1. A Copula Based-Approach for Image Splicing Detection
- Author
-
Aniq Atiqi Rohmawati and Rimba Whidiana Ciptasari
- Subjects
Pixel ,business.industry ,Computer science ,Copula (linguistics) ,Probability density function ,Pattern recognition ,Image processing ,Transformation (function) ,Joint probability distribution ,Computer Science::Computer Vision and Pattern Recognition ,RGB color model ,Artificial intelligence ,Marginal distribution ,business - Abstract
Measuring properties in pixels to reveal an image description is of great essential for image processing. One of obstacles in image splicing detection is an unavailability of complete-reference image. In this study, a statistical method is proposed using a cumulative joint distribution of colours pixel. Consider an image consisting of particular pixels with RGB (Red, Green, Blue) components, each colour provides unique behaviour of probability function that implicitly sticking in marginal distributions. Copulas provide a cumulative joint distribution that consist of the univariate marginals and load their dependency structure, since the deficiency of linear dependency under non-linear strictly transformation. Copulas can determine the distinguish the cumulative joint distribution in the edge boundaries of image splicing. According to experimental results, Gaussian and Archimedean copulas provide deeper description to detect the contrast copula parameters of splicing image distribution.
- Published
- 2021