本文選題：偽造圖像 + 圖像噪聲　； 參考：《計(jì)算機(jī)應(yīng)用研究》2017年01期

【摘要】：針對不同來源合成偽造數(shù)字圖像提出了一種盲檢測方法,不同數(shù)字圖像背景噪聲存在差異,因而偽造圖像區(qū)域噪聲方差不同。從待測圖像小波分解后的高頻子帶中去除相應(yīng)邊緣區(qū)域的高頻干擾,改進(jìn)了噪聲方差估計(jì)算法,并對所得噪聲圖像進(jìn)行分塊處理估計(jì)每一個(gè)分塊的噪聲方差,將方差相近的塊進(jìn)行融合,最后比較圖像中紋理接近的同質(zhì)區(qū)域,找出方差異常的位置。通過實(shí)驗(yàn)研究了方差估計(jì)精度,對不同來源的偽造圖像進(jìn)行了檢測,結(jié)果表明算法提高了圖像噪聲方差的估計(jì)精度,在圖像紋理接近的同質(zhì)區(qū)域中可以定位圖像的偽造區(qū)域。
[Abstract]:A blind detection method is proposed for synthesizing fake digital images from different sources. Because of the difference of background noise in different digital images, the regional noise variance of forged images is different. The noise variance estimation algorithm is improved by removing the high frequency interference in the corresponding edge region from the high frequency subband of the wavelet decomposition of the image to be tested, and the noise variance of each block is estimated by the block processing of the noise image. The blocks with similar variances are fused. Finally, the homogenous regions of the images with close texture are compared to find out the location of the variance anomalies. The precision of variance estimation is studied experimentally and the false image from different sources is detected. The results show that the estimation accuracy of noise variance is improved and the false region of image can be located in the homogeneous region of image texture.
【作者單位】： 西北工業(yè)大學(xué)計(jì)算機(jī)學(xué)院;
【基金】：陜西省科技統(tǒng)籌創(chuàng)新工程資助項(xiàng)目(2015KTTSGY04-05,2015KTZDGY04-01)
【分類號】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前9條

1 陳宗民;周治平;;噪聲方差和紋理復(fù)雜度分析的源相機(jī)識別[J];計(jì)算機(jī)應(yīng)用;2012年06期

2 曲巍崴;高峰;;基于噪聲方差估計(jì)的小波閾值降噪研究[J];機(jī)械工程學(xué)報(bào);2010年02期

3 李天翼;王明輝;吳亞娟;�；�;;圖像噪聲方差的小波域估計(jì)算法[J];北京工業(yè)大學(xué)學(xué)報(bào);2012年09期

4 劉鵬;劉定生;;基于噪聲方差來計(jì)算變分圖像反卷積的規(guī)整化參數(shù)[J];光學(xué)學(xué)報(bào);2009年09期

5 劉艷麗;郭延文;王進(jìn);彭群生;;噪聲方差未知的小波域中非局部均值圖像去噪[J];計(jì)算機(jī)輔助設(shè)計(jì)與圖形學(xué)學(xué)報(bào);2009年04期

6 徐鑫;田逢春;姬艷麗;龍紅梅;;基于小波系數(shù)層間相關(guān)性的圖像噪聲方差估計(jì)[J];計(jì)算機(jī)應(yīng)用;2009年10期

7 賴施成;賈l，

本文編號：1792060