局部搜索的音频数据检索

PDF下载HTML

Title:

An audio data retrieval method based on local search trees

文章编号:

1673-4785(2008)03-0259-06

作者:

李应;

福州大学数学与计算机科学学院; 福建福州;

Author(s):

LI Ying

College of Mathematics and Computer Science; Fuzhou University; Fuzhou 350108; China

关键词:

音频数据检索;  局部搜索树;  小波包最好基;  塔型算法

Keywords:

audio data retrieval;  local search tree（LS-kdtree）;  wavelet packet best base;  pyramidal algorithm

分类号:

TP391.3;TP311.12

文献标志码:

A

摘要:

根据多媒体音频数据的特点,提出一种适用于快速音频数据检索的局部搜索数据结构,即局部搜索树(local search tree,LS-tree).在局部搜索树中,分别以音频数据小波变换系数的过零率和平均幅度作为主、次关键码,基于局部范围对作为索引的其他系数进行组织.其次,基于局部搜索树,提出采用小波包最好基小波塔型算法实现音频数据检索.最后,把采用局部搜索树的小波包最好基—小波塔型算法的搜索和基于小波不同级系数的检索方法相比较,结果表明,这种方法对音频数据检索的快速和有效性.

Abstract:

For fast audio data retrieval, a local search tree (LS2kd tree) structure is p roposed according to the char2 acteristics of multimedia audio data. In an LS2kd tree, the zero2kd crossing rate and the average magnitude of the wavelet transform coefficients in the audio data are taken as main and secondary key codes respectively, and the other coefficients used for the index are organized in the local range. On the basis of thisLS2kd tree, an audio data retrieval app roach is p resented using the wavelet packet best base and the wavelet pyramidal algorithm. Finally, the research results obtained from the p roposed app roach are compared with those obtained using the different2kd level wavelet transform coefficients. It is found that the p roposed app roach is effective and fast for audio data retrieval