权重自适应的多特征组合三维工程模型检索算法

摘要

To solve the issues of “semantic gap”and single feature limited search capability,an engineering triangular mesh model retrieval algorithm based on relevance feedback and features combination was proposed.Sever feature descriptors were combined for higher retrieval ability.To narrow the gap between high-level semantic knowledge and low-level content features and improve semantic relevance of retrieval results,the weight of each feature descriptor was determined dynamically using particle swarm optimization (PSO)based on user’s feedback information.Results of experiments on publicly available 3D model database engi-neering shape benchmark (ESB)suggest that the proposed approach has better retrieval ability.%针对三维模型检索领域的“语义鸿沟”以及单特征检索能力有限的问题，提出一种基于用户相关反馈信息和多特征组合的三维工程网格模型检索算法。通过联合多个特征算子增强算法检索能力，利用用户相关反馈信息指导非线性优化算法，动态分配各特征算子在相似度环节的权重，缩小模型高层语义与底层特征之间的差距，提高检索结果的语义相关性。采用工程标准模型库对算法进行验证，验证结果表明，该算法可有效提高三维工程网格模型的检索准确度。