Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Low complexity shape-adaptive DCT for coding of arbitrarily shaped image segments

: Sikora, T.


Image communication 7 (1995), No.4-6, pp.381-95
ISSN: 0923-5965
Journal Article
Fraunhofer HHI ()
adaptive codes; computational complexity; discrete cosine transforms; image coding; image segmentation; statistical analysis; transform coding; video coding; low complexity shape-adaptive dct; generic input sequences; arbitrarily shaped image segments; transform algorithm; transform efficiency; gilge method; standard mpeg-1 coder; segment based coding; object based coding; content-based functionality; intra coding; mpeg-4

A low complexity shape-adaptive DCT transform algorithm for coding pels in arbitrarily shaped image segments is presented. The proposed algorithm is compared to the well established generalized shape-adaptive transform method introduced by Gilge et al. (1989) in terms of transform efficiency and computational complexity. Results obtained under both theoretical and experimental conditions show that the new algorithm achieves a transform efficiency close to that of the Gilge method with considerably reduced computational complexity. The proposed shape-adaptive DCT algorithm was implemented into a standard MPEG-1 coder to provide object or segment based coding of images and video with additional content-based functionality. The extended MPEG-1 object based coding scheme can handle generic input sequences and can readily provide MPEG-1 backward compatibility if no contour data is transmitted for a given video sequence. Results for INTRA coding of images indicate that the algorithm allows efficient coding over a wide range of coding parameters-thus pro