A video image defogging method based on temporal label transfer
A video image and label technology, applied in the field of video image defogging based on time series label transmission, can solve the problems of inconspicuous and unsatisfactory changes in the background area, and achieve enhanced feature extraction capabilities, simple steps, temporal and spatial continuity and real-time performance Good results
Active Publication Date: 2021-10-22
ROCKET FORCE UNIV OF ENG
View PDF10 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The method based on the general defogging model can improve the solution efficiency of the video defogging algorithm. However, various defogging models rely on different assumptions, such as the background area in the video does not change significantly or the distribution of the image fog in each frame of the video is the same. For The assumption of real video is usually not satisfied, so the effect of this type of method needs to be further improved
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] Such as figure 1 As shown, a kind of video image defogging method based on time series tag transfer of the present invention comprises the following steps:
[0033] Step 1. Establish a training set of foggy video images. The process is as follows:
[0034] Step 101, utilize the clear video image data set {X of known depth i}, according to the atmospheric scattering model to synthesize the foggy video image training set {Y i}, wherein, i is the frame number of the clear video image and i=1, 2,..., N, N is the total number of frames of the clear video image and N is not less than 1000;
[0035] In this embodiment, the clear video image data set with known depth includes the NYU image data set.
[0036] Step 102, will (X j ,Y j+t ,X j+t ) to form a set of sample images to construct a training set of foggy video images {(X j ,Y j+t ,X j+t )}, where, the training set of foggy video images {(X j ,Y j+t ,X j+t )}, the number of sample image groups is not less than 10...
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 discloses a video image defogging method based on time series label transfer, comprising the steps of: 1. Establishing a training set of foggy video images; 2. Designing a defogging network to defog the foggy video images in a single group of sample images; 3. 1. Calculate the pixel-level absolute value loss function of the defogged image; 4. Update the weight parameter set; 5. Call the next set of sample images, and repeat steps 2 to 4 until the training is stopped; Defog the foggy image; 7. Defog the actual non-first frame of the test video; 8. Repeat step 7 multiple times to achieve continuous dehazing of the entire video image sequence. The present invention uses temporally adjacent defogging results as a priori to guide subsequent video frames for defogging, and continuously updates the guiding image, thereby realizing continuous and natural changes in the time dimension of the defogged video, while maintaining the spatial dimension of the image It can accurately remove fog and output clear video sequences with consistent time and space.
Description
technical field [0001] The invention belongs to the technical field of video image defogging, and in particular relates to a video image defogging method based on time series tag transfer. Background technique [0002] Video images collected in severe weather such as fog and haze will suffer from quality degradation due to atmospheric scattering, which will make the image color off-white, reduce the contrast, and make it difficult to identify object features, which not only deteriorates the visual effect, but also reduces the appreciation of the image. , It will also lead to deviations in the understanding of image content. For example, in the field of automatic driving, pedestrians or lane lines in the distance may not be detected due to the influence of smog, which may cause traffic accidents. Video image defogging refers to the use of specific methods and means to reduce or even eliminate the adverse effects of suspended particles in the air on video images, and to keep i...
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 Patents(China)
IPC IPC(8): G06T5/00G06K9/62G06N3/04G06N3/08
CPCG06T5/003G06N3/04G06N3/08G06T2207/10016G06T2207/20081G06T2207/20084G06F18/214
Inventor 苏延召崔智高李爱华王涛姜柯蔡艳平王念
Owner ROCKET FORCE UNIV OF ENG