Source Detection Of Mobile Phone Pictures Based On Pattern Noise

Digital image source has been an important branch in the field of image forensics.With the development of technology,more and more powerful camera equipment,especially mobile phone camera is becoming more and more popular.Therefore,the detection of pictures of the device is also becoming more important.Camera reference pattern noise is widely used in the detection of image source.However,camera reference pattern noise is computed by averaging sensor pattern noise value of all image,not considering each image proportion.In our paper we propose an improved camera reference pattern extraction algorithm for image source based on the characteristics of noise variance and difference image have difference texture and lightness.First we extract all image sensor pattern noise,then remove interpolation point to extract image’s really sensor pattern noise according to the characteristics of the interpolation point have smaller noise variance.We use the product of image texture and lightness as the weighted of each image sensor pattern noise in camera reference pattern noise.Finally,we use a weighted averaging to estimate camera reference pattern noise.The experimental results show that our method achieve good performance for image source detection.In addition,this paper is aimed at a lot of algorithms in the image scene is the night of the situation will fail,we propose the night scene of the image source detection.Firstly,the image is divided into several regions,then the images are divided into 4 categories according to the texture and brightness of the image.Processing each category separately.To give the texture complex block extraction pattern noise to suppress the interference of the texture,and the brightness of the dark scene to enhance the pattern noise.And the brightness of the dark scene and texture complex block extraction pattern noise should to suppress the interference of texture and enhance pattern noise.