基于Kinect的虚拟试衣系统的设计与实现

发布时间：2018-03-03 17:16

本文选题：虚拟试衣　切入点：Kinect　出处：《东南大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着互联网时代到来和人机交互技术的发展,基于体感的交互系统受到了广泛关注和研究。体感虚拟试衣系统有着操作简便,交互自然等优点,能够给用户带来更好的购衣体验。以往的虚拟试衣系统大多通过用户手动输入用户身体参数来建立虚拟的用户模型,再依据这个用户模型来实现虚拟试衣,用户的代入感和试衣效果不强。在服装仿真方面,单纯采用2D贴图的服装模拟的试衣方式对用户试衣过程中的动作和角度有很大限制,并不能很好的展示试衣效果。对服装采用3D建模的服装模拟方式需要对每件实体服饰进行3D建模,存在着建模困难,成本高的问题。系统维护的成本高,产品周期长,不能很好的应用在实际项目中。本文设计实现了一套基于Kinect的虚拟试衣系统,针对上述问题,本文在系统中进行了一些新的尝试,主要有以下几点。1.通过将服装2D贴图在关节处进行适当裁剪,然后根据用户具体动作进行旋转变换,然后拼接而成,这样用户试衣过程中可以做出一些动作来提高试衣效果。2.将透视变换算法应用在服装图像上,并跟踪用户骨骼关节点距离和深度值来计算用户转身角度,二者结合来实现了试衣过程中服装的立体效果。3.调用Kinect工具类库来进行体感捕获,获取用户骨骼数据和手部轮廓信息实现动作识别和手势判定的功能,包括如何模拟鼠标移动来操作菜单,体感按压按钮的实现按钮点击事件,手部抓握动作的识别实现列表滚动的功能,提升了人机交互效果。通过系统试衣效果测试表明,系统有着良好的试衣效果,可满足用户日常试衣的基本要求。
[Abstract]:With the advent of the Internet era and the development of human-computer interaction technology, the interaction system based on body sensation has received extensive attention and research. In the past, most of the virtual fitting system built virtual user model by manually input the user's body parameters, and then realized the virtual fitting according to this user model. The effect of substitution and fitting is not strong. In the aspect of clothing simulation, the method of clothing simulation based on 2D map has a great limitation on the movement and angle of the user in the process of fitting. It is difficult to model and cost high. The cost of system maintenance is high, and the product cycle is long. This paper designs and implements a set of virtual fitting system based on Kinect, aiming at the above problems, this paper makes some new attempts in the system. The main points are as follows: 1. By cutting the 2D map of the garment at the joint properly, then rotating and transforming according to the specific action of the user, then stitching it together. In this way, the user can do some actions to improve the fitting effect. 2. The perspective transform algorithm is applied to the clothing image, and the distance and depth of the user's skeletal node are tracked to calculate the turning angle of the user. The two methods are combined to realize the stereoscopic effect of clothing in the process of fitting. 3. The function of motion recognition and gesture determination is realized by calling Kinect tool class library to capture the body sensation, to obtain the user's skeleton data and the information of the contour of the hand, and to realize the function of motion recognition and gesture determination. It includes how to simulate mouse movement to operate menu, how to realize button click event of body feeling pressing button, how to recognize hand grip action to realize the function of list rolling, and how to improve the effect of human-computer interaction. The system has good fitting effect and can meet the basic requirements of daily fitting.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP391.9

【参考文献】