一种三维模型特征提取与压缩算法

特征提取是三维模型检索最主要的问题,提取的特征决定检索系统的性能和效率,如何对提取的特征进行有效压缩成为保证性能的关键.首先采用透视投影的方法把预处理后的模型从三维空间转换成二维图像,然后对图像进行两次一维傅立叶变换转换到频域,并对频域采样得到的高维向量进行有效压缩.算法解决了特征信息量大导致速度低下和特征信息量小导致性能低下的矛盾.实验结果表明,该算法能在速度和性能上同时取得较好的效果.

An Algorithm for Feature Extraction and Compression of 3-D Model

Feature extraction is a major problem in 3-D model retrieval and extracted features decide its performance and efficiency.Therefore,how to compress the extracted features efficiently become the keys to guarantee the retrieval＇s performance and speed.Firstly,perspective projection was used to transform the standard 3-D model to 2-D images;secondly,these images were transformed to frequency domain by Fourier Transformation to get the vectors feature;lastly,higher dimensional vectors obtained from frequency domain samples were compressed.This algorithm has solved the contradictions of low speed resulting from large information and low performance resulting from little information.The experimental results show that the solution has good effects both in speed and performance.