基于扫描数据的个性化动态数字人体模型生成方法研究

发布时间：2018-06-20 23:45

本文选题：数字人体模型 + 三维人体扫描　；参考：《哈尔滨工业大学》2014年硕士论文

【摘要】：数字人体模型作为计算机辅助人体工学领域重要的工具，可应用于产品与系统设计、工效学分析、性能评估与故障诊断等。传统的数字人体建模技术多采用参数化手段，，建模过程需要人体尺寸统计学规律作为外部约束条件，并需要建立庞大的人体模板数据库，存在开发周期长、成本高昂且难以保证模型逼真度等问题。近年来，随着三维测量技术的发展，出现了基于三维人体扫描数据的建模方式，此类方法无需外部约束条件或人体模板，构建的模型具有较高的逼真度。目前，基于扫描数据所生成的模型多不具备运动性能或运动性能较差。针对该问题，本文提出了一种基于扫描数据的个性化数字人体运动模型生成方法，该方法包括模型结构设计与模型构建两部分。首先，介绍了数字人体模型结构的设计方法。整个模型为骨架-皮肤双层结构：内层的骨架模型为链式结构，具有多个关节与自由度，用于整个模型的运动驱动；外层的皮肤模型由三角网格组成，用于表达模型的体表信息。模型的运动控制分为两部分：对骨架模型进行刚性变形控制；对皮肤模型采用刚性变形与局部变形结合的变形控制方法。其次，介绍了数字人体模型的构建方法。通过从扫描数据中提取的人体骨骼特征点计算骨架模型的各关节位置，生成骨架模型。以骨架模型的各骨骼段为单位，将外部扫描数据分段，筛选出各段扫描数据中的特定位置、特定数量的数据点构造该骨骼段的外部皮肤网格。最后，介绍了基于本文所提方法的数字人体模型生成软件。该软件可实现数字人体模型的生成、运动控制与图形化显示。利用该软件完成了数字人体模型构建实验并对实验结果进行了测试与分析。实验分析表明，本文方法所构建的模型具有如下优点：具有较好的运动性能，可以模拟真实人体较为复杂的姿态变化，且在运动过程中具有良好的皮肤变形特性；与真实对象具有较高的几何相似度；具有较小的数据量与运算量，可支持实时运动控制。该模型在计算机辅助人体工学领域的运动仿真与姿态模拟研究中具有较大的应用价值。
[Abstract]:As an important tool in the field of computer aided human engineering, digital human model can be used in product and system design, ergonomics analysis, performance evaluation and fault diagnosis. The traditional digital human body modeling technology mostly uses parameterized means, the modeling process needs the human body size statistics law as the external constraint condition, and needs to establish the huge human body template database, the existence development period is long, The cost is high and it is difficult to guarantee the fidelity of the model. In recent years, with the development of 3D measurement technology, a modeling method based on 3D human body scanning data has emerged. This method does not require external constraints or human templates, and the constructed model has a high fidelity. At present, most models based on scanning data have no or poor motion performance. To solve this problem, this paper proposes a personalized digital human motion model generation method based on scanning data, which includes two parts: model structure design and model construction. Firstly, the design method of digital human model structure is introduced. The whole model is a skeleton-skin double layer structure: the inner skeleton model is a chain structure with multiple joints and degrees of freedom, which is used to drive the whole model, and the outer skin model is composed of triangular meshes. The body surface information used to express the model. The motion control of the model is divided into two parts: rigid deformation control for skeleton model and deformation control method combining rigid deformation and local deformation for skin model. Secondly, the construction method of digital human body model is introduced. The joint position of the skeleton model is calculated by extracting the human skeleton feature points from the scanning data, and the skeleton model is generated. Taking each skeleton segment of the skeleton model as a unit, the external scanning data is segmented, and the specific position and number of data points in each segment scan data are selected to construct the external skin mesh of the bone segment. Finally, the digital human model generation software based on the proposed method is introduced. The software can realize the generation of digital human body model, motion control and graphical display. The software is used to build the digital human body model and the experimental results are tested and analyzed. Experimental analysis shows that the model constructed by this method has the following advantages: it has better motion performance, can simulate the complex posture changes of real human body, and has good skin deformation characteristics in the process of motion; It has high geometric similarity with real objects, small amount of data and computation, and can support real-time motion control. The model has great application value in the research of motion simulation and attitude simulation in the field of computer aided ergonomics.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TB18;TP391.7

【参考文献】