基于简约三轴切片视图的三维建模算法数据

在三维建模和3D打印应用中，通过从XYZ三轴的切片视图生成完整的三维模型，适用于复杂的多层结构生成。此方法通过分析三轴切片视图的简约线条图特征，以实现三维结构的重构，可广泛应用于工业设计、3D打印和教育领域。1.数据收集：在工业设计和3D内容生成应用中，收集设计师绘制的XYZ三轴方向的切片图。这些切片图涵盖模型的多角度信息，能更全面地捕捉复杂结构细节。每个三维模型包含三维渲染图、三轴切片渲染图、简约三轴切片线条图，以保证多视角信息用于后续的三维重建。

2.数据预处理：在预处理阶段，对简约三轴切片线条图进行特征提取，以生成多角度的特征向量。首先，将每个轴的切片图分别编码，生成特征向量F_xslice, F_yslice和F_zslice，过程为F_xslice = Encoder_x(XSlice_img), F_yslice = Encoder_y(YSlice_img), F_zslice = Encoder_z(ZSlice_img)，其中Encoder_x、Encoder_y和Encoder_z分别是用于提取各方向特征的预训练模型，XSlice_img、YSlice_img和ZSlice_img是三轴方向上的CAD模型切片图像。然后将这些特征向量结合形成一个统一的特征向量F_3axis = [F_xslice, F_yslice, F_zslice]，以便后续用于模型输入。

3.模型构建：构建一个多输入的深度学习模型，该模型使用三轴切片的联合特征向量F_3axis作为输入，以生成完整的三维结构。模型包含多通道的编码解码模块，分别解码X、Y、Z轴特征，生成三维结构，公式如下：Volume_gen = Decoder_3axis(F_3axis)，其中Decoder_3axis为解码生成三维结构的模块。之后计算平均CD（倒角距离）和平均Voxel_IoU（体素交并比），以评估模型性能。

In 3D modeling and 3D printing applications, complete 3D models are generated from slice views along the X, Y, and Z axes, which is suitable for the generation of complex multi-layer structures. This method realizes 3D structure reconstruction by analyzing the features of simplified line drawings of triaxial slice views, and can be widely applied in industrial design, 3D printing and education fields.
1. Data Collection: In industrial design and 3D content generation applications, slice maps drawn by designers along the X, Y, and Z axes are collected. These slice maps cover multi-angle information of the model, which can comprehensively capture the details of complex structures. Each 3D model includes 3D renderings, triaxial slice renderings, and simplified triaxial slice line drawings, so as to ensure that multi-view information is used for subsequent 3D reconstruction.
2. Data Preprocessing: In the preprocessing stage, feature extraction is performed on the simplified triaxial slice line drawings to generate multi-angle feature vectors. First, the slice maps of each axis are encoded separately to generate feature vectors F_xslice, F_yslice and F_zslice. The process is as follows: F_xslice = Encoder_x(XSlice_img), F_yslice = Encoder_y(YSlice_img), F_zslice = Encoder_z(ZSlice_img), where Encoder_x, Encoder_y and Encoder_z are pre-trained models used to extract features in each direction respectively, and XSlice_img, YSlice_img and ZSlice_img are the slice images of CAD models along the triaxial directions. Then these feature vectors are combined to form a unified feature vector F_3axis = [F_xslice, F_yslice, F_zslice] for subsequent model input.
3. Model Construction: A multi-input deep learning model is constructed, which uses the joint feature vector F_3axis of triaxial slices as input to generate a complete 3D structure. The model includes multi-channel encoder-decoder modules, which decode the features of X, Y, and Z axes respectively to generate a 3D structure. The formula is as follows: Volume_gen = Decoder_3axis(F_3axis), where Decoder_3axis is the module for decoding and generating the 3D structure. Then the average CD (Chamfer Distance) and average Voxel_IoU (Voxel Intersection over Union) are calculated to evaluate the model performance.

搜集汇总

数据集介绍

特点

该数据集包含6277条三维建模相关数据，格式为xlsx，用于通过XYZ三轴切片视图生成完整三维模型，适用于工业设计、3D打印和教育领域。数据集由湖州创感科技有限公司自行产生，包含三维模型编号、渲染图、切片图等字段，并采用多输入深度学习模型进行三维结构生成。