作者: Stefan Moser , Michael Schroeder
DOI: 10.1117/12.463430
关键词: Image (mathematics) 、 Image processing 、 Computer vision 、 Meta element 、 Feature detection (computer vision) 、 Still camera 、 Artificial intelligence 、 Computer science 、 Contextual image classification 、 Digital image processing 、 Automatic image annotation
摘要: In this contribution we show how meta information is included in almost every digital still camera (DSC) image can be utilized for classification and automatic enhancement a photo finishing system. All DSC manufacturers have realized the importance of tag all models support tags. However, nowhere literature found an application tags building Here, way to use tagged purpose. The additional about capture conditions allows us classify each earlier learned classes apply more appropriate processing form optimized algorithms. We such system could work also explain obtained restrict plausibility scene type classification. Finally present experimental results on general performance