人机交互手势特征识别及边缘修复算法研究  

Research on Human-computer Interaction Gesture Feature Recognition and Edge Repair Algorithm

作　　者：梁英[1] LIANG Ying(Xi'an Railway Vocational&Technical Institute,Xi'an 710000 China)

机构地区：[1]西安铁路职业技术学院,陕西西安710000

出　　处：《自动化技术与应用》2022年第11期43-46,共4页Techniques of Automation and Applications

摘　　要：如何精确地从人脸区域中检测并分离出手势部分,是人机交互领域中重要的研究部分及研究热点。人手的纹理与脸的纹理高度相似,对手势识别的稳定性造成很大的影响。针对人脸与手会出现重叠的现象,本文提出一种基于计算机边缘修复的手势识别算法,在Chamfer距离匹配算法的基础之上,通过边缘修复算法修复人脸与手重合交界处的模糊边缘处。研究结果显示,边缘修复算法的正确率高达94.6%,而其他算法的正确率仅为8.7%,该算法对头部运动以及手势变换频繁的情况下等具有较强的鲁棒性。How to accurately detect and separate the gesture part from the face area is an important research part and research hotspot in the field of human-computer interaction.The texture of the human hand is highly similar to the texture of the face,which has a great influence on the stability of gesture recognition.Aiming at the situation of face and hand overlap,this paper proposes a comput-er-based edge repair algorithm based on Chamfer distance matching algorithm.Based on the Chamfer distance matching algo-rithm,the edge repair algorithm is used to repair the blurred edge of the intersection of face and hand.The research results show that the correct rate of the edge repair algorithm is as high as 94.6%,while the correct rate of other algorithms is only 8.7%.The algorithm is robust to head motion and frequent gesture changes.

关 键 词：手势识别 手势分割 边缘修复 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]