Temporal event clustering for digital photo collections

• DOI: 10.1145/957013.957093
• Publication Date: 2004-04-19T17:18:43Z
• Authors: Matthew Cooper; Jonathan Foote; Andreas Girgensohn; Lynn Wilcox
• Keywords: Similarity (geometry)

We present similarity-based methods to cluster digital photos by time and image content. The approach is general, unsupervised, makes minimal assumptions regarding the structure or statistics of photo collection. results for algorithm based solely on temporal similarity, jointly content-based similarity. also describe a supervised learning vector quantization. Finally, we include experimental proposed algorithms several competing approaches two test collections.