Traitement en cours

Veuillez attendre...

Paramétrages

Paramétrages

Aller à Demande

1. CN110175953 - Image super-resolution method and system

Office
Chine
Numéro de la demande 201910439532.8
Date de la demande 24.05.2019
Numéro de publication 110175953
Date de publication 27.08.2019
Type de publication A
CIB
G06T 3/40
GPHYSIQUE
06CALCUL; COMPTAGE
TTRAITEMENT OU GÉNÉRATION DE DONNÉES D'IMAGE, EN GÉNÉRAL
3Transformation géométrique de l'image dans le plan de l'image
40Changement d'échelle d'une image entière ou d'une partie d'image
G06N 3/04
GPHYSIQUE
06CALCUL; COMPTAGE
NSYSTÈMES DE CALCULATEURS BASÉS SUR DES MODÈLES DE CALCUL SPÉCIFIQUES
3Systèmes de calculateurs basés sur des modèles biologiques
02utilisant des modèles de réseaux neuronaux
04Architecture, p.ex. topologie d'interconnexion
G06N 3/08
GPHYSIQUE
06CALCUL; COMPTAGE
NSYSTÈMES DE CALCULATEURS BASÉS SUR DES MODÈLES DE CALCUL SPÉCIFIQUES
3Systèmes de calculateurs basés sur des modèles biologiques
02utilisant des modèles de réseaux neuronaux
08Méthodes d'apprentissage
CPC
G06T 3/4053
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
3Geometric image transformation in the plane of the image
40Scaling the whole image or part thereof
4053Super resolution, i.e. output image resolution higher than sensor resolution
G06N 3/0454
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
04Architectures, e.g. interconnection topology
0454using a combination of multiple neural nets
G06N 3/08
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
08Learning methods
Déposants PENG CHENG LABORATORY
鹏城实验室
GRADUATE SCHOOL AT SHENZHEN, TSINGHUA UNIVERSITY
清华大学深圳研究生院
Inventeurs XIA SHUTAO
夏树涛
DAI TAO
戴涛
LI QING
李清
LIN DONG
林栋
WANG YI
汪漪
Mandataires 北京集佳知识产权代理有限公司 11227
Titre
(EN) Image super-resolution method and system
(ZH) 一种图像超分辨方法和系统
Abrégé
(EN) The invention provides an image super-resolution method and system, and the method comprises the steps of enabling a to-be-processed image to serve as the input of a convolutional neural network super-resolution model, and enabling the convolutional neural network super-resolution model to be composed of four execution modules which are connected in sequence, enabling the first execution module toprocess the to-be-processed image to obtain a first processed image, enabling the second execution module to process the first processing image and output a second processing image, enabling the third execution module to process the second processed image and output a third processed image, and enabling the fourth execution module to process the third processing image and output a super-resolution image. Based on the invention, the convolutional neural network hyper-division model sets the weighted features for the to-be-processed image, through learning the weighted features, the important features in the to-be-processed image are determined, and the super-resolution processing is carried out according to the important features, so that the feature expression capability of the convolutional neural network super-resolution model is improved, and the detail quality of the super-resolution image obtained after the super-resolution processing is greatly improved.
(ZH) 本发明提供一种图像超分辨方法和系统,该方案包括:将待处理图像作为卷积神经网络超分模型的输入,卷积神经网络超分模型由四个依次连接的执行模块构成;第一执行模块对待处理图像进行处理,得到第一处理图像;第二执行模块对第一处理图像进行处理,输出包含第二处理图像;第三执行模块对第二处理图像进行处理,输出第三处理图像;第四执行模块对第三处理图像进行处理,输出超分辨率图像。基于本发明,卷积神经网络超分模型为待处理图像设置加权特征,通过对加权特征的学习,确定待处理图像中的重要特征,并依据重要特征进行超分辨处理,从而提高卷积神经网络超分模型的特征表达能力,使得超分辨处理后所得到的超分辨率图像的细节质量大大提高。
Documents de brevet associés