Video deblurring method based on iterative neural network
An iterative neural network and deblurring technology, applied in biological neural network models, neural learning methods, neural architectures, etc., can solve the problems of ignoring frame-to-frame continuity, difficult application of video deblurring models, and discontinuity in time domain of videos, etc. problem, to achieve the effect of reducing the parameter scale
Active Publication Date: 2020-04-17
WENZHOU UNIVERSITY
View PDF6 Cites 7 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] However, the above methods have achieved certain results in the field of video deblurring, but most of them rely on huge convolutional neural networks, making it difficult for video deblurring models to be applied to actual scenes
In addition, the above methods often focus on how to model the timing relationship between the input frame sequence, while ignoring the frame-to-frame continuity in the generated video, resulting in a certain temporal discontinuity in the recovered video
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
[0032] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings.
[0033] Such as figure 1 As shown, in the embodiment of the present invention, a kind of video deblurring method based on iterative neural network proposed, comprises the following steps:
[0034] Step S1, building a video deblurring model; wherein, the video deblurring model includes a non-local temporal domain module, an iterative module formed based on a convolutional neural network and a recurrent neural network, and several convolutional layers;
[0035] The specific process is as figure 2 As shown, a video deblurring model is constructed. Video deblurring models include non-local temporal modules (such as image 3 shown), based on convolutional neural network (such as Figure 5 shown) and recurrent neural networks (such as Figure 4 Shown) form an itera...
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 video deblurring method based on an iterative neural network. The video deblurring method comprises the following steps: constructing a video deblurring model; obtaining an original video sequence, calculating local and global similarity between frames of the original video sequence, and further comparing modeling time domain information; performing deblurring operation onthe fuzzy sequence in the feature space according to the time domain information to obtain deblurring features; recovering the features after the deblurring operation from the feature space to a clear image sequence; and calculating a time domain loss function according to the clear image sequence and the target video sequence, and performing back propagation to train the network. Through implementation of the invention, model parameters are reduced through the coupling effect of a convolutional neural network and a recurrent neural network, time domain information is modeled by calculating the similarity between global and local frames in a video, and a more continuous and clearer image sequence is further generated by the model through a time domain loss function, so that the problems in the prior art can be solved.
Description
technical field [0001] The invention relates to the technical field of image processing, in particular to a video deblurring method based on an iterative neural network. Background technique [0002] With the development of computer processors, simple image processing tasks have been widely used in practical scenes, and researchers began to study how to obtain more information from videos to get better results. However, due to object movement or camera shaking during shooting, the input video often contains a certain degree of blurring, and these blurring contents often seriously affect the applicability of the video. Therefore, in order to make the obtained video better Application, video restoration tasks such as video deblurring are essential prerequisite tasks. [0003] Video deblurring is a classic video restoration problem. The earliest proposed video deblurring algorithm is an inverse filter (Inverse Filter) deconvolution algorithm. For example, Nathan uses a two-bi...
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): G06T5/00G06N3/04G06N3/08
CPCG06N3/084G06T2207/10016G06T2207/20221G06T2207/20081G06T2207/20084G06N3/045G06T5/73
Inventor 张笑钦蒋润华王涛王金鑫赵丽
Owner WENZHOU UNIVERSITY