Multi-frame adaptive fusion video super-resolution method based on deep learning
A deep learning and super-resolution technology, applied in the field of video super-resolution algorithms, can solve the problems of difficult image registration, can not make full use of redundant information of adjacent frames, affect user experience and other problems, achieve strong robustness, and solve image matching problems. The effect of increasing the difficulty of quasi-difficulty and avoiding flickering
Pending Publication Date: 2020-02-28
TIANJIN UNIV
View PDF0 Cites 4 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, this multi-frame super-resolution algorithm based on a fixed number of frames has the following two problems: 1) When the image content difference between adjacent frames is very large, if the selected number of frames is too large, it will cause problems for image registration. It brings great difficulties, and the fused video is prone to bad flickering phenomenon, which affects the user experience; 2) When the number of frames is too small, the redundant information of adjacent frames cannot be fully utilized
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
[0030] The mathematical model and specific implementation of the video super-resolution algorithm based on deep learning based on multi-frame adaptive fusion of the patent will be described in detail below with reference to examples and accompanying drawings. The specific flow chart is shown in Figure 4 gives:
[0031] The first step is to construct the data set required for training the network of the present invention, that is, the video in the Vimeo-90k video data set is read frame by frame into an image and saved, which is recorded as a high-resolution image set Y HR , and then convert the high-resolution image set Y through matlab HR Each image is downsampled to get the corresponding low-resolution image set Y LR .
[0032] The second step is to build a multi-frame adaptive fusion video super-resolution network through the deep learning framework TensorFlow. like figure 1 As shown, this figure is the overall framework of the network of the present invention, and the ...
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 provides a multi-frame adaptive fusion video super-resolution method based on deep learning and electronic equipment thereof. The method comprises the following steps of: 1, constructinga data set required for training a network of the invention; 2, establishing a multi-frame adaptive fusion video super-resolution network through a deep learning framework TensorFlow, dividing a multi-frame adaptive fusion video super-resolution network into two parts, namely a multi-frame adaptive fusion video super-resolution network and a frame adaptive fusion video super-resolution network, wherein the multi-frame adaptive registration network can distort adjacent frames of a key frame needing super-resolution, so that the content of the adjacent frames tends to be the same as that of thekey frame so as to provide more detail information for the algorithm, and the super-resolution network super-resolutions the output of the multi-frame adaptive registration network into a high-resolution frame image; training.
Description
technical field [0001] The invention relates to a video super-resolution algorithm based on a convolutional neural network, and relates to a multi-frame adaptive fusion video image registration algorithm. Background technique [0002] High-resolution video will bring users a clearer and more comfortable visual experience, so the related technical research has also received extensive attention from scholars. As a new technology for obtaining high-definition images at low cost, video super-resolution technology, which has developed rapidly in recent years, contains huge commercial value in many industries such as security, finance, and modern logistics, and has become a cutting-edge technology that many large companies are competing for. The basic task of super-resolution technology is to reconstruct the corresponding high-resolution (HR) image or video from the original low-resolution (LR) image or video, which is a typical pathological condition. question. At present, scho...
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): G06T3/40G06T7/246G06T7/33
CPCG06T3/4053G06T7/251G06T7/344G06T2207/10016G06T2207/20081G06T2207/20084G06T2207/20221Y02T10/40
Inventor 曾明马金玉吴雨璇李祺王湘晖
Owner TIANJIN UNIV