基于HSV和HE的复制粘贴篡改检测算法

针对块匹配检测算法不能准确提取图像面积较小区域、自然相似区域以及平滑区域特征的问题，提出一种基于HSV(hue,saturation,value)颜色空间和直方图均衡化算法的复制粘贴篡改精确检测与定位的方法。将待测图像从RGB(red,green,blue)颜色空间转换到HSV颜色空间，并通过直方图均衡化算法对图像HSV颜色空间的V通道进行特征增强；提取基于圆谐波变换的特征构建特征向量；使用最近邻搜索算法对构建的特征向量进行快速随机搜索得到稠密的偏移场；采用稠密线性拟合的方法滤除错误匹配，并通过形态学操作标识定位出篡改区域。实验结果表明，所提方法能够有效降低面积较小区域、自然相似区域和平滑区域的误检与漏检像素，从而提升检测结果并准确定位出图像的篡改区域。

Algorithm of copy move forgery detection based on HSV and HE

With the problem that block matching algorithm cannot accurately extract the feature of small region, natural similar region and smooth region of the image, this paper proposes a method of copy move forgery exact detection and localization based on hue, saturation, value (HSV) color space and histogram equalization algorithm. Firstly, the image is transformed from red, green, blue (RGB) color space to HSV color space, and the feature enhancement of the V channel of the HSV color space is performed by the histogram equalization algorithm. Secondly, extract feature based on circular harmonic transforms to construct eigenvectors. Then, the nearest neighbor search algorithm is used to perform a fast random search on the constructed eigenvectors to obtain a dense offset field. Finally, a dense linear fitting is used to filter out the false matches and locate the tamper region by morphological manipulation. The experimental results show that this paper can effectively reduce false detection and missed detection pixels of small region, natural similar region and smooth region, thereby, improving detecting results and accurately locating the forgery region of the image.