Methods and devices for coding and decoding images, computer program implementing them and information carrier enabling their implementation

A method of decoding a digital image, that comprises, to form at least one current macroblock of a higher resolution level of said image, a step of determining at least one low frequency coefficient on the basis of a macroblock, designated as “second predictor”, of a resolution level lower than the resolution level of the current macroblock to form, a step of decoding at least one motion vector, a step of determining at least one first predictor on the basis of each decoded motion vector and of at least one image designated as “reference”, a step of determining at least one high frequency coefficient on the basis of coefficients of at least one transform, to the frequential domain, of at least one predictor representing at least one first predictor and a step of transforming, to the spatial domain, said low and high frequency coefficients.

A device for coding a digital image, that comprises a motion estimator that performs an estimation of motion between at least one macroblock designated as “current” of said image and at least on image designated as “reference”, to obtain at least one first predictor in each reference image, a means for obtaining, for said current macroblock, a second predictor of a spatial scalability level lower than the spatial scalability level of the current macroblock, a means for transforming, from the spatial domain to the frequential domain, at least one predictor representing at least one first predictor, a low frequency predictor for predicting at least one low spatial frequency coefficient on the basis of the second predictor and a high frequency predictor for predicting, for the current macroblock, at least one high spatial frequency coefficient on the basis of at least one high frequency coefficient of at least one predictor representing at least one first predictor transformed to the frequential domain.

A device for decoding a digital image that comprises a means for determining at least one low frequency coefficient on the basis of a macroblock designated as “second predictor”, of a resolution level lower than the resolution level of a current macroblock of the image to form, a decoder for decoding at least one motion vector for the current macroblock, a means for determining at least one first predictor for the current macroblock, on the basis of each decoded motion vector and of at least one image designated as “reference”, a means for determining, for the current macroblock, at least one high frequency coefficient on the basis of coefficients of at least one transform, to the frequential domain, of at least one predictor representing at least one first predictor and a means for transforming, for the current macroblock, to the spatial domain, said low and high frequency coefficients.