A novel approach for computing and pooling structural similarity index in the discrete wavelet domain

Rezazadeh, Soroosh et Coulombe, Stéphane. 2009. « A novel approach for computing and pooling structural similarity index in the discrete wavelet domain ». In 16th IEEE International Conference on Image Processing (ICIP) (Cairo, Egypt, Nov. 7-12, 2009) pp. 2209-2212. Piscataway, NJ, USA : Institute of Electrical and Electronics Engineers.
Compte des citations dans Scopus : 14.

Prévisualisation	PDF Coulombe S. 2009 3021 A novel approach for computing and pooling structural similarity index.pdf - Version acceptée Télécharger (316kB) | Prévisualisation
Prévisualisation	PDF Coulombe S. 2009 3021 ppt A novel approach for computing.pdf - Présentation Licence d'utilisation : Tous les droits réservés aux détenteurs du droit d'auteur. Télécharger (2MB) | Prévisualisation

Résumé

The structural similarity (SSIM) index is an objective metric that gives relatively accurate similarity prediction scores with reasonable complexity. In this paper, an excellent trade-off between accuracy and complexity is presented in the form of a wavelet structural similarity index (WSSI), which is more accurate and less complex than the spatial SSIM index. Like the spatial SSIM index, the WSSI has the feature of boundedness. It computes an edge structural similarity map and an approximation structural similarity map to obtain the final similarity score. A contrast map is introduced in the wavelet domain for pooling structural similarity maps. Experimental results show that the low-complexity WSSI gives a correlation coefficient of 0.9548 between objective and subjective scores, and competes with visual information fidelity (VIF) performance.

Type de document:	Compte rendu de conférence
ISBN:	9781424456536
Professeur:	Professeur Coulombe, Stéphane
Affiliation:	Génie logiciel et des technologies de l'information
Date de dépôt:	30 oct. 2012 18:19
Dernière modification:	05 juill. 2017 13:37
URI:	https://espace2.etsmtl.ca/id/eprint/3021

Actions (Authentification requise)

Dernière vérification avant le dépôt