Method, system, device and storage medium for improving image quality of tourism scenes

CN119741205APending Publication Date: 2025-04-01CTRIP TRAVEL NETWORK TECH SHANGHAI0
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411885660.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-19
Publication Date
2025-04-01

AI Technical Summary

Technical Problem

The existing image super-resolution method based on diffusion priors has problems with output instability when generating low-resolution images, resulting in different outputs generated by the same low-resolution images, making it difficult to ensure the consistency of image content.

Method used

Multi-dimensional filtration and degradation treatment methods are used to establish mapping relationship tables for low-quality and high-quality images, and train them using Latent Diffusion model to ensure the consistency of content in the image quality improvement process.

Benefits of technology

It effectively improves the quality of low-quality images, ensures the stability of output and the consistency of content, and significantly improves the stability and quality of image generation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119741205A_ABST
    Figure CN119741205A_ABST
Patent Text Reader

Abstract

The present invention provides a method, system, device and storage medium for improving the image quality of a tourist scene. The method includes: collecting pictures of tourist scenes and establishing a tourist picture collection; performing multi-dimensional filtering on the pictures in the tourist picture collection to obtain high-quality tourist pictures, at least including filtering of the preset length and width dimensions of the pictures and the picture quality threshold; degrading the filtered pictures to obtain low-quality tourist pictures, and establishing a mapping relationship table between low-quality tourist pictures and high-quality tourist pictures; inputting the low-quality tourist pictures, high-quality tourist pictures and the mapping relationship table into an image quality improvement model for training; inputting a low-quality tourist picture into the trained image quality improvement model to obtain a tourist picture with improved quality. The present invention can effectively convert low-quality pictures provided by merchants into high-quality display images, greatly improving the image quality of hotel content and improving user experience.
Need to check novelty before this filing date? Find Prior Art