Image Splicing Detection Based on Complex Features of Camera Noise Characteristics

The image captured by the camera has different camera noise compared to the splicing image. First, the splicing image and the original image are inconsistent in internal statistical characteristics. Secondly, the image splicing affects the image content, resulting in inconsistent texture information. Most importantly, the difference in the source of the pictures leads to inconsistencies between the splicing image and the original image due to camera noise. These inconsistencies in characteristics can be used as a means of image splicing detection. Therefore, this paper combines the advantages of the three methods and proposes a method for tampering detection of composite image splicing based on the three features of camera, texture and statistics. The first step is to extract the statistical features of the gray-level histogram and gray-level co-occurrence matrix of the image. The second step is to extract the LBP of the grayscale image as the texture feature. Third, we adopt the wavelet filtering-based light response non-uniform noise (PRNU) as image features, and then extract PRNU statistical information and texture features as noise features. Finally, we adopt the SVM method for classification. Experimental demonstrate the effectiveness of the proposed scheme. © 2022 Taiwan Ubiquitous Information CO LTD. All rights reserved.