| Titre : |
Nonparametric approach to digital image restoration |
| Type de document : |
texte imprimé |
| Auteurs : |
Fadli, Djamel-Eddine, Auteur ; Djeddi, M., Directeur de thèse |
| Editeur : |
Institut National d'Electricité et d'Electronique INELEC |
| Année de publication : |
1995 |
| Importance : |
92 f. |
| Présentation : |
ill. |
| Format : |
30 cm. |
| Note générale : |
Mémoire de Magister : Électronique : Boumerdès, Institut National d’Électricité et d’Électronique : 1995
Bibliogr. f. I - IV . Annexe [7] f |
| Langues : |
Anglais (eng) |
| Mots-clés : |
Nonparametric approach Digital image restoration Kernel regression smoothing of noisy images estimation density |
| Index. décimale : |
M004895 |
| Résumé : |
Image analysis is concerned with restoration and interpretation of images that have been contaminated by noise and possibly some (nonlinear transformation). The purpose of our research work is to implement the statistical nonparametric methods to the restoration of noisy images which are characterized by the range of possible values for the intensity function evaluated at different regular points of a lattice plane D. The nonparametric approach is more easier to be mastered by engineers than the parametric one, and it does not require a priori strong assumptions concerning the unknown model of the image. The first contribution consists of regarding the restoration of a given degraded image as the estimation of the intensity function from the observed degraded colors. This task was accomplished by the use of the kernel method of regression. It was noticed that the implemented method does not perform satisfactorily in the regions where the intensity function fluctuates rapidly (sharp changes in the coloring). This drawback can be reduced to a certain extent by using the variable kernel method. In order to overcome the drawback of the curve fitting approach, a stochastic nonparametric approach was developed. In this approach, the value of the intensity function at each pixel was regarded as an observation generated from a random variable which is locally dependent on the neighboring random variables. Therefore, the restoration of images reduces to the estimation of an unknown stochastic process using one of the suggested nonparametric methods. The first method considers the observed image as a realization of a spatial process, whereas the second one is based on the estimation of the density function of the true unknown color of a pixel using the neighboring principle. The two developed methods have been found to be very efficient. Furthermore, their performances were found to be better than that of the parametric weiner approach. |
Nonparametric approach to digital image restoration [texte imprimé] / Fadli, Djamel-Eddine, Auteur ; Djeddi, M., Directeur de thèse . - Institut National d'Electricité et d'Electronique INELEC, 1995 . - 92 f. : ill. ; 30 cm. Mémoire de Magister : Électronique : Boumerdès, Institut National d’Électricité et d’Électronique : 1995
Bibliogr. f. I - IV . Annexe [7] f Langues : Anglais ( eng)
| Mots-clés : |
Nonparametric approach Digital image restoration Kernel regression smoothing of noisy images estimation density |
| Index. décimale : |
M004895 |
| Résumé : |
Image analysis is concerned with restoration and interpretation of images that have been contaminated by noise and possibly some (nonlinear transformation). The purpose of our research work is to implement the statistical nonparametric methods to the restoration of noisy images which are characterized by the range of possible values for the intensity function evaluated at different regular points of a lattice plane D. The nonparametric approach is more easier to be mastered by engineers than the parametric one, and it does not require a priori strong assumptions concerning the unknown model of the image. The first contribution consists of regarding the restoration of a given degraded image as the estimation of the intensity function from the observed degraded colors. This task was accomplished by the use of the kernel method of regression. It was noticed that the implemented method does not perform satisfactorily in the regions where the intensity function fluctuates rapidly (sharp changes in the coloring). This drawback can be reduced to a certain extent by using the variable kernel method. In order to overcome the drawback of the curve fitting approach, a stochastic nonparametric approach was developed. In this approach, the value of the intensity function at each pixel was regarded as an observation generated from a random variable which is locally dependent on the neighboring random variables. Therefore, the restoration of images reduces to the estimation of an unknown stochastic process using one of the suggested nonparametric methods. The first method considers the observed image as a realization of a spatial process, whereas the second one is based on the estimation of the density function of the true unknown color of a pixel using the neighboring principle. The two developed methods have been found to be very efficient. Furthermore, their performances were found to be better than that of the parametric weiner approach. |
|
M000100 
Ajouter le résultat dans votre panier
Visionner les documents numériques
Faire une suggestion Affiner la rechercheNonparametric approach to digital image restoration (1995)
