Image super-resolution method based on hybrid Transform super-resolution network

The invention relates to an image super-resolution method based on a hybrid Transform super-resolution network, and the method comprises the steps: firstly obtaining an image pair as a training sample, the image pair being composed of an image ILR and an image IHR; a hybrid Transform super-resolutio...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: ZHENG LING, HUANG LINSHENG, LIU YULONG, SHIM CHAN-JOON, LI SHICHANG, ZHANG SHENGYU, ZHU JINCHEN, HOU ZHIMENG, ZHANG MINGJIAN, WENG SHIZHUANG, ZHAO JINLING
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention relates to an image super-resolution method based on a hybrid Transform super-resolution network, and the method comprises the steps: firstly obtaining an image pair as a training sample, the image pair being composed of an image ILR and an image IHR; a hybrid Transform super-resolution network model is built, and the hybrid Transform super-resolution network model is built; the training sample is used to train the hybrid Transform super-resolution network model, and a trained hybrid Transform super-resolution network model is obtained; and importing an image to be processed as an LR image into the trained hybrid Transform super-resolution network model, wherein a result output by the hybrid Transform super-resolution network model is an HR image. According to the method, the hybrid Transform super-resolution network model is constructed, and parameters and performance are properly balanced; according to the method, the local information extraction capability of the network is enhanced, the glob