论文部分内容阅读
目的用手势控制家电是智能家居发展的趋势之一,传统的静态手势识别算法难以适应复杂的居家环境,特别当使用广角相机或环境干扰大时,为此提出一种动态的挥手识别算法,可以对视频序列中的挥手动作做出响应,以达到控制家电的目的。方法挥手动作具有周期性且频率相对稳定,算法首先调整长滤波器和短滤波器使其检测到视频内周期性运动的区域,然后利用人手识别算法对周期性运动区域进行验证并确认人手。结果通过与主流的手势识别算法的对比,在复杂环境下,本文算法将成功次数提高了3%,误触发次数降低了44%,响应时间也降低了近0.4 s。结论实验结果表明,算法能够满足实际应用需求。此外,算法不基于运动目标检测,运算量极低,可以在较高的图像分辨率下实时运行,并能被移植到嵌入式平台下。
The purpose of using gestures to control home appliances is one of the trends in the development of smart home. The traditional static gesture recognition algorithms are difficult to adapt to complex home environments, especially when using wide-angle cameras or the environment is disturbing. To this end, a dynamic waving recognition algorithm Responds to the waving motion in the video sequence to achieve the purpose of controlling appliances. The method waving is cyclical and the frequency is relatively stable. The algorithm first adjusts the long filter and the short filter to detect the periodic motion in the video. Then, the algorithm verifies and confirms the periodic motion region by using the human hand recognition algorithm. Results Compared with the mainstream gesture recognition algorithms, the proposed algorithm improves the number of successes by 3%, reduces the number of false triggers by 44% and the response time by 0.4s in a complex environment. Conclusion The experimental results show that the algorithm can meet the practical application requirements. In addition, the algorithm is not based on the moving object detection, the computation is very low, it can run in real time with higher image resolution, and can be transplanted to the embedded platform.