目的 从保护和传承传统皮影文化的角度，使用Kinect设备和带有颜色标识的皮影签子，提出一种类似于传统皮影的交互方式来控制数字皮影。方法 首先，利用Kinect获得的深度图以及彩色图并配合训练已得的颜色标识，计算出签子头部的位置，且为每个签子设置卡尔曼滤波器的控制，为皮影动画提供较为稳定的控制信息。其次，将数字皮影组织成3根签子控制特有的关节层次模型。在当前模型下，结合输入的控制信息分析皮影的动作原理，通过OpenGL绘制技术，最终呈现出皮影动画。结果 实验中，用户手执皮影签子面对Kinect，操纵绘制在计算机上的皮影人物，随着签子的移动，皮影可以在构造出的悟空摘桃的皮影场景里进行相应的任意移动、挥舞胳膊、腿部下跪弯曲等动作。用户调查表明该操控方式与传统皮影类似，但该方式更简单，皮影动画也美观有趣味性。结论 仿真传统皮影交互的新的数字皮影的交互方式，对于想体验传统皮影戏表演的非专业人士来说，操作更加容易，且相比于其他数字皮影的交互更有继承传统皮影交互的力度。

Objective The traditional shadow play is a cultural treasure of Chinese folk art, combining sculpture, painting, drama, music, performances, and other arts, and it was once considered as a spiritual food. However, studies show that traditional shadow play is on the verge of extinction. Therefore, to protect and uphold the culture of traditional shadow play, numerous researchers combine shadow play with current computer technologies, inherit the shadow charming artistry, and create the possibility of a live shadow protection-digital shadow play. In this study, we manipulate the digital shadow play and the traditional shadow play interaction using Kinect and color shadow play control rods, respectively. We also analyze the essential technologies for the input of the interaction and the output of the digital shadow play actions. Method According to the characteristics of the Kinect depth map and control rod movements, the regions for the control rods are segmented to serve as mask basis in detecting the rod head position in a frame. The colors marks in the control rods are trained in advance using the Nave Bayesian method to obtain a corresponding color category table. These marks are identified while a single frame is detected to calculate the rod head positions. Each rod is set through the control of the Kalman filter to provide stable control information for shadow animation. After analyzing the traditional shadow play action characteristics and the human body joints, we organize the digital shadow figure into a joint hierarchical model by controlling the three rods. We analyze the action principles of the current model combined with the input control information in terms of moving, waving arms, and changing legs for the shadow play figure. The shadow animation is then rendered by OpenGL to present to the user. Result Facing the Kinect, a user holds three rods to control the shadow play figure rendered by the computer. Along with the rods' movements, the shadow play figure correspondingly acts through movements, such as arbitrarily moving, waving arms, and changing legs, in the shadow scene of Wukong picking peaches. User surveys show that our control method is similar to the traditional shadow play, except that ours is simpler to operate. In addition, the animation in our control method is interesting and beautiful. Conclusion This study proposes a novel way to interact with the digital shadow play as a traditional interaction simulation. This strategy allows nonprofessionals to experience controlling traditional shadow figures. Compared with other digital shadow play interactions, the proposed method has more traditional shadow play culture efforts.