颜色解耦光响应非均匀性噪声融合凸优化的图像伪造检测算法

Study on Image Forgery Detection Algorithm Based on Color Decoupled Photo Response Non-Uniformity Noise Integration with Convex Optimization

当前基于光响应非均匀性噪声的伪造检测算法使用了滤色器阵列,容易产生颜色插值噪声,严重影响了其噪声的提取精度及相关性检测的分辨率;且没有考虑源相机的相关性,借助局部像素统计决策进行目标识别,削弱了算法检测精度。提出了颜色解耦光响应非均匀性(,CD-PRNU color decoupled photo response non-uniformity)噪声融合凸优化方案的图像伪造检测算法。设计颜色解耦光响应非均匀性噪声;并构造了CD-PRNU噪声残留的数学计算模型;再嵌入Bayes原理,形成贝叶斯最小风险决策,利用图像全局像素,完成伪造目标检测。同时提出凸优化方案,将真伪决策演变为凸问题,降低算法复杂度。仿真结果显示:该算法能够更有效地检测出微小尺寸伪造目标,且其误检率更低。

Low detection precision and correlation resolution are induced by using local pixels to make decision authenticity, and ignoring the strong spatial dependence of the source in current image forgery detection algorithms based on photo response non-uniformity（CD-PRNU） noise, the small real area of forgery can not be detected. Therefore the image forgery detection algorithm based on color decoupled photo response non-uniformity noise integration with convex optimization mechanism is proposed. The CD-PRNU noise is designed, and the mathematical calculation model of CD-PRNU noise residual is constructed. The Bias minimum risk decision is formed by introducing the Bayes rules, and combining the decisions of all pixels, the forgery is detected. Taking image forgery detection into convex optimization problem, the convex optimization mechanism is designed to improve the detection efficiency. Simulation results show that this algorithm can detect the small size forgery area, and has lower detection error rate compared with current algorithm.