【摘要】 为了提高模糊润饰篡改取证技术的准确率和召回率,提出新的盲取证技术。采用能量比和信息熵构建模糊边缘属性约简规则来检测源图像自然强边缘;设计保持边缘平滑的滤波结构增强源图像模糊区域,对滤波图像进行全边缘检测;设计人工模糊隶属度函数,计算源图像人工模糊隶属度集合;对所得边缘图像进行融合,利用人工模糊隶属度集合对融合图像进行处理,得到准确清晰的人工模糊边缘。实验结果显示,本文算法盲取证准确率和召回率较高,能较好区别人工模糊和自然模糊,篡改区域定位效果较好。更多还原

【Abstract】 A new forensics technology was proposed in order to improve the accuracy and recall rate vague of evidence tampering retouching techniques.By detecting the origin image’s naturally strong edges on attribute reduction rules built by energy ratio and information entropy,using edge preserving smoothing filter structure to enhanced fuzzy region and detect the image’s edges,designing artificial fuzzy membership function to compute every pixel’s fuzzy membership and use it to detect the origin image’s fuzzy edges,merging all the images and processing them through artificial collection of fuzzy membership,the accurately and clearly artificial fuzzy edges were obtained.Experimental results showed that the new algorithm gets a higher precision and recall rate,distinguishes artificial and natural blur better,and detects tampered region better.更多还原