基于机器视觉的皮影动作交互系统设计  

Design of shadow play interaction system based on machine vision

作　　者：王帆 WANG Fan(Xi’an Vocational and Technical College,Xi’an 710017,China)

机构地区：[1]西安职业技术学院,西安710017

出　　处：《自动化与仪器仪表》2023年第2期131-136,共6页Automation & Instrumentation

基　　金：院级课题《科普类短视频在当下新媒体领域中的艺术转型与传播研究》(2022YB14)。

摘　　要：为进一步提高我国皮影表演智能化水平,提出一款基于机器视觉的皮影交互控制系统,以实现在复杂条件下进行准确、稳定的智能化皮影表演。其中,使用分块PCA+SVM算法进行人脸检测;使用改进的Camshift算法结合Kalman滤波预测算法进行人脸的跟踪。实验仿真结果表明,使用分块PCA+SVM算法能够进行准确的人脸检测,与其他识别算法相比,该算法的识别准确率最高能够达到90.21%;使用改进的Camshift算法结合Kalman滤波预测算法能够进行稳定的人脸跟踪;使用HSV颜色空间能够准确地进行人手识别。以上实验结果验证了提出的皮影动作交互系统的可行性和优越性,具有一定的实用价值。In order to further improve the intelligent level of China’s shadow play, a machine vision based interactive control system for shadow play is proposed to achieve accurate and stable intelligent shadow play under complex conditions. Among them, block PCA+SVM algorithm is used for face detection;The improved Camshift algorithm combined with Kalman filter prediction algorithm is used to track the face. The experimental simulation results show that the segmented PCA+SVM algorithm can be used for accurate face detection. Compared with other recognition algorithms, the highest recognition accuracy of this algorithm can reach 90.21%;The improved Camshift algorithm combined with Kalman filter prediction algorithm can perform stable face tracking;Human hand recognition using HSV color space can accurately identify human hands. The above experimental results verify the feasibility and superiority of the shadow action interaction system proposed in this study, which has certain practical value.

关 键 词：皮影表演 机器视觉 PCA法 MEANSHIFT算法 KALMAN滤波 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]