A Multiscale Image Inpainting System Based on Wireframe and Edge Structure

Abstract

The present invention provides a multi-scale image restoration system based on wireframe and edge structure, which is used to complete structural information of images to be processed in indoor and outdoor scenes including buildings or furniture, and complete image restoration or object removal tasks , is characterized in that it includes: the wireframe and edge extraction module uses a predetermined wireframe extraction method to perform wireframe extraction on the image to be processed to obtain a wireframe image, and uses a predetermined edge extraction method to perform edge extraction on the image to be processed to obtain an edge image; structure Based on the image to be processed, the wireframe image and the edge image, the restoration module performs structural restoration through the pre-trained structural restoration model, so as to obtain the repaired image to be processed as the structural restoration image; the image processing module completes the image restoration task based on the structural restoration image or object removal tasks. The image restoration system of the present invention has better image restoration performance for artificial scenes (such as indoor furniture and outdoor buildings).

Description

technical field [0001] The invention belongs to the field of image generation, and in particular relates to a multi-scale image restoration system based on wireframe and edge structures. Background technique [0002] Image inpainting is a task of filling and patching default areas in an image. The purpose of image restoration is to ensure the authenticity and rationality of texture details in the repaired image area. And with different image masking systems, image inpainting can serve many real-world applications, such as image restoration, image editing, object removal, etc. [0003] As a research hotspot in computer vision in recent years, there have been many valuable works on image inpainting algorithms based on deep learning. Many of these works are devoted to improving the texture details of filled regions by improving the model. For example, the paper [1] proposes an image inpainting algorithm based on gated convolution, which effectively improves the image inpaint...

AI Technical Summary

Problems solved by technology

However, these auxiliary edge information are often based on gradient operations, so there are some disadvantages, such as straight-line edges are often intermittent, and due to dependence on gradients, structural information in some color-insensitive regions is lost

Images