Clustering techniques for personal photo album management

摘要: We propose a novel approach for the automatic representation of pictures achieving more effective organization of personal photo albums. Images are analyzed and described in multiple representation spaces, namely, faces, background, time capture. Faces automatically detected, rectified, represented, projecting face itself common low-dimensional eigenspace. Backgrounds represented with low-level visual features based on an RGB histogram Gabor filter bank. Faces, time, background information each image collection is automatically organized using mean-shift clustering technique. Given particular domain personal photo libraries, where most pictures contain faces relatively small number different individuals, clusters tend to be semantically significant besides containing visually similar data. report experimental results based on data set of about 1000 images detection rectification of faces lead approximately 400 faces. Significance clustering has been evaluated, very encouraging.