Non-uniformity correction method and system based on spatial domain and frequency domain characteristics

Abstract

The invention discloses a non-uniformity correction method and system based on spatial domain and frequency domain characteristics, and relates to the technical field of infrared image processing, and the method comprises the correction steps: inputting a to-be-corrected image into a trained infrared image non-uniformity correction model, and outputting a corrected image; the model training comprises the following steps: constructing a training data set based on a visible light data set; an initial infrared image non-uniformity correction model is constructed, a U-Net structure is adopted to construct a non-uniformity correction trunk module, and a spatial domain feature extraction module and a frequency domain feature extraction module are arranged in the non-uniformity correction trunk module; and performing iterative training on the initial infrared image heterogeneity correction model by using the training data set to obtain an infrared image heterogeneity correction model. Local pixel-level information is captured through the spatial domain feature extraction module, and in combination with the global spectrum features extracted by the frequency domain feature extraction module, the cooperative effect of local noise suppression and global structure optimization is realized, and the precision and robustness of non-uniformity correction are improved.

Description

Technical Field

[0001] The present invention relates to the field of infrared image processing technology, and more particularly to a non-uniformity correction method and system based on spatial domain and frequency domain features. Background Art

[0002] Currently, traditional non-uniformity correction methods based on mask calibration remain the mainstream technology in infrared image processing. This method uses a mask image captured at the start of a thermal infrared camera as a background reference, and subsequently eliminates fixed-pattern noise through pixel-by-pixel subtraction. Its operational workflow is simple and easy to implement. However, this method has significant drawbacks. First, due to the time-varying nature of detector noise, the background image and real-time noise cannot be fully matched, resulting in residual gauze-like artifacts in the corrected image, particularly in low-contrast areas, with severe loss of detail. Second, detector response drift with ambient temperature (temperature drift effect) can cause background feature shifts, forcing the system to frequently interrupt operation to re-acquire mask images, severely limiting its application in continuous operation scenarios such as security monitoring and industrial inspection. Furthermore, the interplay of pixel response non-uniformity due to manufacturing process variations in infrared focal plane arrays, radiometric distortion caused by the uneven spatial distribution of atmospheric refractive index, and center-to-edge response differences caused by vignetting in the optical system complicate the development of an accurate global correction model using traditional methods. Therefore, addressing these issues is a pressing need for those skilled in the art. Summary of the Invention

[0003] In view of this, the present invention provides a non-uniformity correction method and system based on spatial and frequency domain features, which overcome the above-mentioned defects.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] A non-uniformity correction method based on spatial and frequency domain features, specifically comprising the following steps:

[0006] The calibration steps are:

[0007] Acquire the image to be corrected;

[0008] Inputting the image to be corrected into a trained infrared image non-uniformity correction model to output a corrected image;

[0009] The model training steps are:

[0010] Constructing a training data set based on the visible light data set, wherein the training data set includes a training data pair of an image to be non-uniformity corrected and a pseudo infrared image;

[0011] Constructing an initial infrared image non-uniformity correction model, wherein the initial infrared image non-uniformity correction model adopts a U-Net structure to construct a non-uniformity correction backbone module, wherein the non-uniformity correction backbone module is equipped with a spatial domain feature extraction module and a frequency domain feature extraction module;

[0012] The initial infrared image non-uniformity correction model is iteratively trained using the training data set until convergence, thereby obtaining the infrared image non-uniformity correction model.

[0013] Furthermore, the steps of constructing the training data set are:

[0014] Converting a visible light image in the visible light dataset into a grayscale image;

[0015] Adjusting pixel value distribution of the grayscale image to simulate thermal infrared features to form a pseudo infrared image;

[0016] superimposing an infrared background image on the pseudo infrared image to generate an image to be corrected for non-uniformity, and constructing a training data pair;

[0017] The training data set is constructed based on the training data pairs.

[0018] Furthermore, the non-uniformity correction backbone module includes: an encoder, a bottleneck layer and a decoder connected in sequence; the layers of the encoder are connected through a downsampling module; the layers of the decoder are connected through an upsampling module; the encoder and the decoder are connected through a jump connection; the encoder, the bottleneck layer and the decoder are all provided with the spatial domain feature extraction module and the frequency domain feature extraction module.

[0019] Furthermore, the data processing steps of the spatial feature extraction module are as follows:

[0020] The input image is processed in sequence through the layer normalization function, the first convolution, the depth convolution, the activation function, the multi-weight channel attention module and the second convolution to generate preliminary spatial features;

[0021] The preliminary spatial domain features are superimposed on the input image to obtain the final spatial domain features.

[0022] Furthermore, the data processing steps in the multi-weight channel attention module are:

[0023] Perform multi-scale convolution on the input features to obtain multi-scale features;

[0024] After globally pooling the features of each scale, multi-scale channel features are generated through convolution and Sigmoid function;

[0025] After splicing the multi-scale channel features, perform probabilistic processing to generate probabilistic features;

[0026] Performing convolution processing on the probability features to obtain final channel features;

[0027] The final channel feature is multiplied by the input feature to obtain a multi-weight channel feature.

[0028] Furthermore, the frequency domain feature extraction module extracts frequency domain features by combining real fast Fourier transform and real inverse fast Fourier transform.

[0029] Furthermore, the data processing steps of the frequency domain feature extraction module are:

[0030] The input image is sequentially processed by a third convolution, a first batch of normalization functions, and a first ReLu activation function to obtain a preprocessed feature tensor;

[0031] The preprocessed feature tensor is sequentially processed by real fast Fourier transform, fourth convolution, second batch normalization function, second ReLu activation function and real inverse fast Fourier transform to generate an initial frequency domain feature tensor;

[0032] After the initial frequency domain feature tensor and the preprocessed feature tensor are superimposed, a final frequency domain feature tensor is obtained by convolution processing.

[0033] A non-uniformity correction system based on spatial and frequency domain features, comprising:

[0034] A training set construction module, configured to construct a training data set based on a visible light data set, wherein the training data set includes a training data pair of an image to be non-uniformity corrected and a pseudo infrared image;

[0035] A model construction module is used to construct an initial infrared image non-uniformity correction model, wherein the initial infrared image non-uniformity correction model adopts a U-Net structure to construct a non-uniformity correction backbone module, wherein the non-uniformity correction backbone module is equipped with a spatial domain feature extraction module and a frequency domain feature extraction module;

[0036] a model training module, configured to iteratively train the initial infrared image non-uniformity correction model using the training data set until convergence, thereby obtaining the infrared image non-uniformity correction model;

[0037] The image correction module is used to obtain an image to be corrected, input the image to be corrected into a trained infrared image non-uniformity correction model, and output a corrected image.

[0038] From the above technical solutions, it can be seen that the present invention provides a method and system for non-uniformity correction based on spatial and frequency domain features, which has the following beneficial effects compared with the prior art:

[0039] 1. The spatial domain feature extraction module captures local pixel-level information, and combined with the global spectrum features extracted by the frequency domain feature extraction module, it achieves the synergistic effect of local noise suppression and global structure optimization, significantly improving the accuracy and robustness of non-uniformity correction.

[0040] 2. For stubborn vertical stripe noise in infrared images, frequency domain feature analysis can accurately locate periodic noise components. Combined with spatial domain adaptive filtering, it can completely remove the noise and avoid the residual problems common in traditional methods.

[0041] 3. During the denoising process, through the refined extraction of local spatial features and a multi-scale fusion strategy, the texture, edges, and weak target information in the image are effectively protected, detail loss caused by over-smoothing is avoided, and image usability is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0043] Figure 1 A schematic flow chart of the method provided by the present invention;

[0044] Figure 2 This is a schematic diagram of the structure of the infrared image non-uniformity correction model provided by the present invention;

[0045] Figure 3 A schematic diagram of the structure of the spatial feature extraction module provided by the present invention;

[0046] Figure 4 A schematic diagram of the structure of the multi-weight channel attention module provided by the present invention;

[0047] Figure 5 This is a schematic diagram of the structure of the frequency domain feature extraction module provided by the present invention;

[0048] Figure 6(a) shows the single-point correction result of the first image; Figure 6(b) shows the HINet network correction result of the first image; Figure 6(c) shows the Restormer network correction result of the first image; Figure 6(d) shows the MPRNet network correction result of the first image; Figure 6(e) shows the NAFNet network correction result of the first image; Figure 6(f) shows the correction result of the first image using the infrared image non-uniformity correction model provided by the present invention;

[0049] Figure 7(a) is the single-point correction result of the second image; Figure 7(b) is the HINet network correction result of the second image; Figure 7(c) is the Restormer network correction result of the second image; Figure 7(d) is the MPRNet network correction result of the second image; Figure 7(e) is the NAFNet network correction result of the second image; Figure 7(f) is the correction result of the second image using the infrared image non-uniformity correction model provided by the present invention. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0051] The embodiment of the present invention discloses a non-uniformity correction method based on spatial and frequency domain features, such as Figure 1 As shown, the specific steps include:

[0052] The calibration steps are:

[0053] Step 1: Obtain the image to be corrected;

[0054] Step 2: Input the image to be corrected into the trained infrared image non-uniformity correction model and output the corrected image;

[0055] The model training steps are:

[0056] Step 3: construct a training data set based on the visible light data set, where the training data set includes training data pairs of the image to be non-uniformity corrected and the pseudo infrared image;

[0057] Step 4: construct an initial infrared image non-uniformity correction model. The initial infrared image non-uniformity correction model adopts a U-Net structure to construct a non-uniformity correction backbone module. The non-uniformity correction backbone module is equipped with a spatial domain feature extraction module and a frequency domain feature extraction module.

[0058] Step 5: Iteratively train the initial infrared image non-uniformity correction model using the training data set until convergence, thereby obtaining the infrared image non-uniformity correction model.

[0059] In one embodiment, the steps for constructing the training dataset are:

[0060] Convert the visible light images in the visible light dataset into grayscale images;

[0061] Adjust the pixel value distribution of the grayscale image to simulate thermal infrared characteristics and form a pseudo infrared image;

[0062] The pseudo infrared image is superimposed with the infrared background image to generate an image to be corrected for non-uniformity, and a training data pair is constructed;

[0063] Construct a training dataset based on the training data pairs.

[0064] Furthermore, this embodiment simulates and generates a large number of pseudo infrared images based on the existing visible light data set, and then superimposes the infrared background image to obtain training data pairs of the image to be non-uniformity corrected and the clean pseudo infrared image. The specific method is as follows:

[0065] First, the images in the visible light dataset are converted into grayscale images. Then, the pixel values ​​of the grayscale images are further adjusted to make them closer to the real thermal infrared images, thereby obtaining pseudo infrared images. Next, the image to be corrected for non-uniformity is obtained by superimposing the infrared background image on the clean pseudo infrared image, thereby obtaining training data pairs of the image to be corrected for non-uniformity and the clean pseudo infrared image.

[0066] Furthermore, the network structure diagram of the infrared image non-uniformity correction model is as follows: Figure 2 As shown in FIG, it mainly includes a spatial domain feature extraction module, a frequency domain feature extraction module and a non-uniformity correction backbone module.

[0067] The spatial domain feature extraction module is responsible for extracting image information in the pixel domain and completing the non-uniformity correction task based on the local information of the input image; the frequency domain feature extraction module focuses on extracting the frequency domain information of the input image and achieving non-uniformity correction based on the global information of the input image.

[0068] In one embodiment, the data processing steps of the spatial feature extraction module are as follows:

[0069] The input image is processed in sequence through the layer normalization function, the first convolution, the depth convolution, the activation function, the multi-weight channel attention module and the second convolution to generate preliminary spatial features;

[0070] The preliminary spatial features are superimposed on the input image to obtain the final spatial features.

[0071] Furthermore, the spatial feature extraction module structure is as follows Figure 3As shown in the figure, specifically: the input feature image tensor (i.e., input image) is processed in sequence by layer normalization function, convolution with a convolution kernel size of 1 (i.e., the first convolution), depth convolution with a convolution kernel size of 3, activation function, multi-weight channel attention module and convolution with a convolution kernel size of 1 (i.e., the second convolution), and the obtained feature image tensor (i.e., preliminary spatial features) and the input feature image tensor are added as the output of the spatial feature extraction module (i.e., the final spatial features); among them, the first convolution with a convolution kernel size of 1 doubles the channel dimension of the feature image tensor, and the second convolution with a convolution kernel size of 1 halves the channel dimension of the feature image tensor. The remaining modules do not change the dimension of the feature image tensor.

[0072] In one embodiment, the data processing steps in the multi-weight channel attention module are:

[0073] Perform multi-scale convolution on the input features to obtain multi-scale features;

[0074] After globally pooling the features of each scale, multi-scale channel features are generated through convolution and Sigmoid function processing;

[0075] After splicing the multi-scale channel features, perform probabilistic processing to generate probabilistic features;

[0076] Perform convolution processing on the probability features to obtain the final channel features;

[0077] Multiply the final channel feature with the input feature to obtain the multi-weight channel feature.

[0078] Furthermore, the structure diagram of the multi-weight channel attention module in the spatial feature extraction module is as follows Figure 4 As shown in Figure 1, it first convolves the input feature image tensor (i.e., input feature) with convolution kernel sizes of 3, 5, 7, and 9 to obtain feature image tensors of four scales. 、 、 、 , whose dimensions are consistent with the input image matrix, both , and then processed by pooling, convolution with a convolution kernel size of 1, activation function (ReLU), convolution with a convolution kernel size of 1, and Sigmoid function to obtain multi-scale channel features 、 、 、 , whose dimensions are . Use the Concat function to merge the multi-scale channel features into a matrix , whose dimensions are ; Then use the Reshape function to adjust the dimension to ; Then, apply the Softmax function to the second channel, which converts the multi-scale channel features into a probability distribution, and the sum of all probabilities is 1; then use the Reshape function to change the output dimension of the Softmax function back to , and the final channel feature is obtained by convolution with a convolution kernel size of 1, whose dimension is The input feature image tensor and the final channel feature are multiplied to obtain the output of the multi-weight channel attention module (i.e., multi-weight channel feature).

[0079] In one embodiment, the frequency domain feature extraction module extracts frequency domain features by combining real fast Fourier transform and real inverse fast Fourier transform.

[0080] In one embodiment, the data processing steps of the frequency domain feature extraction module are as follows:

[0081] The input image is processed in sequence by the third convolution, the first batch of normalization functions, and the first ReLu activation function to obtain the preprocessed feature tensor;

[0082] The preprocessed feature tensor is processed in sequence by real fast Fourier transform, fourth convolution, second batch normalization function, second ReLu activation function and real inverse fast Fourier transform to generate the initial frequency domain feature tensor;

[0083] After the initial frequency domain feature tensor and the preprocessed feature tensor are superimposed, the final frequency domain feature tensor is obtained through convolution processing.

[0084] Furthermore, the frequency domain feature extraction module structure is as follows Figure 5 As shown in the figure, specifically, the input feature image tensor (i.e., input image) is sequentially processed through convolution with a convolution kernel size of 1, batch normalization, and ReLu activation to obtain a preprocessed feature tensor. The preprocessed feature tensor is then sequentially processed through a real fast Fourier transform, convolution with a convolution kernel size of 1, batch normalization, ReLu activation, and inverse real fast Fourier transform to obtain a frequency-domain processed feature image tensor (i.e., the initial frequency-domain feature tensor). The preprocessed feature tensor and the frequency-domain processed feature image tensor are added together, and then convolution with a convolution kernel size of 1 is performed to obtain the output feature image tensor (i.e., the final frequency-domain feature tensor). After the real fast Fourier transform, the imaginary and real parts are merged in the channel dimension, doubling the dimension of the feature image tensor. Before the inverse real fast Fourier transform, the imaginary and real parts are split in the channel dimension, halving the dimension of the feature image tensor. The remaining modules do not change the dimension of the feature image tensor.

[0085] In one embodiment, the non-uniformity correction backbone module adopts a U-Net architecture with a five-layer structure. Each encoder layer is connected by a downsampling module, and each decoder layer is connected by an upsampling module. Skip connections are added between the encoder and decoder. Spatial and frequency domain feature extraction modules are used in the encoder, decoder, and bottleneck layers. The channel dimensions of each layer are 32, 64, 128, 256, and 512, respectively. The number of modules in the encoder is 2, 2, 4, and 8; the number of modules in the decoder is 2, 2, 2, and 2; and the number of modules in the bottleneck layer is 12.

[0086] Furthermore, the data processing steps are:

[0087] First, the input feature image tensor is convolved with a convolution kernel size of 1 to change the dimension of the feature image tensor to 32.

[0088] The feature image tensor is divided into two parts in the channel, and the two parts are processed by the spatial domain feature extraction module and the frequency domain feature extraction module respectively. Then the two parts are merged in the channel dimension and passed through the downsampling module. The downsampling module uses a convolution with a kernel size of 2 and a stride of 2. The size of the feature image tensor can be reduced from become ,in, Indicates the number of channels, Indicates the height of the image. Indicates the width of the image. Using this module can achieve the effect of halving the image size and doubling the number of channels.

[0089] Repeat the above operation three times to obtain the output of the four-layer encoder.

[0090] The bottleneck layer divides the feature image tensor into two parts in the channel, passes these two parts through the spatial domain feature extraction module and the frequency domain feature extraction module respectively, and then merges the two parts in the channel dimension.

[0091] The decoder first passes through the upsampling module, which includes a convolution with a convolution kernel size of 1 and a PixelShuffle module. The image tensor size is reduced from become , and then through PixelShuffle into The output of the corresponding layer encoder is then superimposed, and the feature image tensor is divided into two parts in the channel. These two parts are processed by the spatial domain feature extraction module and the frequency domain feature extraction module respectively, and then the two parts are merged in the channel dimension.

[0092] Repeat the operation three times to obtain the output of the four-layer decoder.

[0093] Finally, the feature image tensor is convolved with a convolution kernel size of 1 to change the dimension of the feature image tensor to 1 and obtain the final output image.

[0094] The usage process of the encoder, decoder in the non-uniformity correction backbone module and the spatial domain feature extraction module and the frequency domain feature extraction module in the bottleneck layer is as follows: the feature image tensor is divided into two parts on the channel, the two parts are passed through the spatial domain feature extraction module and the frequency domain feature extraction module respectively, and finally the two parts are merged in the channel dimension to obtain the final output tensor.

[0095] The peak signal-to-noise ratio loss function is used as the loss function in the infrared image non-uniformity correction model. The loss function is defined as: ,in, Indicates the number of image bits, represents a clean infrared image, Infrared image representing the output of the model.

[0096] Figures 6(a)-6(f) and 7(a)-7(f) show the comparison results of various methods on different scenes. The comparison shows that the method of this embodiment not only completely removes non-uniform noise, but also well preserves the details and edges in the image, effectively improving image quality.

[0097] On the other hand, this embodiment discloses a non-uniformity correction system based on spatial and frequency domain features, including:

[0098] A training set construction module is used to construct a training data set based on a visible light data set, where the training data set includes training data pairs of an image to be non-uniformity corrected and a pseudo infrared image;

[0099] A model construction module is used to construct an initial infrared image non-uniformity correction model. The initial infrared image non-uniformity correction model adopts a U-Net structure to construct a non-uniformity correction backbone module. The non-uniformity correction backbone module is equipped with a spatial domain feature extraction module and a frequency domain feature extraction module;

[0100] The model training module is used to iteratively train the initial infrared image non-uniformity correction model using the training data set until convergence to obtain the infrared image non-uniformity correction model.

[0101] The image correction module is used to obtain the image to be corrected, input the image to be corrected into the trained infrared image non-uniformity correction model, and output the corrected image.

[0102] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0103] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A non-uniformity correction method based on spatial and frequency domain features, characterized in that: The specific steps include: The calibration steps are: Acquire the image to be corrected; Inputting the image to be corrected into a trained infrared image non-uniformity correction model to output a corrected image; The model training steps are: Constructing a training data set based on the visible light data set, wherein the training data set includes a training data pair of an image to be non-uniformity corrected and a pseudo infrared image; Constructing an initial infrared image non-uniformity correction model, wherein the initial infrared image non-uniformity correction model adopts a U-Net structure to construct a non-uniformity correction backbone module, wherein the non-uniformity correction backbone module is equipped with a spatial domain feature extraction module and a frequency domain feature extraction module; The initial infrared image non-uniformity correction model is iteratively trained using the training data set until convergence, thereby obtaining the infrared image non-uniformity correction model.

2. The non-uniformity correction method based on spatial and frequency domain features according to claim 1, characterized in that: The steps for constructing the training dataset are: Converting a visible light image in the visible light dataset into a grayscale image; Adjusting pixel value distribution of the grayscale image to simulate thermal infrared features to form a pseudo infrared image; superimposing an infrared background image on the pseudo infrared image to generate an image to be corrected for non-uniformity, and constructing a training data pair; The training data set is constructed based on the training data pairs.

3. The non-uniformity correction method based on spatial and frequency domain features according to claim 1, characterized in that: The non-uniformity correction backbone module includes: an encoder, a bottleneck layer and a decoder connected in sequence; the layers of the encoder are connected through a downsampling module; the layers of the decoder are connected through an upsampling module; the encoder and the decoder are connected through a jump connection; the encoder, the bottleneck layer and the decoder are all provided with the spatial domain feature extraction module and the frequency domain feature extraction module.

4. The non-uniformity correction method based on spatial and frequency domain features according to claim 1, characterized in that: The data processing steps of the spatial feature extraction module are as follows: The input image is processed in sequence through the layer normalization function, the first convolution, the depth convolution, the activation function, the multi-weight channel attention module and the second convolution to generate preliminary spatial features; The preliminary spatial domain features are superimposed on the input image to obtain the final spatial domain features.

5. The non-uniformity correction method based on spatial and frequency domain features according to claim 4, characterized in that: The data processing steps in the multi-weight channel attention module are: Perform multi-scale convolution on the input features to obtain multi-scale features; After globally pooling the features of each scale, multi-scale channel features are generated through convolution and Sigmoid function; After splicing the multi-scale channel features, perform probabilistic processing to generate probabilistic features; Performing convolution processing on the probability features to obtain final channel features; The final channel feature is multiplied by the input feature to obtain a multi-weight channel feature.

6. The non-uniformity correction method based on spatial and frequency domain features according to claim 4, characterized in that: The frequency domain feature extraction module extracts frequency domain features by combining real fast Fourier transform and real inverse fast Fourier transform.

7. The non-uniformity correction method based on spatial and frequency domain features according to claim 6, characterized in that: The data processing steps of the frequency domain feature extraction module are as follows: The input image is processed sequentially by a third convolution, a first batch of normalization functions, and a first ReLu activation function to obtain a preprocessed feature tensor; The preprocessed feature tensor is sequentially processed by real fast Fourier transform, fourth convolution, second batch normalization function, second ReLu activation function and real inverse fast Fourier transform to generate an initial frequency domain feature tensor; After the initial frequency domain feature tensor and the preprocessed feature tensor are superimposed, a final frequency domain feature tensor is obtained by convolution processing.

8. A non-uniformity correction system based on spatial and frequency domain features, characterized in that: include: A training set construction module, configured to construct a training data set based on a visible light data set, wherein the training data set includes a training data pair of an image to be non-uniformity corrected and a pseudo infrared image; A model construction module is used to construct an initial infrared image non-uniformity correction model, wherein the initial infrared image non-uniformity correction model adopts a U-Net structure to construct a non-uniformity correction backbone module, wherein the non-uniformity correction backbone module is equipped with a spatial domain feature extraction module and a frequency domain feature extraction module; a model training module, configured to iteratively train the initial infrared image non-uniformity correction model using the training data set until convergence, thereby obtaining the infrared image non-uniformity correction model; The image correction module is used to obtain an image to be corrected, input the image to be corrected into a trained infrared image non-uniformity correction model, and output a corrected image.

Images