【作者】 李晓丽； 王彤； 杜振龙；

【机构】 兰州理工大学计算机与通信学院；

【摘要】 <正>1引言随着数字技术的进步,产生和积累了大量的媒体数据资料,这些媒体资料中包含着大量的、有用的信息。如何高效地索引和简化音频资料已成为亟待解决的问题。本文主要研究怎样构造音频媒体中的特征选择集。音频媒体有它自身的特性。首先,一般的音频更多还原

【Abstract】 Audio features are contained in sets of audio frames,for the huge number of frames causes audio indexing require heavy computation cost,so to how to extract those frames most concerned to audio features from audio is very significant.Feature selection is an important step in data mining which is successfully used to reduct the dimension of the features and the irrelated data.The discernibility of conventional Rough Sets is based on definite comparing, in this paper we present new discernibility definition on basis of relational comparing,and discernibility of multiple ordinal attributes,meanwhile apply it to construct audio feature selection set.Experimental evaluations indicate that the proposed algorithms could produce selection sets most faith to original data and improve the efficiency of audio indexing.更多还原