摘要
提出一种基于最大熵功率谱估计的Hadoop云平台下网络音视频数据特征挖掘方法,实现对数据信息的高速访问。构建数据挖掘Hadoop云平台和数据挖掘访问模型,设计最大熵功率谱特征提取算法,采用分段思想将同一时间段的视音频数据进行群体分割,分段提取最大熵功率谱特征。将提取的特征信息进行维度匹配分箱和溯源处理,实现信息恢复,最终完成高速数据访问。仿真测试表明,该算法能有效地实现对网络音视频数据的特征挖掘,提高访问效率,访问响应时间较当前方法缩短明显。
An improved feature mining method of network audio data was proposed based on maximum entropy spectral esti-mation in Hadoop cloud platform. High-speed access of data information is realized. The Hadoop cloud platform is con-structed, and the extraction algorithm of maximum entropy power spectra feature is designed. The idea of segmentation is used to extract the feature. The dimensionality matching box and traceability process are used for the extracted feature. The information recovery and high-speed data access are realized. Simulation results shows that the new algorithm can effective-ly realize the feature mining of audio and video data, the access efficiency is improved greatly. Access response time is re-duced significantly.
出处
《科技通报》
北大核心
2014年第8期59-61,共3页
Bulletin of Science and Technology
基金
内江师院课题项目(08NJS-99)
