Recognising an image by transforming it using fishers linear discriminant analysis, generating an image key and comparing it with other keys.

摘要

A camera produces a 3D natural image. The data space occupied by the image is reduced by Fisher's Linear Discriminant Analysis (FLDA) and the reduced image provides an image key representative of the image. The image key is compared with stored image keys of known images, and the output comparison is sorted to produce a final list of potential matches. In a verification process, just a single image key may be compared.