Two-stage image restoration method based on texture structure perception
A repair method and technology for structural diagrams, which are applied in image enhancement, image analysis, image data processing, etc., and can solve problems such as low image quality, blurred texture details, and structural distortion.
Pending Publication Date: 2021-05-14
BEIJING UNIV OF TECH
View PDF0 Cites 6 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, most of these existing methods fail to simultaneously generate plausible structures and fine texture details
[0004] In summary, the existing image inpainting algorithms generate images with low quality, distorted structures, and blurred texture details, which have certain limitations.
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0046] In order to describe the technical content of the present invention more clearly, further description will be given below in conjunction with specific examples:
[0047] Example results of the present invention are as figure 1 shown.
[0048] In this invention, we propose a two-stage architecture for image inpainting, which divides the image inpainting task into a structure generation network and an image completion network. Each network is based on a generative confrontation network, and the generator part consists of an encoder, The residual block and decoder are composed, and the discriminator is designed according to the PatchGAN architecture, which effectively solves practical problems in image restoration. The frame diagram of the present invention is as figure 2 shown.
[0049] The mask, structure map and grayscale image of the damaged image are input into the encoder composed of three convolutional layers, and then input to the decoder composed of three conv...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention relates to a two-stage image restoration method based on texture structure perception, which is used for solving the technical problem of image restoration, and specifically comprises two stages: an image structure generation stage for generating structure information of an image missing region; in the image content generation stage, the obtained structure information is used as prior, and the texture and color information of the damaged image is combined to complete the image; each stage corresponds to one generative adversarial network, and the two generative adversarial networks have the same structure; in the image structure generation stage, a grey-scale image, a structure image and a mask of a damaged image are input into an encoder which is trained in the image structure generation stage, and a reconstructed structure image is obtained through nine residual blocks and a decoder; and an image content generation stage: inputting the structure diagram obtained by reconstruction, the damaged image and the mask of the damaged image into an encoder trained in the image content generation stage, and obtaining a repaired image through nine residual blocks and a decoder in sequence.
Description
Technical field: [0001] The invention relates to the field of computer image processing, in particular to a two-stage image restoration method based on texture structure perception. Background technique: [0002] Image inpainting is the process of filling missing regions with visually true and semantically similar content. This is a classic and challenging image processing topic widely used in image editing, image-based rendering, and computational photography. Traditional patch-based image inpainting methods search and copy the best matching patch from known regions to missing regions. This traditional image inpainting method works well on static textures, but has limited effect on textures with complex or non-repetitive structures such as faces, and is not suitable for capturing high-level semantic information. [0003] In recent years, methods based on convolutional neural networks have achieved great success in the field of image inpainting. Pathak et al. first traine...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06T5/00G06T7/00G06N3/08
CPCG06T7/0002G06N3/08G06T2207/20081G06T2207/30168G06T5/77
Inventor 王瑾张熙王琛高颖朱青
Owner BEIJING UNIV OF TECH