Unlock instant, AI-driven research and patent intelligence for your innovation.

image inpainting method

A repair method and image technology, applied in the field of image repair, can solve problems such as poor local detail repair effect, and achieve the effect of satisfying subjective feelings, improving repair effect, and improving image quality

Active Publication Date: 2021-11-16
CHENGDU SHULIANYUNSUAN TECH CORP
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the above methods are all to give people a clear feeling in the overall visual effect of the image, and the repair effect on the local details of the object in the image is not good.

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • image inpainting method
  • image inpainting method
  • image inpainting method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0041] Please refer to figure 1 , figure 1 It is a schematic structural diagram of an image restoration neural network. Embodiment 1 of the present invention provides an image restoration method. The method includes: inputting an image to be repaired into an image restoration neural network, and using the image restoration neural network to process the image to be repaired , to obtain the repaired image;

[0042] The image restoration neural network includes:

[0043] A rough repair sub-network, where the rough repair sub-network is used to perform overall repair processing on the image to be repaired to obtain a rough repair image;

[0044] A feature sub-network, where the feature sub-network is used to extract target features from the rough inpainted image to obtain a first feature vector map;

[0045] Segmentation sub-network, the segmentation sub-network is used to extract each component image of the target from the rough repair image, and obtain a segmentation map of t...

Embodiment 2

[0070] Please refer to image 3 , image 3 It is a schematic flow chart of applying the image repair neural network in the present invention to carry out image repair, and the specific method is:

[0071] Data annotation:

[0072] Data annotation is the process of artificially labeling the parts of the object in the image. In the embodiment here, an image containing an airplane will be taken as an example, and it is assumed that the image size is ,in is the scaling factor. In the process of data labeling, it is necessary to label each component in the aircraft image, for example: fuselage, left and right wings, and left and right tail, a total of five components. And mark the more important key points, for example: four key points of nose, tail, left and right wings. The number of the above components and key points is not unique and depends on personal judgment.

[0073] data preprocessing

[0074] Data preprocessing is the process of processing images and labeling ...

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

No PUM Login to View More

Abstract

The invention discloses an image restoration method, which relates to the field of image processing, comprising: inputting an image to be repaired into an image repair neural network to process the repaired image; repair processing, to obtain a rough repair image; a feature sub-network, used to extract target features from the rough repair image, to obtain a first feature vector map; a segmentation sub-network, used to extract each component image of the target from the rough repair image, Obtain the segmentation map of the target; the key point sub-network is used to extract the key point coordinates from the rough repair image, and obtain the key point map based on the key point coordinates; the fine repair sub-network is used to convert the first feature vector map, the segmentation map Fusion with the key point map to obtain the repaired image; the present invention focuses on the restoration effect of the local details of the target in image super-resolution reconstruction, and can improve the restoration effect of the local details of the target.

Description

technical field [0001] The present invention relates to the field of image processing, in particular to an image restoration method. Background technique [0002] The purpose of the digital image super-resolution reconstruction task is to improve the quality of the image, and to improve the human visual effect by using a software algorithm to transform one or more frames of image reconstruction into a higher resolution image or video. technology. Due to the limitation of technical process, cost or shooting status and other factors, noise and blurring will appear on the image during the imaging process, resulting in image degradation. The image super-resolution reconstruction algorithm can appropriately and flexibly increase the quality of the imaged image, and has played an important role in many fields such as military applications, medical analysis, and public security. In the task of digital image super-resolution reconstruction, the input is a low-quality (low-resoluti...

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
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06T5/00G06T3/40G06T7/10G06T9/00G06N3/04G06N3/08
CPCG06T3/4038G06T3/4046G06T3/4053G06T7/10G06T9/002G06N3/04G06N3/08G06T2207/20081G06T2207/20084G06T2207/20221G06T5/77
Inventor 不公告发明人
Owner CHENGDU SHULIANYUNSUAN TECH CORP