CN-121985081-A - Method and system for establishing synthetic video noise reduction sample set and image processor

CN121985081ACN 121985081 ACN121985081 ACN 121985081ACN-121985081-A

Abstract

The invention provides a method, a system and an image processor for establishing a synthetic video noise reduction sample set, wherein the method comprises the steps of adding noise and dynamically clipping based on a first data set to obtain the first sample set; the method comprises the steps of obtaining a first data set, obtaining a second data set, obtaining a video noise reduction sample set, wherein the first data set is a dynamic noise reduction data set, carrying out noise addition and static clipping processing based on the second data set, obtaining the second sample set, wherein the second data set is a static noise reduction data set, and establishing the video noise reduction sample set based on the first sample set and the second sample set. The method and the system for establishing the synthesized video noise reduction sample set and the image processor provided by the embodiment of the invention improve the efficiency of establishing the video noise reduction sample set.

Inventors

LI ZHENG
FENG JINLI
SHEN CHAO

Assignees

北京清微智能科技股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260109

Claims (15)

1. A method for creating a composite video noise reduction sample set, comprising: Noise adding and dynamic clipping processing are carried out on the basis of a first data set to obtain the first data set, wherein the first data set is a dynamic noise reduction data set; performing noise addition and static clipping processing based on a second data set to obtain a second sample set, wherein the second data set is a static noise reduction data set; and establishing a video noise reduction sample set based on the first sample set and the second sample set.
2. The method of claim 1, wherein the noise adding based on the first data set comprises: Performing compensation black frame quantization truncation processing on the first data set to obtain a first sampling data set; black frame sampling is carried out on the first sampling data set, and noise which corresponds to each frame of image in the first data set and is irrelevant to illumination is obtained; And obtaining a first noise data set according to the noise related to illumination, the noise related to illumination corresponding to each frame of image in the first data set and each frame of image in the first data set, wherein the noise related to illumination is obtained in advance.
3. The method of claim 2, wherein the deriving a first noise dataset from the illumination-related noise, the illumination-independent noise corresponding to each frame of image in the first dataset, and each frame of image in the first dataset comprises: And adding noise to each frame of image in the first data set according to a formula D 1 = KI 1 + N 1 , wherein D 1 represents one frame of image in the first noise data set, I 1 represents one frame of image in the first data set, N 1 represents illumination-independent noise corresponding to one frame of image in the first data set, and K represents illumination-related noise and is obtained by carrying out noise calibration based on one frame of image of 24-color card video stream data.
4. The method of claim 2, wherein performing a compensated black frame quantization truncation process on the first data set to obtain a first sampled data set comprises: screening the first data set based on a first sensitivity threshold to obtain a first temporary data set; drawing a brightness histogram of each frame of image in the first temporary data set to obtain a brightness histogram corresponding to each frame of image in the first temporary data set; Performing truncation processing on a brightness histogram corresponding to each frame of image in the first temporary data set to obtain a truncated brightness histogram corresponding to each frame of image in the first temporary data set; And compensating the truncated brightness histogram corresponding to each frame of image in the first temporary data set, so that the compensated brightness histogram corresponding to each frame of image in the first temporary data set accords with Gaussian distribution.
5. The method of claim 2, wherein dynamically clipping based on the first noise data set comprises: and performing motion region clipping on the continuous preset number of frame images of the first noise data set.
6. The method of claim 1, wherein noise adding based on the second data set comprises: Performing compensation black frame quantization truncation processing on the second data set to obtain a second sampling data set; black frame sampling is carried out on the second sampling data set, and noise which corresponds to each frame of image in the second data set and is irrelevant to illumination is obtained; And obtaining a second noise data set according to the noise related to illumination, the noise related to illumination corresponding to each frame of image in the second data set and each frame of image in the second data set, wherein the noise related to illumination is obtained in advance.
7. The method of claim 6, wherein the deriving a second noise dataset from the illumination-related noise, the illumination-independent noise corresponding to each frame of image in the second dataset, and each frame of image in the second dataset comprises: And adding noise to each frame of image in the second data set according to a formula D 2 = KI 2 + N 2 , wherein D 2 represents one frame of image in the second noise data set, I 2 represents one frame of image in the second data set, N 2 represents illumination-independent noise corresponding to one frame of image in the second data set, and K represents illumination-related noise and is obtained by carrying out noise calibration based on one frame of image of 24-color card video stream data.
8. The method of claim 6, wherein performing a compensated black frame quantization truncation process on the second data set to obtain a second sampled data set comprises: screening the second data set based on a second sensitivity threshold to obtain a second temporary data set; Drawing a brightness histogram of each frame of image in the second temporary data set to obtain a brightness histogram corresponding to each frame of image in the second temporary data set; Performing truncation processing on the brightness histogram corresponding to each frame of image in the second temporary data set to obtain a truncated brightness histogram corresponding to each frame of image in the second temporary data set; And compensating the truncated brightness histogram corresponding to each frame of image in the second temporary data set, so that the compensated brightness histogram corresponding to each frame of image in the second temporary data set accords with Gaussian distribution.
9. The method of claim 6, wherein performing static clipping based on the second noise data set comprises: And cutting each frame of image in the second noise data set according to a fixed size sequentially from a preset position, randomly selecting a starting position to cut according to the fixed size, or randomly moving a preset distance by the random selection starting position to cut according to the fixed size, so as to obtain a multi-frame cutting image corresponding to each frame of image.
10. A training system of a noise reduction network model, characterized in that a video noise reduction sample set obtained by adopting the method for establishing a composite video noise reduction sample set according to any one of claims 1 to 9 comprises: the acquisition module is used for acquiring the video noise reduction sample set as training data; and the training module is used for training the original model based on the training data to obtain a noise reduction network model.
11. An image processor, characterized in that the noise reduction network model obtained by the training system for noise reduction network model according to claim 10 comprises: The receiving module is used for receiving the video stream data to be noise reduced; And the noise reduction module is used for carrying out noise reduction processing on the video stream data to be noise reduced through the noise reduction network model.
12. A board comprising at least one image processor according to claim 11.
13. An electronic device comprising at least one board as claimed in claim 12.
14. A computer device comprising a memory, a processor and a computer program stored on the memory, characterized in that the processor executes the computer program to implement the method of any one of claims 1 to 9.
15. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the method of any one of claims 1 to 9.

Description

Method and system for establishing synthetic video noise reduction sample set and image processor Technical Field The invention relates to the technical field of image processing, in particular to a method and a system for establishing a synthetic video noise reduction sample set and an image processor. Background At present, a noise reduction model can be adopted to carry out noise reduction treatment on video data, and the noise reduction model can be obtained through deep learning training. In the prior art, a large amount of training sample data is required for training the noise reduction model. In order to obtain a high-quality training sample, a real noisy-clean Raw video frame data pair can be acquired as training data based on a screen shot image, but a real noisy-clean Raw video denoising data set is acquired, high-quality display and shooting equipment is highly dependent, a large amount of video data needs to be shot, and a large amount of time is consumed. Therefore, how to improve the construction efficiency of training sample data is an important issue to be solved in the art. Disclosure of Invention Aiming at the problems in the prior art, the embodiment of the invention provides a method and a system for establishing a synthetic video noise reduction sample set and an image processor, which can at least partially solve the problems in the prior art. In a first aspect, the present invention provides a method for establishing a noise reduction sample set of a synthesized video, including: Noise adding and dynamic clipping processing are carried out on the basis of a first data set to obtain the first data set, wherein the first data set is a dynamic noise reduction data set; performing noise addition and static clipping processing based on a second data set to obtain a second sample set, wherein the second data set is a static noise reduction data set; and establishing a video noise reduction sample set based on the first sample set and the second sample set. Further, the noise adding based on the first data set includes: Performing compensation black frame quantization truncation processing on the first data set to obtain a first sampling data set; black frame sampling is carried out on the first sampling data set, and noise which corresponds to each frame of image in the first data set and is irrelevant to illumination is obtained; And obtaining a first noise data set according to the noise related to illumination, the noise related to illumination corresponding to each frame of image in the first data set and each frame of image in the first data set, wherein the noise related to illumination is obtained in advance. Further, the obtaining a first noise data set according to the noise related to illumination, the noise related to illumination corresponding to each frame of image in the first data set, and each frame of image in the first data set includes: And adding noise to each frame of image in the first data set according to a formula D 1 = KI1 + N1, wherein D 1 represents one frame of image in the first noise data set, I 1 represents one frame of image in the first data set, N 1 represents illumination-independent noise corresponding to one frame of image in the first data set, and K represents illumination-related noise and is obtained by carrying out noise calibration based on one frame of image of 24-color card video stream data. Further, the performing the black frame quantization truncation process on the first data set to obtain a first sampled data set includes: screening the first data set based on a first sensitivity threshold to obtain a first temporary data set; drawing a brightness histogram of each frame of image in the first temporary data set to obtain a brightness histogram corresponding to each frame of image in the first temporary data set; Performing truncation processing on a brightness histogram corresponding to each frame of image in the first temporary data set to obtain a truncated brightness histogram corresponding to each frame of image in the first temporary data set; And compensating the truncated brightness histogram corresponding to each frame of image in the first temporary data set, so that the compensated brightness histogram corresponding to each frame of image in the first temporary data set accords with Gaussian distribution. Further, performing dynamic clipping processing based on the first noise data set includes: and performing motion region clipping on the continuous preset number of frame images of the first noise data set. Further, noise adding based on the second data set includes: Performing compensation black frame quantization truncation processing on the second data set to obtain a second sampling data set; black frame sampling is carried out on the second sampling data set, and noise which corresponds to each frame of image in the second data set and is irrelevant to illumination is obtained; And obtaining a second noise data set accordi