Abstract: A content-based classification of shapes is proposed according to the usual chinese nouns describing the shapes. A hierachical calssification is proposedd such that some simple parameters are used to classification of two-calss shapes and complex parameters to fuzzy classification of multi-class shapes. The primary experiment on 17 kinds of usual shapes is tried and the results are satisfied.

Key words: concept-based, classification of shapes, invariable moments

摘要： 基于中文自然语言中的形状名词，提出了一种基于概念的形状分类方法，利用简单的形状描述参数进行二分类与利用复杂的形状参数进行模糊分类相结合的原则，实现形状分层分类。给出了17类基本形状的分类实验结果，表明本文方法的可行性。

关键词: 基于概念, 形状分类, 不变矩