Two-color image fusion anti-interference method and device based on adaptive guide filtering

Abstract

The invention discloses a two-color image fusion anti-interference method and device based on adaptive guide filtering, and relates to the technical field of image processing and target recognition. The method comprises the following steps: performing feature extraction on a medium-wave image and a long-wave image to obtain a feature map of the medium-wave image and a feature map of the long-wave image; denoising the feature map to obtain a low-noise feature map; performing same-coordinate pixel comparison on the low-noise feature map to obtain an initial attention weight map; performing guided filtering on the initial attention weight map to obtain high and low frequency component fusion weights of the medium-wave image and the long-wave image; and according to the adaptive fusion parameters of the high and low frequency components, carrying out weighted fusion on the fusion weights of the high and low frequency components to obtain a high and low frequency fusion result. According to the invention, when image fusion is carried out, calculation resources are reduced, time cost is reduced, the correlation degree with a sensor is reduced, and the application range is expanded.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing and target recognition, and in particular to a two-color image fusion anti-interference method and device based on adaptive guided filtering. Background Art

[0002] Infrared imagery is a widely used image type in civil and industrial fields, such as for nighttime airport surveillance and surface patrol operations. Because it can penetrate inclement weather such as fog, haze, rain, and snow, and produce images in complete darkness, it holds significant value for target detection and recognition in complex environments. However, in practical applications, commonly used infrared detectors typically only image specific short-, medium-, and long-wavelength radiation sources. These images are often affected by various noise and interference factors, resulting in weak target imaging characteristics or excessive background interference, leading to reduced target detection accuracy. Due to differences in target, interference, and atmospheric transmittance across different wavelengths, fusing infrared images captured at different wavelengths can increase the information entropy of target features and enhance image anti-interference capabilities. For example, image fusion techniques can be used to fuse information from multiple medium- and long-wavelength infrared images to enhance image quality and target detection accuracy, thereby improving the automation of nighttime surveillance and detection in these scenarios.

[0003] Existing image fusion methods mainly include fusion methods based on wavelet transform and fusion methods based on deep learning.

[0004] For the fusion method based on wavelet transform, the computational complexity of wavelet transform is large, requiring high computing resources and time costs, so it may be limited for application scenarios with high real-time requirements.

[0005] For deep learning-based fusion methods, the trained models are often only suitable for infrared images acquired by the current sensor. To adapt to new sensors, new images must be collected, resulting in weak algorithm transferability. Furthermore, acquiring large-scale medium-wave and long-wave infrared data pairs for training is not easy. Summary of the Invention

[0006] In response to the defects in the existing technology, the technical problem solved by the present invention is: when performing image fusion, how to reduce computing resources and time costs while reducing the correlation with the sensor and expanding the scope of application.

[0007] To achieve the above objectives, in a first aspect, an embodiment of the present application provides a two-color image fusion anti-interference method based on adaptive guided filtering, the method comprising the following steps:

[0008] Acquire medium-wave and long-wave images that need to be fused;

[0009] After feature extraction of the medium wave image and the long wave image, a feature map of the medium wave image and a feature map of the long wave image are obtained; after denoising the feature map, a low-noise feature map is obtained; after comparing pixels with the same coordinates of the low-noise feature map, an initial attention weight map is obtained;

[0010] After performing guided filtering on the initial attention weight map, a fusion weight of low-frequency components of the medium-wave image and the long-wave image, and a fusion weight of high-frequency components of the medium-wave image and the long-wave image are obtained;

[0011] After scale decomposition of the medium-wave image and the long-wave image, the low-frequency component and the high-frequency component of the medium-wave image, as well as the low-frequency component and the high-frequency component of the long-wave image are obtained; according to the adaptive fusion parameters of the low-frequency components, the fusion weights of the low-frequency components are weightedly fused to obtain the low-frequency fusion result; according to the adaptive fusion parameters of the high-frequency components, the fusion weights of the high-frequency components are weightedly fused to obtain the high-frequency fusion result.

[0012] In conjunction with the first aspect, in one embodiment, the process of extracting features from the medium-wave image and the long-wave image includes:

[0013] After Laplace filtering of the medium wave image and the long wave image, the feature map of the medium wave image and the feature map of the long wave image F are obtained. n (x, y), n represents the feature map of the medium wave image or the feature map of the long wave image, (x, y) represents the current pixel coordinates;

[0014] F n (x,y)=I n (x,y)*L(x,y)

[0015]

[0016] L represents the Laplace operator.

[0017] In combination with the first aspect, in one embodiment, the low noise characteristic graph F n The calculation formula for ′(x,y) is:

[0018] Where G is a low-pass Gaussian filter with parameters r and σ.

[0019] In combination with the first aspect, in one embodiment, the initial attention weight map The calculation formula is:

[0020]

[0021] In combination with the first aspect, in one embodiment, the low-frequency component fusion weight The calculation formula is: Among them I n represents a medium-wave image or a long-wave image; r1 is 0.075 times the shortest side of the initial attention weight map, and ò1 is 0.4;

[0022] The high frequency component fusion weight The calculation formula is: r2 is 0.01 times the shortest side of the initial attention weight map, ò2=10 -5 .

[0023] In conjunction with the first aspect, in one embodiment, the process of determining the adaptive fusion parameter of the low-frequency component includes:

[0024] The image size is defined as w×n, and the maximum pixel value of the low-frequency component B1 of the medium-wave image and the low-frequency component B2 of the long-wave image is and The logarithmic mean of the brightness values ​​of the low-frequency components of the medium-wave image and the long-wave image is The calculation formula is:

[0025]

[0026] δ is a small amount that is not zero, and the brightness gain is defined as g n :

[0027]

[0028] Then the adaptive fusion parameter of B1 is Adaptive fusion parameters of B2

[0029] The calculation formula of the low-frequency fusion result is: B = |α1W1 l B1-α2W2 l B2|;

[0030] The process of determining the adaptive fusion parameters of the high-frequency component includes:

[0031] Calculate the logarithmic mean of the brightness values ​​of the high-frequency component D1 of the medium-wave image and the high-frequency component D2 of the long-wave image

[0032]

[0033] Adaptive fusion parameters of D1 Adaptive fusion parameters of D2

[0034] The calculation formula of the high-frequency fusion result is:

[0035] In combination with the first aspect, in one embodiment, the process of performing scale decomposition on the medium-wave image and the long-wave image includes: performing Gaussian filtering on the medium-wave image and the long-wave image respectively to obtain a low-frequency component of the medium-wave image and a low-frequency component of the long-wave image; subtracting the medium-wave image from the low-frequency component of the medium-wave image to obtain a high-frequency component of the medium-wave image; and subtracting the long-wave image from the low-frequency component of the long-wave image to obtain a high-frequency component of the long-wave image.

[0036] In combination with the first aspect, in one embodiment, the method further includes the following steps: performing background removal on the low-frequency fusion result and the high-frequency fusion result to obtain a fused image.

[0037] In combination with the first aspect, in one embodiment, the calculation formula of the fused image A is: A=|γD-B|; wherein γ is an adaptive adjustment factor, if the average value of the pixels with the brightness ranking last 10% to 20% in the high-frequency fusion result is t D , the average brightness of the top 10% to 20% pixels in the low-frequency fusion result is t B ,but

[0038] In the second aspect, an embodiment of the present application provides a two-color image fusion and anti-interference device based on adaptive guided filtering, wherein the two-color image fusion and anti-interference device based on adaptive guided filtering includes a processor, a memory, and a two-color image fusion and anti-interference program based on adaptive guided filtering stored on the memory and executable by the processor, wherein when the two-color image fusion and anti-interference program based on adaptive guided filtering is executed by the processor, the steps of the method provided in the first aspect are implemented.

[0039] Compared with the prior art, the advantages of the present invention are:

[0040] (1) Compared with learning-based algorithms that need to learn specific sensor imaging results, this algorithm has stronger algorithm versatility and is less correlated with sensor characteristics. It can be flexibly applied to medium and long-wave imaging sensors with different imaging modes, different manufacturers, and different optoelectronic devices.

[0041] (2) Based on the adaptive guided filtering, adaptive fusion weight judgment based on feature intensity can be performed in different image scale spaces. This weight is obtained by calculating the brightness gain of the high-frequency component and the low-frequency component of the image, which can achieve the effect of eliminating the background while retaining the target. In addition, the adaptive fusion parameters utilize the statistical characteristics of the image, reducing the algorithm limitations brought by fixed parameters.

[0042] (3) The fusion result achieves a relatively ideal background elimination effect, which helps human visual attention and computer vision processing, and realizes the improvement of detection accuracy.

[0043] (4) The present invention uses box filtering to achieve rapid solution of guided filtering. The processing time on a single-channel image of any size does not exceed 10ms, which can achieve real-time processing in the engineering field.

[0044] (5) The fusion results can be used for target recognition based on a general network for infrared image target recognition, completing the decoupling of recognition and image fusion, so that the fusion algorithm can be flexibly applied to various detection tasks as a pre-processing of other algorithms, thereby increasing the scope of application of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0046] Figure 1 Flowchart of a two-color image fusion anti-interference method based on adaptive guided filtering in an embodiment of the present invention;

[0047] Figure 2 is a fusion image of a medium-wave image and a long-wave image in different scenes in an embodiment of the present invention;

[0048] Figure 3 Detection results of merchant ships with a universal model in different images according to the embodiment of the present invention;

[0049] Figure 4 This is a hardware structure diagram of the device for the adaptive guided filtering dual-color image fusion anti-interference method involved in the embodiment of the present application. DETAILED DESCRIPTION

[0050] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0051] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, combined, or partially merged, so the actual execution order may vary depending on the actual situation.

[0052] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0053] First, the research and development process of the present invention is briefly described.

[0054] The inventors discovered that:

[0055] Traditional wavelet-based image fusion algorithms (such as the Chinese invention patent with publication number CN105844606A) perform wavelet decomposition on the image to be fused to obtain its low-frequency and high-frequency components, then use a custom fusion rule to fuse the frequency bands, and finally perform wavelet inverse transformation on each fused frequency band to obtain the final fused image. However, the decomposition and fusion of different images by wavelet-based image fusion algorithms are affected by the characteristics of the images themselves, so in some cases, over-fusion or under-fusion may occur. Secondly, it is difficult to distinguish between interference and the target to be detected in the decomposed frequency components, resulting in the fused image being unable to suppress interference and achieve the purpose of improving the target detection rate.

[0056] Fusion algorithms based on deep learning (such as the Chinese invention patent with publication number CN113033630A) use an encoding network to encode the image to be fused, then fuse the encoded features in the fusion layer, and finally decode the fused feature map by the decoder to obtain the final fused image. To facilitate target detection, commonly used fusion algorithms based on deep learning (such as the Chinese invention patent with publication number CN113610180A) also use a network based on target detection or target segmentation to train the fusion network together. This will cause the target image of the fusion network to be affected by the detection network, thereby reducing the types of images that can be processed. Secondly, the trained model is often only suitable for infrared images obtained by the current sensor. If you want to adapt to a new sensor, you need to collect new images, resulting in weak algorithm portability. Acquiring large-scale medium-wave and long-wave infrared data pairs for training is also not easy.

[0057] On this basis, in a first aspect, an embodiment of the present application provides a two-color image fusion anti-interference method based on adaptive guided filtering, the method comprising the following steps:

[0058] Step A: Obtain the medium-wave image and the long-wave image to be fused. The medium-wave image and the long-wave image can be full-size images or images of a specified area.

[0059] Step B: After extracting features from the medium wave image and the long wave image respectively, the feature maps of the medium wave image and the long wave image are obtained; after denoising the feature maps of the medium wave image and the long wave image respectively, the low-noise feature maps of the medium wave image and the long wave image are obtained; after comparing the low-noise feature maps of the medium wave image and the long wave image with the same coordinate pixels respectively, the initial attention weight maps of the medium wave image and the long wave image are obtained.

[0060] Step C: After performing guided filtering with adaptive parameters on the initial attention weight maps of the medium wave image and the long wave image, the low-frequency component fusion weights of the medium wave image and the long wave image, as well as the high-frequency component fusion weights of the medium wave image and the long wave image are obtained.

[0061] Step D: After performing scale decomposition on the medium-wave image and the long-wave image, the low-frequency component and high-frequency component of the medium-wave image, as well as the low-frequency component and high-frequency component of the long-wave image, are obtained; according to the adaptive fusion parameters of the low-frequency components, the fusion weights of the low-frequency components of the medium-wave image and the low-frequency components of the long-wave image are weightedly fused to obtain a low-frequency fusion result; according to the adaptive fusion parameters of the high-frequency components, the fusion weights of the high-frequency components of the medium-wave image and the high-frequency components of the long-wave image are weightedly fused to obtain a high-frequency fusion result.

[0062] Step E: After background removal, the low-frequency fusion result and the high-frequency fusion result in step D are subjected to a fusion image.

[0063] In one embodiment, the process of extracting features from the medium-wave image and the long-wave image in step B includes:

[0064] After Laplace filtering of the medium wave image and the long wave image, the feature map of the medium wave image and the feature map of the long wave image F are obtained. n (x, y), n represents the feature map of the medium wave image or the feature map of the long wave image. In this embodiment, n = 1, 2, 1 represents the medium wave image, and 2 represents the long wave image; (x, y) represents the current pixel coordinates;

[0065] F n (x,y)=I n (x,y)*L(x,y)

[0066]

[0067] Where L represents the Laplace operator; the Laplace operator can be used to effectively extract the edge and corner features of the image.

[0068] In one embodiment, the process of determining the low-noise feature maps of the medium-wave image and the long-wave image in step B includes: using a Gaussian filter to perform denoising on the feature map of the medium-wave image and the feature map of the long-wave image, respectively, to obtain the low-noise feature map F of the medium-wave image and the low-noise feature map F of the long-wave image. n ′(x,y), the calculation formula is:

[0069] Where G is a low-pass Gaussian filter with parameters r and σ.

[0070] In one embodiment, the initial attention weight map in step B The calculation formula is:

[0071] That is, the sizes of each pixel with the same coordinates in the low-noise feature maps of medium wave and long wave are compared, and the larger value is assigned 1 and the smaller value is assigned 0.

[0072] In one embodiment, the process of determining the low-frequency component fusion weight in step C includes: taking the initial attention weight map of the medium-wave image and the initial attention weight map of the long-wave image as input images, using the medium-wave image and the long-wave image in S1 as guide images, and performing guided filtering using a guided filter with r1 and ò1 as parameters to obtain the low-frequency component fusion weight The calculation formula is:

[0073] Among them I n represents a medium-wavelength image or a long-wavelength image. As mentioned above, n = 1 represents a medium-wavelength image, and n = 2 represents a long-wavelength image. r1 is set to 0.075 times the short side of the input image, and ò1 is set to 0.4. Guided filtering is a locally optimal linear filter. It preserves the edge information of the guide map in the output image. By guiding the original image, guided filtering maintains the consistency of the image's spatial structure. This filtering process can be quickly implemented using a box filter, achieving a processing speed of 10 ms per frame for a 640×640 image.

[0074] Similarly, the high-frequency component fusion weight in step C The calculation formula is:

[0075] r2 is set to 0.01 times the short side, ò2=10 -5 .

[0076] In one embodiment, the scale decomposition process of the medium-wave image and the long-wave image in step D includes: performing Gaussian filtering on the medium-wave image and the long-wave image to obtain a low-frequency component of the medium-wave image and a low-frequency component of the long-wave image; subtracting the medium-wave image from the low-frequency component of the medium-wave image to obtain a high-frequency component of the medium-wave image; and subtracting the long-wave image from the low-frequency component of the long-wave image to obtain a high-frequency component of the long-wave image. This decomposition method is simple to operate, lossless, and reversible, and has lower computational complexity than wavelet decomposition.

[0077] In one embodiment, the process of determining the adaptive fusion parameters of the low-frequency components of the medium-wavelength image and the long-wavelength image in step D includes:

[0078] Calculate the logarithmic mean of the brightness values ​​of the low-frequency components of the medium-wave image and the long-wave image. If the image size is w×n, the maximum pixel value of the low-frequency component B1 of the medium-wave image and the low-frequency component B2 of the long-wave image is and Then the logarithmic mean of the brightness value of the low-frequency component can be expressed as n=1,2;

[0079]

[0080] Where δ is a small amount that is not equal to 0, then the brightness gain is defined as g n ,n=1,2:

[0081]

[0082] At this time, the adaptive fusion parameters of B1 Adaptive fusion parameters of B2

[0083] On this basis, the calculation formula of the low-frequency fusion result in step D is: B = |α1W1 l B1-α2W2 l B2|.

[0084] Similarly, the process of determining the adaptive fusion parameters of the high-frequency components of the medium-wave image and the long-wave image in step D includes:

[0085] Calculate the logarithmic mean of the brightness values ​​of the high-frequency component D1 of the medium-wave image and the high-frequency component D2 of the long-wave image n=1,2;

[0086]

[0087] At this time, the adaptive fusion parameters of D1 Adaptive fusion parameters of D2

[0088] On this basis, the calculation formula of the high-frequency fusion result in step D is:

[0089] In one embodiment, in step E, the calculation formula of the fused image A is: A = |γD-B|; where γ is an adaptive adjustment factor, which is determined by the brightness ratio of some pixels in the high-frequency and low-frequency fused images. If the average brightness of the pixels in the last 10% to 20% of the high-frequency fusion result is t D , the average brightness of the top 10% to 20% pixels in the low-frequency fusion result image is t B ,but

[0090] The above method is described below by a specific embodiment. Figure 1 As shown, the steps of the method specifically include:

[0091] S1. Image input: Obtain the medium-wave image and long-wave image to be fused, that is, input the long-wave infrared image and medium-wave infrared image to be fused respectively. In one embodiment, the medium-wave image and long-wave image can be full-size images or images of a specified area (that is, the area of ​​interest is cropped after the image is input).

[0092] S2. Multiscale Image Decomposition: Gaussian filtering is performed on both the medium-wave and long-wave images to obtain the low-frequency components of the medium-wave image and the long-wave image. The medium-wave image is then subtracted from the low-frequency component to obtain the high-frequency component of the medium-wave image. The long-wave image is then subtracted from the low-frequency component to obtain the high-frequency component of the long-wave image. This decomposition method is simple, lossless, and reversible, and has lower computational complexity than wavelet decomposition.

[0093] S3. Feature extraction: After performing Laplace filtering on the medium wave image and the long wave image, the feature map of the medium wave image and the feature map of the long wave image F are obtained. n (x, y), n represents the feature map of the medium wave image or the feature map of the long wave image. In this embodiment, n = 1, 2, 1 represents the medium wave image, and 2 represents the long wave image; (x, y) represents the current pixel coordinates;

[0094] F n (x,y)=I n (x,y)*L(x,y)

[0095]

[0096] Where L represents the Laplace operator; the Laplace operator can be used to effectively extract the edge and corner features of the image.

[0097] S4. Feature map denoising: After denoising the feature map of the medium wave image and the feature map of the long wave image respectively using Gaussian filter, the low noise feature map of the medium wave image and the low noise feature map of the long wave image are obtained. n ′(x,y), the calculation formula is:

[0098] Where G is a low-pass Gaussian filter with parameters r and σ.

[0099] S5: After comparing the low-noise feature map of the medium-wave image and the low-noise feature map of the long-wave image with the same coordinate pixels, the initial attention weight map of the medium-wave image and the initial attention weight map of the long-wave image are obtained. The calculation formula is:

[0100] That is, the sizes of each pixel with the same coordinates in the low-noise feature maps of medium wave and long wave are compared, and the larger value is assigned 1 and the smaller value is assigned 0.

[0101] S6: Take the initial attention weight map of the medium wave image and the initial attention weight map of the long wave image as the input map, use the medium wave image and the long wave image in S1 as the guide map, and use the guide filter with r1 and ò1 as parameters to perform guided filtering to obtain the low frequency component fusion weight The calculation formula is:

[0102] Among them I n represents a medium-wavelength image or a long-wavelength image. As mentioned above, when n is 1, it represents a medium-wavelength image, and when n is 2, it represents a long-wavelength image. r1 is set to 0.075 times the shortest side of the input image (i.e., the shortest side of the initial attention weight map), and ò1 is set to 0.4. Guided filtering is a locally optimal linear filter. It preserves the edge information of the guide map in the output image. Guided filtering, guided by the original image, maintains the consistency of the image's spatial structure. This filtering process can be quickly implemented using box filtering, achieving a processing speed of 10 ms per frame for a 640×640 image.

[0103] S7: After performing guided filtering with adaptive parameters on the initial attention weight map of the medium wave image and the initial attention weight map of the long wave image, the high frequency component fusion weight of the medium wave image and the high frequency component fusion weight of the long wave image are obtained.

[0104] Similar to S6, the high-frequency component fusion weights in S7 The calculation formula is:

[0105] r2 is set to 0.01 times the shortest side, ò2=10 -5 .

[0106] S8: Determine the adaptive fusion parameters of the low-frequency components of the medium-wave image and the long-wave image in S2:

[0107] Calculate the logarithmic mean of the brightness values ​​of the low-frequency components of the medium-wave image and the long-wave image. If the image size is w×n, the maximum pixel value of the low-frequency component B1 of the medium-wave image and the low-frequency component B2 of the long-wave image is and Then the logarithmic mean of the brightness value of the low-frequency component can be expressed as n=1,2;

[0108]

[0109] Where δ is a small amount that is not equal to 0, then the brightness gain is defined as g n ,n=1,2:

[0110]

[0111] At this time, the adaptive fusion parameters of B1 Adaptive fusion parameters of B2

[0112] According to the adaptive fusion parameters of the low-frequency components of the medium-wave image and the low-frequency components of the long-wave image, the low-frequency component fusion weights of the medium-wave image and the low-frequency component fusion weights of the long-wave image are weighted fused to obtain the low-frequency fusion result: B = |α1W1 l B1-α2W2 l B2|.

[0113] S9: Determine adaptive fusion parameters of the high frequency components of the medium wave image and the long wave image in S2. Similar to S8, the process of determining the adaptive fusion parameters of the high frequency components of the medium wave image and the long wave image in S9 includes:

[0114] Calculate the logarithmic mean of the brightness values ​​of the high-frequency component D1 of the medium-wave image and the high-frequency component D2 of the long-wave image n=1,2;

[0115]

[0116] At this time, the adaptive fusion parameters of D1 Adaptive fusion parameters of D2

[0117] According to the adaptive fusion parameters of the high-frequency components of the medium-wave image and the long-wave image, the high-frequency component fusion weights of the medium-wave image and the high-frequency component fusion weights of the long-wave image are weightedly fused to obtain the high-frequency fusion result:

[0118] S10: After background removal, the low-frequency fusion result in S8 and the high-frequency fusion result in S9 are subjected to the fusion image A, where A = |γD-B|; γ is an adaptive adjustment factor, which is determined by the brightness ratio of some pixels in the high-frequency and low-frequency fusion images. If the average brightness of the last 15% of the pixels in the high-frequency fusion result is t D , the average brightness of the top 15% pixels in the low-frequency fusion result image is t B ,but See also Figure 2 It can be seen that the target in the fused image of this embodiment is very obvious.

[0119] S11: Training a general infrared image target detection network. This embodiment performs target recognition detection on the fused image. The YoloV5 target recognition network is trained on the infrared image target recognition public dataset to obtain a general infrared image target recognition training model. The background-removed image is fed into the model for inference to obtain the target detection result. Figure 3 It can be seen that compared with the medium-wave and long-wave image detection (0 ships detected), the fused image detection accuracy of this embodiment is significantly improved (2 ships were detected).

[0120] It can be concluded that compared with learning-based algorithms that need to learn for specific sensor imaging results, this algorithm has stronger algorithm versatility, has little correlation with sensor characteristics, and can be flexibly applied to medium and long-wave imaging sensors with different imaging modes, different manufacturers, and different optoelectronic devices; in addition, based on adaptive guided filtering, feature-based weight judgment can be performed on fusion weights in different image scale spaces, while retaining the spatial structural consistency of the fusion weights. The fused image has a stronger interpretative fusion result based on structural semantics, achieving a more ideal background removal effect, which is helpful for human visual attention and computer vision processing.

[0121] The present invention adopts box filtering to realize the rapid solution of guided filtering. The processing time on single-channel images of different sizes does not exceed 10ms, which can realize real-time processing in the engineering field. The fusion result can be used for target recognition by a general network based on infrared image target recognition, completing the decoupling of recognition and image fusion, and increasing the reusability of the fusion algorithm.

[0122] On the third aspect, an embodiment of the present application provides a two-color image fusion and anti-interference device based on adaptive guided filtering. The two-color image fusion and anti-interference device based on adaptive guided filtering can be a personal computer (PC), a laptop computer, a server, or other device with data processing capabilities.

[0123] Reference Figure 4 , Figure 4The hardware structure diagram of the dual-color image fusion and anti-interference device based on adaptive guided filtering involved in the embodiment of the present application is shown in FIG. In the embodiment of the present application, the dual-color image fusion and anti-interference device based on adaptive guided filtering may include a processor, a memory, a communication interface, and a communication bus.

[0124] The communication bus may be of any type and is used to interconnect the processor, memory, and communication interface.

[0125] Communication interfaces include input / output (I / O) interfaces, physical interfaces, and logical interfaces, which are used to interconnect components within the adaptive guided filtering-based two-color image fusion and anti-interference device, as well as interfaces used to interconnect the device with other devices (such as other computing devices or user devices). Physical interfaces can include Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc. User devices can include displays, keyboards, etc.

[0126] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0127] The processor may be a general-purpose processor that can call a two-color image fusion anti-interference program based on adaptive guided filtering stored in a memory and execute the two-color image fusion anti-interference method based on adaptive guided filtering provided in an embodiment of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the two-color image fusion anti-interference program based on adaptive guided filtering is called can refer to the various embodiments of the two-color image fusion anti-interference method based on adaptive guided filtering in the present application, and will not be repeated here.

[0128] Those skilled in the art will understand that Figure 4 The hardware structure shown in the figure does not constitute a limitation to the present application and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0129] In a third aspect, an embodiment of the present application also provides a computer-readable storage medium.

[0130] The computer-readable storage medium of the present application stores a two-color image fusion and anti-interference program based on adaptive guided filtering, wherein when the two-color image fusion and anti-interference program based on adaptive guided filtering is executed by a processor, the steps of the two-color image fusion and anti-interference method based on adaptive guided filtering as described above are implemented.

[0131] Among them, the method implemented when the two-color image fusion anti-interference program based on adaptive guided filtering is executed can refer to the various embodiments of the two-color image fusion anti-interference method based on adaptive guided filtering in this application, and will not be repeated here.

[0132] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0133] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device to execute the methods described in each embodiment of the present application.

[0134] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit the "first", "second" and "third" to different types.

[0135] In the description of the embodiments of this application, the words "exemplary," "for example," or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary," "for example," or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0136] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0137] In some processes described in the embodiments of the present application, multiple operations or steps are included that appear in a specific order. However, it should be understood that these operations or steps may not be performed in the order in which they appear in the embodiments of the present application or may be performed in parallel. The sequence numbers of the operations are only used to distinguish between different operations, and the sequence numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations or steps may be performed in sequence or in parallel, and these operations or steps may be combined.

[0138] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device to execute the methods described in each embodiment of the present application.

[0139] The above are only specific implementations of the embodiments of the present invention, but the scope of protection of the embodiments of the present invention is not limited to them. Any person skilled in the art can easily conceive of various equivalent modifications or replacements within the technical scope disclosed in the embodiments of the present invention, and such modifications or replacements should be included in the scope of protection of the embodiments of the present invention. Therefore, the scope of protection of the embodiments of the present invention should be based on the scope of protection of the claims.

Claims

1. A dual-color image fusion anti-interference method based on adaptive guided filtering, characterized in that: The method comprises the following steps: Acquire medium-wave and long-wave images that need to be fused; After feature extraction of the medium wave image and the long wave image, a feature map of the medium wave image and a feature map of the long wave image are obtained; after denoising the feature map, a low-noise feature map is obtained; after comparing pixels with the same coordinates of the low-noise feature map, an initial attention weight map is obtained; After performing guided filtering on the initial attention weight map, a fusion weight of low-frequency components of the medium-wave image and the long-wave image, and a fusion weight of high-frequency components of the medium-wave image and the long-wave image are obtained; After scale decomposition of the medium-wave image and the long-wave image, the low-frequency component and the high-frequency component of the medium-wave image, as well as the low-frequency component and the high-frequency component of the long-wave image are obtained; according to the adaptive fusion parameters of the low-frequency components, the fusion weights of the low-frequency components are weightedly fused to obtain the low-frequency fusion result; according to the adaptive fusion parameters of the high-frequency components, the fusion weights of the high-frequency components are weightedly fused to obtain the high-frequency fusion result.

2. The dual-color image fusion anti-interference method based on adaptive guided filtering according to claim 1, characterized in that: The process of extracting features from medium-wave images and long-wave images includes: After Laplace filtering of the medium wave image and the long wave image, the feature map of the medium wave image and the feature map of the long wave image F are obtained. n (x, y), n represents the feature map of the medium wave image or the feature map of the long wave image, (x, y) represents the current pixel coordinates; F n (x,y)=I n (x,y)*L(x,y) L represents the Laplace operator.

3. The method for dual-color image fusion and anti-interference based on adaptive guided filtering according to claim 2, characterized in that: The low noise characteristic graph F n The calculation formula for ′(x,y) is: Where G is a low-pass Gaussian filter with parameters r and σ.

4. The method for dual-color image fusion and anti-interference based on adaptive guided filtering according to claim 3, characterized in that: The initial attention weight map The calculation formula is:

5. The method for dual-color image fusion and anti-interference based on adaptive guided filtering according to claim 4, characterized in that: The low-frequency component fusion weight The calculation formula is: Among them I n represents a medium-wave image or a long-wave image; r1 is 0.075 times the shortest side of the initial attention weight map, is 0.4; The high frequency component fusion weight The calculation formula is: r2 is 0.01 times the shortest side of the initial attention weight map, 6. The method for dual-color image fusion and anti-interference based on adaptive guided filtering according to claim 5, characterized in that: The process of determining the adaptive fusion parameters of the low-frequency component includes: The image size is defined as w×n, and the maximum pixel value of the low-frequency component B1 of the medium-wave image and the low-frequency component B2 of the long-wave image is and The logarithmic mean of the brightness values ​​of the low-frequency components of the medium-wave image and the long-wave image is The calculation formula is: δ is a small amount that is not 0, and the brightness gain is defined as g n : Then the adaptive fusion parameter of B1 is Adaptive fusion parameters of B2 The calculation formula of the low-frequency fusion result is: The process of determining the adaptive fusion parameters of the high-frequency component includes: Calculate the logarithmic mean of the brightness values ​​of the high-frequency component D1 of the medium-wave image and the high-frequency component D2 of the long-wave image Adaptive fusion parameters of D1 Adaptive fusion parameters of D2 The calculation formula of the high-frequency fusion result is:

7. The method for dual-color image fusion and anti-interference based on adaptive guided filtering according to any one of claims 1 to 5, characterized in that: The process of performing scale decomposition on the medium-wave image and the long-wave image includes: performing Gaussian filtering on the medium-wave image and the long-wave image respectively to obtain a low-frequency component of the medium-wave image and a low-frequency component of the long-wave image; subtracting the medium-wave image from the low-frequency component of the medium-wave image to obtain a high-frequency component of the medium-wave image; and subtracting the long-wave image from the low-frequency component of the long-wave image to obtain a high-frequency component of the long-wave image.

8. The method for dual-color image fusion and anti-interference based on adaptive guided filtering according to any one of claims 1 to 5, characterized in that: The method further comprises the following steps: performing background removal on the low-frequency fusion result and the high-frequency fusion result to obtain a fused image.

9. The method for dual-color image fusion and anti-interference based on adaptive guided filtering according to claim 8, characterized in that: The calculation formula of the fused image A is: A = |γD-B|; where γ is an adaptive adjustment factor, if the average value of the pixels with the brightness ranking of the last 10% to 20% in the high-frequency fusion result is t D , the average brightness of the top 10% to 20% pixels in the low-frequency fusion result is t B ,but 10. A dual-color image fusion anti-interference device based on adaptive guided filtering, characterized in that: The two-color image fusion and anti-interference device based on adaptive guided filtering includes a processor, a memory, and a two-color image fusion and anti-interference program based on adaptive guided filtering stored on the memory and executable by the processor. When the two-color image fusion and anti-interference program based on adaptive guided filtering is executed by the processor, the steps of the two-color image fusion and anti-interference method based on adaptive guided filtering as described in any one of claims 1 to 9 are implemented.

Images