| Titre : |
Codage des imahes vidéo couleur par compensation temporelle et spectrale |
| Type de document : |
texte imprimé |
| Auteurs : |
Mohammed Mekideche, Auteur ; Mohammed Khamadja, Directeur de thèse |
| Editeur : |
Constantine : Université Mentouri Constantine |
| Année de publication : |
2007 |
| Importance : |
73 f. |
| Note générale : |
01 Disponible à la salle de recherche 02 Disponibles au magazin de la B.U.C. 01 CD |
| Langues : |
Français (fre) |
| Catégories : |
Français - Anglais
Electronique
|
| Tags : |
Compensation Codage Image vidéo couleur |
| Index. décimale : |
621 Electronique |
| Résumé : |
AbstractIn term of video coding, the standards MPEGx and H.26x use the YUV model of colour images spectrally decorrelated, after a linear transformation of RGB images. For spatial-temporal decorrelation, these standards adopt the Hybride method. This method is based on temporal compensated prediction of current image from preceding image using the motion vector (MV), and transform coding of residual image. In the case of still multispectral images, there are several methods developped for spatial decorrelation of the neighbouring pixels in the same band, and spectral decorrelation of the pixels in the same position into different bands. There is some algorithms of compression use vector quantization. Other algorithms based on combining orthogonal transforms such as (DCT, DWT, KLT) . And others combine predictive methods intra or inter bands with transform coding. In fact, RGB images are three bands multispectral images, to which the preceding algorithms can be applied. But, these algorithms consider only multispectral images. In our work, we have developed a video coder for RGB model of colour images, based on spectral method. For comparison, we had integrated to our coder a program for video coding YUV model of colour images, based on temporal compensation method. Then we obtain a coder operating with several modes. This first one MC (Motion-Compensated) mode which is coding YUV image with temporal compensation shows the best results that proofs the temporal compensated efficiency. This algorithm is used as a reference to compare our algorithms with The second mode SC ( Shift- compensated) which Codes RGB images using shift vectors ( between pixels value in the same position in G and R bands on the one hand, and in B and R bands on the other hand), yields good results. Then, we improved the SC from which we derived the new mode denoted by SC* (where the shift vectors are ignored in coding phase because, they can be deducted at decoder from motion vector and reference frame). Then, the performances of SC* mode are more close to those for MC mode. |
| Diplôme : |
Magistère |
| En ligne : |
../theses/electronique/MEK4753.pdf |
| Permalink : |
index.php?lvl=notice_display&id=2717 |
Codage des imahes vidéo couleur par compensation temporelle et spectrale [texte imprimé] / Mohammed Mekideche, Auteur ; Mohammed Khamadja, Directeur de thèse . - Constantine : Université Mentouri Constantine, 2007 . - 73 f. 01 Disponible à la salle de recherche 02 Disponibles au magazin de la B.U.C. 01 CD Langues : Français ( fre)
| Catégories : |
Français - Anglais
Electronique
|
| Tags : |
Compensation Codage Image vidéo couleur |
| Index. décimale : |
621 Electronique |
| Résumé : |
AbstractIn term of video coding, the standards MPEGx and H.26x use the YUV model of colour images spectrally decorrelated, after a linear transformation of RGB images. For spatial-temporal decorrelation, these standards adopt the Hybride method. This method is based on temporal compensated prediction of current image from preceding image using the motion vector (MV), and transform coding of residual image. In the case of still multispectral images, there are several methods developped for spatial decorrelation of the neighbouring pixels in the same band, and spectral decorrelation of the pixels in the same position into different bands. There is some algorithms of compression use vector quantization. Other algorithms based on combining orthogonal transforms such as (DCT, DWT, KLT) . And others combine predictive methods intra or inter bands with transform coding. In fact, RGB images are three bands multispectral images, to which the preceding algorithms can be applied. But, these algorithms consider only multispectral images. In our work, we have developed a video coder for RGB model of colour images, based on spectral method. For comparison, we had integrated to our coder a program for video coding YUV model of colour images, based on temporal compensation method. Then we obtain a coder operating with several modes. This first one MC (Motion-Compensated) mode which is coding YUV image with temporal compensation shows the best results that proofs the temporal compensated efficiency. This algorithm is used as a reference to compare our algorithms with The second mode SC ( Shift- compensated) which Codes RGB images using shift vectors ( between pixels value in the same position in G and R bands on the one hand, and in B and R bands on the other hand), yields good results. Then, we improved the SC from which we derived the new mode denoted by SC* (where the shift vectors are ignored in coding phase because, they can be deducted at decoder from motion vector and reference frame). Then, the performances of SC* mode are more close to those for MC mode. |
| Diplôme : |
Magistère |
| En ligne : |
../theses/electronique/MEK4753.pdf |
| Permalink : |
index.php?lvl=notice_display&id=2717 |
|
Affiner la recherche Interroger des sources externes
