三维鱼体参数化建模

因水下环境复杂，鱼类三维模型获取非常困难，基于三维鱼类数据驱动的很多自动化研究工作无法开展。为此，提出一种三维鱼体参数化建模方法。首先，用专业建模师构建的标准鱼体模板注册扫描仪采集三维鱼体数据，构建拓扑结构一致的三维鱼体网格数据集；其次，用主成分分析法对数据集中的三维鱼体网格模型进行建模分析，建立鱼体的参数化表示模型；最后，采集草鱼数据进行相关实验。实验结果表明，注册得到的三维鱼体模型与真实扫描鱼体模型顶点之间的平均均方根误差仅为0.691 3 mm，在可接受范围。通过改变三维鱼体参数化表示模型中的权参数，快速生成了大量三维鱼体数据，从而解决了三维鱼体数据获取困难的问题。

Parametric modeling of 3D fish body

The complex underwater environment makes it difficult to obtain 3D fish models, leading to the fact that many automated research work based on 3D fish data cannot be carried out. This paper proposes a 3D fish body parametric modeling method. Firstly, we use the standard fish body template constructed by a professional modeler to register the 3D fish body data collected by the scanner, and construct a 3D fish body mesh dataset with consistent topological structure; Secondly, we apply the principal component analysis method to model and analyze the 3D fish body mesh model in the dataset establishing a parameterized representation model of the fish body. We take the collected grass carp data to verify the effectiveness of the above approach. The experimental results show that the root mean square error between the vertices of the registered 3D fish model and the real scanned fish model is only 0.691 3 mm. By changing the weight parameters of the constructed 3D fish body parameterized model, a large amount of 3D fish body data can be quickly generated.