A novel image dataset for source camera identification and image based recognition systems

• DOI: 10.1007/s11042-022-13354-5
• Publication Date: 2022-06-17T07:23:36Z
• Authors: Andrea Bruno; Paola Capasso; Giuseppe Cattaneo; Umberto Ferraro P...; Riccardo Improta
• Keywords: Identification; Benchmarking; Digital camera

Abstract Multimodal emotion recognition has attracted a great deal of attention in recent years, with new interesting applications now being considered. One promising application is the digital image forensics fields where, for example, it gives possibility to automatically highlight subjects that are pain, images under examination, by analyzing their facial expressions. However, finding an represents possible crime leaves problem identifying device used take open. Such been addressed Source Camera Identification algorithms (SCI, short). These analyze some features hidden target find traces left sensor captured image. A particularly challenging case when candidate source cameras investigation same manufacturer and model. fair universal assessment these only if standard datasets benchmarking. our comprehensive analysis shown majority proposed so far contain collection taken different types cameras, mostly smartphones. We fill this gap presenting UNISA2020, novel dataset contains large real-world multiple conventional type. The have assembled as avoid artifacts could negatively affect identification process. To validate dataset, we also performed comparative experimental investigate performance SCI reference algorithm running on well other datasets.