Abstract: In order to solving the difficulty of extracting characteristic of the shape of 3D model in content-based retrieval techniques, in this paper, the surface area distribution is used to express the characteristic of the shape of 3D model. Surface area distribution characteristic satisfies translation invariance, rotation invariance, symmetry invariance through using PCA. Ratios between the sum of all triangles’ surface area in one divided block and the whole area of 3D model ensure that the characteristic satisfies scale invariance. A quadratic form distance function is used to calculate the similarity between two models, which can distinguish the different position of vector components. Result shows that 3D model retrieval technique based on area distribution has a good performance.

Key words: content-based retrieval, area-based distribution retrieval, 3D model retrieval

摘要： 基于内容的3D模型检索技术难以提取形状特征，该文通过提取表面面积分布来表示3D模型的形状特征，采用主分量分析方法确保特征满足平移不变性、旋转不变性、对称不变性。以切分块内三角面片面积和与总面积的比值作为特征，确保缩放不变性。二次型形式的距离公式计算的相似度能够有效地区分特征分量的空间位置。实验表明，基于面积分布的检索技术具有较好的检索效果。

关键词: 基于内容检索, 基于面积分布检索, 3D模型检索

CLC Number: 

• TP391.41