and convexity detection algorithms, which can accurately extract the cutting line of the wrist according to convex points.Preliminary experiments of auto- matic gesture recognitions based on convolutional neural network(CNN)were carried out on a benchmark database. The experimental results show that the proposed model can achieve a recognition rate of 97.85%on average for nine dif- ferent static gestures. Keywords:skin-like background;static gesture segmentation;ICWA algorithm;wrist segmentation;gesture recogni- tion;convexity detection;PCA dimension reduction;deep learning 近年来,随着计算机的发展和普及,人机交 互(human-computer interface,HCI)应用正逐渐以 收稿日期:2018-04-18.网络出版日期:2018-05-18. 更加多样化的形式全面融入到人们的学习、工作 基金项目:国家自然科学基金项目(61203172):四川省科技厅 应用基础项目(2018JY0146,2019YFH0187):深圳市 和生活中。其中,由于手势是一种出现较早,使 重大国际合作项目(GHZ20160301164521358) 通信作者:许源平.E-mail:ypxu@cuit.edu.cn 用广泛且较为统一的人类日常交流手段,所以基DOI: 10.11992/tis.201804028 网络出版地址: http://kns.cnki.net/kcms/detail/23.1538.TP.20180516.1009.002.html 基于质心分水岭算法的静态手势分割算法模型 董旭德1 ,许源平1 ,舒红平1 ,张朝龙1,2,卢丽1 ,黄健1 (1. 成都信息工程大学 软件工程学院,四川 成都 610225; 2. 英国哈德斯菲尔德大学 计算与工程学院,西约克 郡 哈德斯菲尔德 HD1 3DH) 摘 要:为了解决在类肤色背景下难以从图像中高效地分割出完整静态手势的问题,提出了基于质心分水岭算 法 (improved centroid watershed algorithm, ICWA) 的静态手势分割模型。该 ICWA 算法可以有效地减少图像梯度 对手势分割的影响并完整地提取出肤色区域。 此外,本文设计了一种将 PCA(principal component analysis) 降维 和凸性检测算法相结合的方法,可以根据对凸点准确提取手腕的割线。同时,利用卷积神经网络 (convolutional neural networks, CNN) 在标准数据库上进行了初步的手势自动识别实验。实验结果表明:该分割模型对于 9 种 静态手势的平均识别率达到了 97.85%。 关键词:类肤色背景;静态手势分割;ICWA 算法;手腕分割;手势识别;凸性检测;PCA 降维;深度学习 中图分类号:TP18 文献标志码:A 文章编号:1673−4785(2019)02−0346−09 中文引用格式:董旭德, 许源平, 舒红平, 等. 基于质心分水岭算法的静态手势分割算法模型[J]. 智能系统学报, 2019, 14(2): 346–354. 英文引用格式:DONG Xude, XU Yuanping, SHU Hongping, et al. Static gesture segmentation algorithm model based on centroid watershed algorithm[J]. CAAI transactions on intelligent systems, 2019, 14(2): 346–354. Static gesture segmentation algorithm model based on centroid watershed algorithm DONG Xude1 ,XU Yuanping1 ,SHU Hongping1 ,ZHANG Chaolong1,2 ,LU Li1 ,HUANG Jian1 (1. School of Software Engineering, Chengdu University of Information Technology, Chengdu 610225, China; 2. School of Comput￾ing and Engineering, University of Huddersfield, Huddersfield HD1 3DH, UK) Abstract: Considering the difficulty in effectively achieving complete static gesture segmentations from skin-like back￾ground regions, this paper proposes an integrated static gesture segmentation model based on an improved centroid wa￾tershed algorithm (ICWA). The ICWA algorithm significantly reduces the interference of image gradient on gesture seg￾mentations such that it can completely extract skin regions from images. Moreover, a novel method is designed and im￾plemented by integrating principal component analysis (PCA) dimension reduction and convexity detection algorithms, which can accurately extract the cutting line of the wrist according to convex points. Preliminary experiments of auto￾matic gesture recognitions based on convolutional neural network (CNN) were carried out on a benchmark database. The experimental results show that the proposed model can achieve a recognition rate of 97.85% on average for nine dif￾ferent static gestures. Keywords: skin-like background; static gesture segmentation; ICWA algorithm; wrist segmentation; gesture recogni￾tion; convexity detection; PCA dimension reduction; deep learning 近年来,随着计算机的发展和普及,人机交 互 (human-computer interface,HCI) 应用正逐渐以 更加多样化的形式全面融入到人们的学习、工作 和生活中。其中,由于手势是一种出现较早,使 用广泛且较为统一的人类日常交流手段,所以基 收稿日期:2018−04−18. 网络出版日期:2018−05−18. 基金项目:国家自然科学基金项目 (61203172);四川省科技厅 应用基础项目 (2018JY0146, 2019YFH0187);深圳市 重大国际合作项目 (GJHZ20160301164521358). 通信作者:许源平. E-mail:ypxu@cuit.edu.cn. 第 14 卷第 2 期 智 能 系 统 学 报 Vol.14 No.2 2019 年 3 月 CAAI Transactions on Intelligent Systems Mar. 2019
向下翻页>>
©2008-现在 cucdc.com 高等教育资讯网 版权所有