Verfahren zum Komprimieren und Dekomprimieren von standardisierten Portraitbildern

摘要

The present method and apparatus forms an addressable collection of standardized image features, representing features found in a selected class of portrait images. A portrait image to be compressed is standardized to be of the same size and feature orientation as the image features making up the collection of standardized features. The standardization process also incarnates a luminance balancing feature to minimize the number of illumination variables that could exist among the images to be compressed and the images used to establish the addressable collection of common features in a class. With that accomplished, the to-be-compressed image is compared, feature-by-feature with the features in the addressable collection of features and the best matching features from the collection are associated with the to-be-compressed image by bits representing addresses of the respective features in the collection. The number of bits needed to accomplished this representation is reduced by the use of vector quantizing techniques and by the recognition that many of the features have symmetry, vertical and/or horizontal, about the central axes of an image such as a face. That is, eyes and ears appear on the right and the left of the faces' center. Recording of only one set of bits to represent the right feature along with an indicator that the feature should be flipped to form the left feature reduces the number of bits. Linkage of like features also provides for a reduction in the number of bits. <IMAGE>