Image Splicing Localization Using SLIC and Noise Level Estimation
With the development of computer technology and artificial intelligence technology,the authenticity of images has been seriously challenged.Therefore,how to determine the authenticity of images has become an important research direction in the field of digital images.At present,image splicing is one of the most common methods of image tampering.Different sources of images have different noise levels,and this paper proposes a method to locate the image splicing tamper region by detecting the inconsistency of the local noise level of the image.Firstly,the image to be detected is segmented into pixel blocks with similar features by using SLIC superpixel segmentation algorithm.Secondly,the noise level estimation method based on principal component analysis is used to calculate the local noise level of each image block.Finally,using three clustering algorithms to cluster the results of the estimated noise level,and the splicing tamper region of the image is located according to the clustering results.The experiment results show that the proposed method in this paper can effectively locate the splicing tamper region and retain more edge information of the detected region.
image splicing localization noise estimation superpixel segmentation algorithm clustering image forensics
Siqian LI Weimin WEI Xueling CHU Xiuru HUA
College of Computer Science and Technology,Shanghai University of Electric Power,Shanghai 200090,China
国内会议
上海
英文
42-52
2019-05-17（万方平台首次上网日期，不代表论文的发表时间）