人体关键点升维算法及其在游戏交互中的应用  

The human key points dimension-ascending algorithm and its application in game interaction

作　　者：李长鑫[1] 黄海于[1] LI Changxin;HUANG Haiyu(School of Computing and Artificial Intelligence,Southwest Jiaotong University,Chengdu 610031,China)

机构地区：[1]西南交通大学计算机与人工智能学院,成都610031

出　　处：《智能计算机与应用》2022年第9期143-148,153,共7页Intelligent Computer and Applications

摘　　要：针对传统基于RGB图像的人体关键点检测算法只能得到关键点2D坐标的问题,本文提出了一种基于双目视觉的人体关键点升维方法。该方法首先对2个RGB摄像头同时拍摄的人体图像进行关键点检测,然后根据检测到的人体关键点的2D坐标,结合摄像头的内外参数,采用几何计算的方法,将人体关键点从像素坐标系下的2D坐标升维到世界坐标系下的3D坐标。该方法可在不需要RGB-D相机和激光雷达等硬件的情况下,实现人体关键点的3D定位。该算法应用到第三人称游戏中,可实现游戏内玩家角色的运动控制及人机交互的目的,在实际测试中达到了较好的演示效果。Aiming at the problem of traditional human key point detection algorithm that can only obtain the 2 D coordinates of the human key points in the pixel coordinate system, a dimension-ascending algorithm based on binocular vision is proposed. The method firstly detects the 2 D position of human key points in the images captured by the two RGB cameras at the same time. Then combined with the internal and external parameters of the camera through geometric calculation, the 2 D position can be transformed to the 3 D position. This method does not require expensive hardware such as RGB-D cameras and lidars and can get high precision position of human key points in 3 D coordinate. The algorithm can be used in third-person games, which could achieve the purpose of human-computer interaction. Through interaction test, it gets the good demonstration effect.

关 键 词：人体关键点检测 升维方法 第三人称游戏 人机交互 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]