一种面向移动端的快速手势分割优化方法

Rapid Optimization Method for Mobile Hand Segmentation

针对移动端手势分割存在的移动设备计算资源有限和手势分割易受到复杂背景干扰等问题,本文提出了一种面向移动端的快速手势分割优化方法.首先结合场景变化率对背景模型的更新速率进行自适应控制,以加强背景模型对环境变化的适应性;然后通过HSV和YCbCr色彩空间构建肤色模型,有效提取肤色区域,排除非肤色运动物体干扰;再通过局部区域定位,减少冗余计算量,同时防止将慢速移动的手部误判为背景.实验结果证明,该方法可以快速排除复杂环境干扰,较准确高效地实现手部分割.

Mobile gesture segmentation is an extremely challenge,not only because its limited computing resource,but also the complex background can easily affect the segmentation result. This study proposes a rapid optimization method for mobile hand segmentation.Firstly,combine the scene change rate and update the background model adaptively to enhance the adaptability of the background model to environmental changes. Secondly,construct skin color model by HSV and YCbCr,effectively extracting skin color regions and eliminating interference from non-skinned moving objects. Finally,reduce the amount of redundant calculations by local area positioning while preventing misunderstanding of slow-moving hand as background. Experimental results illustrate that this method has strong anti-interference ability in complex background,and it is good at practical application.