【摘要】：傳統(tǒng)基于圖像內(nèi)容的圖像數(shù)據(jù)挖掘算法,對(duì)海量圖像特征的分類效率低,對(duì)圖像數(shù)據(jù)的挖掘準(zhǔn)確率受樣本數(shù)量影響較大。因此,提出一種基于圖像特征細(xì)化的海量數(shù)據(jù)挖掘系統(tǒng),其中的人機(jī)界面可賦予系統(tǒng)較高的交互性。圖像搜索引擎能夠智能地從互聯(lián)網(wǎng)海量的圖像數(shù)據(jù)中,采集有價(jià)值圖像數(shù)據(jù)和特征。圖像預(yù)處理模塊對(duì)圖像格式進(jìn)行變換,完成圖像噪聲因素的過濾等操作,并對(duì)采集圖像特征進(jìn)行細(xì)化。數(shù)據(jù)挖掘模塊依據(jù)采集的圖像特征細(xì)化結(jié)果塑造CMQL語句,從圖像數(shù)據(jù)庫中挖掘出有價(jià)值的圖像數(shù)據(jù)。系統(tǒng)實(shí)現(xiàn)部分給出了數(shù)據(jù)挖掘查詢語言CMQL進(jìn)行圖像數(shù)據(jù)的挖掘過程。實(shí)驗(yàn)結(jié)果表明,所設(shè)計(jì)系統(tǒng)具有較高的查準(zhǔn)率和查全率。
[Abstract]:The traditional image data mining algorithm based on image content has low efficiency for the classification of massive image features, and the accuracy of the mining of image data is affected by the number of samples. Therefore, a mass data mining system based on image feature refinement is proposed. The human-computer interface can give the system high interaction. It can collect valuable image data and features intelligently from the massive image data of the Internet. The image preprocessing module transforms the image format, completes the filtering of the image noise factors, and refines the feature of the image acquisition. The data mining module builds the CMQL statement according to the image feature refinement result. In the system realization part, the data mining query language CMQL is used to excavate the image data. The experimental results show that the designed system has high precision and recall.
【作者單位】： 福建教育學(xué)院;浙江理工大學(xué);
【基金】：國家自然科學(xué)基金(50875245)
【分類號(hào)】：TP391.41;TP311.13

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 李正;康立源;范驍輝;;中藥制藥過程數(shù)據(jù)集成、數(shù)據(jù)挖掘與可視化技術(shù)研究[J];中國中藥雜志;2014年15期

【共引文獻(xiàn)】

相關(guān)期刊論文 前5條

1 林凌;許然;;基于圖像特征細(xì)化的海量數(shù)據(jù)挖掘系統(tǒng)設(shè)計(jì)與實(shí)現(xiàn)[J];現(xiàn)代電子技術(shù);2016年24期

2 趙丹;;論中藥制藥工程科技創(chuàng)新方略及其工業(yè)轉(zhuǎn)化[J];黑龍江科技信息;2016年24期

3 李正;吳曉波;姜毅;王臣臣;趙濤;趙步長;;丹紅注射液數(shù)字制藥平臺(tái)設(shè)計(jì)與建設(shè)[J];天津中醫(yī)藥;2015年12期

4 遠(yuǎn)立國;賈坤;孫凌霜;孫W，

本文編號(hào)：2171775