Exposing splicing forgery based on color temperature estimation. (August 2018)

• Record Type: Journal Article
• Authors: Sun, Peng; Lang, Yubo; Fan, Shu; Shen, Zhe; Liu, Lei; Shan, Daguo; Peng, Silong
• Is Part Of: Forensic science international. Volume 289(2018)
• Page Start: 1
• Page End: 11
• Publication Date: 2018-08
• Subjects: Image authentication -- Manipulation localization -- Color shift -- Color temperature estimation; Medical jurisprudence -- Periodicals; Chemistry, Forensic -- Periodicals; Forensic Medicine -- Periodicals; Médecine légale -- Périodiques; Chimie légale -- Périodiques; Gerechtelijke geneeskunde; Gerechtelijke chemie; Gerechtelijke psychiatrie; Chemistry, Forensic; Medical jurisprudence; Electronic journals; Periodicals; Electronic journals; 614.1
• DOI: 10.1016/j.forsciint.2018.04.049
• Languages: English
• ISSNs: 0379-0738

Abstract

Highlights: Color temperature based image authentication method is provided. Random Restricted algorithm is exploited to calculate reference color temperature. OSTU method is applied to further reduce human interactions. Abstract: Splicing is one of the most common tampering techniques for image manipulation in many forensic cases. Normally color shift in images due to color temperature of illumination can be seen as intrinsic features relative to imaging process. In splicing forgeries, copied area and pasted target image come from different imaging process, and are likely to have different color shift. In this paper, a novel automated authentication method is presented to expose splicing manipulation and locate manipulated areas by discriminating the inconsistencies of color shift in an image. In order to minimize human interaction on detection of splicing forgeries as well as localization of manipulated areas, a forensic image is divided into blocks with grid-based strategy. After calculation on color temperature of each blocks with white-point algorithm, reference color temperature is obtained with a random restricted algorithm. Then color temperature distance between each block and reference area is calculated sequentially. At last, by comparing color temperature distance with an optimized threshold determined by OSTU algorithm. This method enables us to judge if splicing has occurred and furthermore localize manipulated area simultaneously. Experiments show that the proposed method can speed up the quantitative detection of possible splicing manipulation and localize manipulated area automatically.