摘要
研究计算机视觉图像优化问题,针对目前三维模型检索较低的的检索精度,为改善检索结果与人视觉上保持一致,减少计算量,提出一种新的基于视觉面距离的模型检索算法。利用模型的整体特征;根据表层视觉特性构建视觉面距离模型将三维模型映射到二维模型;由于二维傅立叶算法不仅对噪音具有很好的鲁棒性,而且对几何变换具有不变性。采用二维傅立叶变换提取二维模型特征向量对特征进行融合,从而有效地评估三维模型之间的相似度。进行仿真分析,仿真结果表明,改进方法较传统其他三维模型检索有更高的精度,并与人视觉的一致性。
According to lower retrieval precision , a new algorithm based on visual distance is propese to improve the retrieval results, which is the same with human visual judgment. At first, The method of model normalization is introduced to obtain the optimal bounding box so that the three - dimensional model is mapped to the two - dimensional model. Secondly, because two -dimensional Fourier algorithm has not only good robustness to noise, but also an invariance on the geometric transformation, which is applied to extract feature vectors and integrate them, so we can assess the similarity effectively between three - dimensional models. Through statistical analysis for Precision and Recall, experimental results show that the proposed method is superior to other methods, and is the same with human visual judgment.
出处
《计算机仿真》
CSCD
北大核心
2010年第5期220-222,313,共4页
Computer Simulation
关键词
视觉面距离
二维傅立叶变换
相似度
三维模型检索
Visual distance
Two - dimensional fourier algorithm
Similarity
3 D Model retrieval
