Image decomposition fusion method and system with adaptive filtering parameters

Abstract

The application relates to the technical field of image fusion, in particular to an image decomposition fusion method and system with adaptive filtering parameters, which comprises the following steps: acquiring each source image to be fused; calculating the pixel intensity quartile skewness of each source image; calculating the Gaussian filtering standard deviation, the spatial domain standard deviation of co-occurrence filtering and the co-occurrence matrix value range Gaussian standard deviation corresponding to each source image based on the pixel intensity quartile skewness; for each source image, performing Gaussian filtering processing based on the Gaussian filtering standard deviation, performing co-occurrence filtering processing based on the spatial domain standard deviation of co-occurrence filtering and the co-occurrence matrix value range Gaussian standard deviation, and then performing difference operation to obtain a base layer, a feature layer and an edge layer; fusing the base layer, the feature layer and the edge layer of each source image respectively, and linearly superimposing three fusion results to obtain a final fusion image. The application can realize fine separation and effective fusion of source image features, and significantly improves the definition and detail retention capability of the fusion image.

Description

Technical Field

[0001] This application relates to the field of image fusion technology, specifically to an image decomposition and fusion method and system with adaptive filtering parameters. Background Technology

[0002] In the field of image processing, image fusion technology, as an important data processing method, is widely used in complex tasks such as target detection, recognition, and tracking. With the increasing amount of multi-source heterogeneous image data (such as images from different sensors, spectral bands, or imaging times), how to effectively integrate this image information to eliminate redundancy, reduce the impact of uncertainty, and comprehensively extract features has become the key to improving the accuracy of subsequent analysis.

[0003] In existing technologies, some image fusion methods are based on multi-scale image transformation. Specifically, the source image is decomposed by fixing filtering parameters to extract feature components at different scales, and then reconstructed using corresponding fusion strategies. This image fusion method based on multi-scale image transformation can, to some extent, separate the fundamental and high-frequency information of the image. By using tools such as Gaussian filtering or co-occurrence filtering to process the image, basic feature layer and edge layer extraction can be achieved, thereby preserving some texture and contour details during the fusion process.

[0004] However, because fixed filtering parameters cannot adapt to the pixel intensity distribution characteristics of different source images, existing image fusion methods based on multi-scale image transformation struggle to finely separate texture details and edge information during the decomposition process. For example, for images with varying degrees of pixel intensity distribution skewness, Gaussian filtering and co-occurrence filtering with fixed parameters may either over-smooth or under-preserve, resulting in insufficient extraction of feature and edge layers. Consequently, the fused image may exhibit visual problems such as unclear targets, blurred edges, or missing textures. Summary of the Invention

[0005] To address the technical problem that existing image fusion methods based on multi-scale image transformation, which use fixed filtering parameters, are unable to adapt to the pixel intensity distribution characteristics of different source images, resulting in insufficient feature extraction and unclear targets and blurred edges in the fused image, this application provides an image decomposition and fusion method and system with adaptive filtering parameters. By adaptively adjusting the filtering parameters according to the interquartile skewness of image pixel intensity, more refined feature separation is achieved, effectively improving the clarity and detail retention of the fused image.

[0006] In a first aspect, this application provides an image decomposition and fusion method with adaptive filtering parameters, comprising the following steps:

[0007] S1. Obtain the source images to be fused, all of which have the same size;

[0008] S2. Calculate the pixel intensity quartile skewness for each source image. The formula is:

[0009]

[0010] in, This indicates the current source image, whose pixel intensity values ​​are in the range [0, 255].

[0011] The truncated mean of pixel intensity values ​​of the remaining pixels in the current source image after removing the top 10% and bottom 10% of pixels with the highest and lowest pixel intensity values.

[0012] This represents the average pixel intensity of all pixels in the current source image;

[0013] This represents the interquartile range of the current source image;

[0014] S3. For each source image, calculate the corresponding Gaussian filter standard deviation, spatial domain standard deviation of co-occurrence filter, and Gaussian standard deviation of co-occurrence matrix value domain based on pixel intensity quartile skewness;

[0015] S4. Gaussian filtering is performed on the source images based on the standard deviation of the Gaussian filter corresponding to each source image to obtain the Gaussian filtering result. Co-occurrence filtering is performed on the source images based on the spatial domain standard deviation and the Gaussian standard deviation of the co-occurrence matrix value domain corresponding to each source image to obtain the co-occurrence filtering result. Then, the difference operation is performed on the Gaussian filtering result and the co-occurrence filtering result to obtain the basal layer, feature layer and edge layer corresponding to each source image.

[0016] S5. Fuse the base layers corresponding to each source image to obtain a fused base layer;

[0017] The feature layers corresponding to each source image are fused to obtain a fused feature layer;

[0018] The edge layers corresponding to each source image are fused to obtain a fused edge layer;

[0019] S6. Linearly stack the fused base layer, fused feature layer, and fused edge layer to obtain the fused image.

[0020] In step S1, by acquiring each source image to be fused and ensuring that each source image has the same size, a standardized data foundation is provided for subsequent unified image decomposition and fusion processing. This avoids registration problems and feature mismatch problems caused by differences in image size, and ensures the spatial consistency of the decomposition and fusion process.

[0021] In step S2, by calculating the pixel intensity quartile skewness of each source image, the truncated mean after removing the influence of extreme values, the average pixel intensity reflecting the overall brightness level, and the interquartile range measuring the degree of data dispersion are combined. This can more comprehensively and resiliently characterize the pixel intensity distribution characteristics of the image itself, providing an accurate and reliable basis for the adaptive setting of subsequent filtering parameters.

[0022] In step S3, for each source image, the standard deviation of the Gaussian filter, the spatial domain standard deviation of the co-occurrence filter, and the Gaussian standard deviation of the co-occurrence matrix value domain are calculated based on the quartile skewness of the pixel intensity. This allows the key filtering parameters to be dynamically adjusted according to the statistical characteristics of different images, overcoming the problem of poor adaptability of fixed parameters to different images and laying the parameter foundation for subsequent targeted decomposition.

[0023] In step S4, Gaussian filtering is performed on the source images based on the standard deviation of the Gaussian filter corresponding to each source image, and co-occurrence filtering is performed based on the co-occurrence filtering parameters corresponding to each source image. The results are then subjected to differential operations to separate the basal layer, feature layer, and edge layer. The adaptive parameter Gaussian filtering achieves targeted smoothing of low-frequency components of the image. At the same time, the advantages of adaptive parameter co-occurrence filtering in edge preservation and noise reduction are utilized. Combined with differential operations, noise can be effectively suppressed while preserving the overall contour and brightness information of the image, providing a clean basal layer, and effectively separating and extracting texture details and sharp edges in the image.

[0024] In step S5, by fusing the base layer, feature layer and edge layer corresponding to each source image respectively to obtain the fused base layer, fused feature layer and fused edge layer, the advantageous information of each source image in each layer can be preserved to the maximum extent, providing a sufficient information foundation for generating a high-quality fused image.

[0025] In step S6, a fused image is obtained by linearly superimposing the fused base layer, fused feature layer, and fused edge layer. The different feature components after optimization and fusion are effectively reconstructed into a complete image, ensuring the comprehensive improvement of the final fusion result in terms of overall contrast, texture detail, and edge sharpness.

[0026] It should be further noted that in step S2, the interquartile range of the current source image The calculation formula is:

[0027]

[0028] in, This represents the first quartile in the distribution of pixel intensity values ​​in the current source image;

[0029] This represents the third quartile in the distribution of pixel intensity values ​​of the current source image.

[0030] By explicitly specifying that the interquartile range of the current source image is calculated using the difference between the first and third quartiles, a specific method for measuring the dispersion of image pixel intensity values ​​is provided. This method is less affected by extreme values, making the subsequently calculated pixel intensity quartile skewness more reliably reflect the distribution characteristics of the image itself.

[0031] It should be further noted that in step S3, the standard deviation of the Gaussian filter... Spatial domain standard deviation of co-occurrence filtering Co-occurrence matrix range Gaussian standard deviation The calculation formula is:

[0032]

[0033]

[0034]

[0035] in, The standard deviation coefficient of Gaussians is the preset value.

[0036] The standard deviation coefficients of the pre-defined co-occurrence filter spatial domain;

[0037] The standard deviation coefficients of the Gaussian coefficients are the range of the predefined co-occurrence matrix.

[0038] By establishing linear calculation formulas between the Gaussian filter standard deviation, the spatial domain standard deviation of the co-occurrence filter, the Gaussian standard deviation of the co-occurrence matrix value domain, and the pixel intensity quartile skewness, and introducing preset coefficients, a specific implementation method is provided to directly and linearly map image statistical features to key filtering parameters, ensuring the computational efficiency and explicitness of the parameter adaptation process.

[0039] It should be further explained that, .

[0040] It should be further noted that in step S4, Gaussian filtering is performed using a Gaussian convolution kernel. The expression is:

[0041]

[0042] in, This represents any point in the Gaussian convolution kernel relative to the center point. coordinates The x-axis is... The vertical axis is used as the coordinate.

[0043] This represents the standard deviation of the Gaussian filter.

[0044] By specifying the use of Gaussian convolution kernels with specific expressions for Gaussian filtering, the core computational tool for basal layer extraction is clarified. Its centrally symmetric exponential weight distribution ensures the smoothness and isotropy of the filtering process, providing a standard operational guarantee for obtaining a good basal layer.

[0045] It should be further noted that in step S4, a co-occurrence filter is used for co-occurrence filtering. The expression for the co-occurrence filter is:

[0046]

[0047] in, Indicates the center pixel;

[0048] Represents the center pixel The new pixel intensity value obtained after co-occurrence filtering;

[0049] The center pixel of the preset size ;

[0050] Indicates in the neighborhood Inside, except for the center pixel Any pixel other than the one mentioned above;

[0051] Represents based on the center pixel With pixels The one-dimensional Gaussian function of the Euclidean distance between them is expressed as:

[0052]

[0053] This represents the spatial domain standard deviation of the co-occurrence filter;

[0054] Represents the center pixel With pixels The Euclidean distance between them;

[0055] Indicates the current source image In the middle, the center pixel Pixel intensity value;

[0056] Indicates the current source image In the middle, pixels Pixel intensity value;

[0057] This represents the normalized co-occurrence matrix, used to measure the center pixel. pixel intensity value with neighboring pixels pixel intensity value The co-occurrence relationship between them is expressed as:

[0058]

[0059] Represents pixel value In the current source image The number of occurrences in is expressed as:

[0060]

[0061] Represents pixel value In the current source image The number of occurrences in is expressed as:

[0062]

[0063] The co-occurrence matrix is ​​expressed as:

[0064]

[0065] Represents pixels With pixels The Euclidean distance between them;

[0066] The range of the co-occurrence matrix represents the Gaussian standard deviation.

[0067] This is an indicator function that executes when the condition within the curly braces is true. ,otherwise .

[0068] By explicitly specifying the use of a co-occurrence filter expression combining a one-dimensional Gaussian function based on Euclidean distance and a co-occurrence matrix for co-occurrence filtering, and utilizing spatial proximity and pixel intensity similarity to jointly determine the filtering weights, this method can effectively preserve edge information while smoothing the internal regions of the image, providing high-quality preliminary processing results for subsequent separation of feature layers and edge layers.

[0069] It should be further explained that step S4 specifically includes:

[0070] S401. Based on the standard deviation of the Gaussian filter corresponding to the source image, apply Gaussian filtering to each source image. conduct The first-order iterative decomposition yields the th Gaussian filtering results ;

[0071] Based on the spatial domain standard deviation and Gaussian standard deviation of the co-occurrence matrix corresponding to the source image, co-occurrence filtering is used to analyze the source image. conduct The first-order iterative decomposition yields the th Level co-occurrence filtering results ;

[0072] Among them, Gaussian filtering is used on the source image Conduct the first The first-order iterative decomposition yields the th Gaussian filtering results The formula is:

[0073]

[0074] Use co-occurrence filtering on the source image Conduct the first The first-order iterative decomposition yields the th Level co-occurrence filtering results The formula is:

[0075]

[0076] in, Represents the standard deviation of the Gaussian filter corresponding to the source image. For the first Gaussian filtering results Perform Gaussian filtering;

[0077] Represents the spatial domain standard deviation of the co-occurrence filter based on the source image. Co-occurrence matrix range Gaussian standard deviation For the first Level co-occurrence filtering results Perform co-occurrence filtering;

[0078] Indicates the total decomposition series;

[0079] Indicates the decomposition level index. ;

[0080] ;

[0081] S402. The first Gaussian filtering results As the base layer :

[0082] ;

[0083] S403. Differential Separation Feature Layer and edge layer :

[0084]

[0085]

[0086] in, Indicates the first Gaussian filtering results.

[0087] By specifically defining the steps for differential operations based on Gaussian filtering and co-occurrence filtering, a clear operational flow is provided that separates the base layer, feature layer, and edge layer, ensuring that texture details and edge structures are effectively separated.

[0088] It should be further explained that the total decomposition series... .

[0089] It should be further explained that in step S5, for the case of fusing two source images, a weighted fusion strategy is used to fuse the base layers corresponding to each source image to obtain a fused base layer.

[0090] In the fusion base layer, coordinate points pixel intensity value The expression is:

[0091]

[0092] in, This represents the coordinates of the base layer corresponding to the first source image. Pixel intensity value;

[0093] This indicates the coordinates of the basal layer corresponding to the second source image. Pixel intensity value;

[0094] The fusion weight coefficients for the basal layer corresponding to the first source image are expressed as follows:

[0095] .

[0096] For the fusion of two source images, an adaptive weighting coefficient based on the pixel intensity value of the basal layer is used for weighted fusion strategy. This allows the contribution of each source image's basal layer to be dynamically adjusted according to the local brightness information of the basal layer when fusing the basal layer. This helps the fused image maintain natural contrast and overall visual effect.

[0097] It should be further explained that in step S5, for the case of fusing two source images, the absolute value maximum fusion strategy is used to fuse the edge layers corresponding to each source image to obtain the fused edge layer.

[0098] In the blended edge layer, coordinate points pixel intensity value The expression is:

[0099]

[0100] in, This represents the coordinates of the edge layer corresponding to the first source image. Pixel intensity value;

[0101] This indicates the coordinates of the edge layer corresponding to the second source image. The pixel intensity value.

[0102] By employing the absolute value maximum fusion strategy in the edge layer fusion for the fusion of two source images, we can directly compare and select the edge information with stronger response at the corresponding position in the edge layer of each source image, ensuring that significant edge structures are preferentially and completely preserved in the fused image.

[0103] It should be further explained that in step S5, for the case of fusing two source images, a product fusion strategy of average gradient and spatial frequency is used to fuse the feature layers corresponding to each source image to obtain a fused feature layer.

[0104] In the fusion feature layer, coordinate points pixel intensity value The expression is:

[0105]

[0106] in, This represents the coordinates of the feature layer corresponding to the first source image. Pixel intensity value;

[0107] This represents the coordinates of the feature layer corresponding to the second source image. Pixel intensity value;

[0108] This indicates that in the feature layer corresponding to the first source image, the coordinate points... The average gradient value within a window of a specified size centered on the target;

[0109] This indicates that in the feature layer corresponding to the first source image, the coordinate points... Spatial frequency values ​​within a specified-size window centered on the target;

[0110] This indicates that in the feature layer corresponding to the second source image, the coordinate points... The average gradient value within a window of a specified size centered on the target;

[0111] This indicates that in the feature layer corresponding to the second source image, the coordinate points... Spatial frequency values ​​within a specified sized window centered on the center.

[0112] By considering the fusion of two source images, the product of average gradient and spatial frequency is used as the fusion weight coefficient in the feature layer fusion. This comprehensively considers the gradient sharpness of local regions and the overall activity, so that regions with rich texture and clear details receive higher weights in the fusion feature layer, effectively enhancing the texture detail representation of the fused image.

[0113] Secondly, this application provides an image decomposition and fusion system with adaptive filtering parameters for implementing the above-mentioned image decomposition and fusion method, comprising:

[0114] The source image acquisition module is used to acquire the source images to be fused.

[0115] The pixel intensity quartile skewness calculation module is used to calculate the pixel intensity quartile skewness of each source image.

[0116] The filter parameter calculation module is used to calculate the corresponding Gaussian filter standard deviation, spatial domain standard deviation of co-occurrence filter, and Gaussian standard deviation of co-occurrence matrix value domain for each source image based on the pixel intensity quartile skewness.

[0117] The hierarchical decomposition module is used to perform Gaussian filtering on the source images based on the Gaussian filter standard deviation corresponding to each source image to obtain the Gaussian filtering result, and to perform co-occurrence filtering on the source images based on the spatial domain standard deviation and the Gaussian standard deviation of the co-occurrence matrix value domain corresponding to each source image to obtain the co-occurrence filtering result. Then, the difference operation is performed on the Gaussian filtering result and the co-occurrence filtering result to obtain the basal layer, feature layer and edge layer corresponding to each source image.

[0118] The layered fusion module is used to: fuse the base layers corresponding to each source image to obtain a fused base layer; fuse the feature layers corresponding to each source image to obtain a fused feature layer; and fuse the edge layers corresponding to each source image to obtain a fused edge layer.

[0119] The fused image generation module is used to linearly superimpose the fused base layer, fused feature layer, and fused edge layer to obtain a fused image.

[0120] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described image decomposition and fusion method.

[0121] Fourthly, this application provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described image decomposition and fusion method.

[0122] As can be seen from the above technical solutions, this application has the following advantages:

[0123] 1. This application calculates the pixel intensity quartile skewness of each source image and adaptively adjusts the parameters of Gaussian filtering and co-occurrence filtering based on the pixel intensity quartile skewness. This enables a more accurate matching of the pixel intensity distribution characteristics of the image, thereby achieving finer feature separation during the decomposition process. This allows the filtering process to dynamically adapt to the feature changes of different images, improving the accuracy and robustness of feature extraction.

[0124] 2. This application utilizes adaptive parameters for Gaussian filtering and co-occurrence filtering, and combines differential operations to separate the basal layer, feature layer, and edge layer. This can more effectively preserve the texture details and edge information in the image, ensure the independence of the basal layer, feature layer, and edge layer, avoid information mixing, and provide purer feature components for image fusion.

[0125] 3. This application employs targeted fusion strategies for the basal layer, feature layer, and edge layer, and performs linear overlay reconstruction. This approach maximizes the preservation of the advantageous features of each layer, enhances the overall clarity and visual effect of the fused image, ensures the effective integration of key information, and significantly improves the detail and edge sharpness of the fused image. Attached Figure Description

[0126] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0127] Figure 1 This is a flowchart of an image decomposition and fusion method with adaptive filtering parameters in one embodiment of this application.

[0128] Figure 2 In one embodiment of this application, the first source image of the first scene is a visible light image.

[0129] Figure 3In one embodiment of this application, the second source image of the first scene is an infrared image.

[0130] Figure 4 This is a fused image of a first scene in one embodiment of this application.

[0131] Figure 5 In one embodiment of this application, the first source image of the second scene is a visible light image.

[0132] Figure 6 In one embodiment of this application, the second source image of the second scene is an infrared image.

[0133] Figure 7 This is a fused image of a second scene in one embodiment of this application.

[0134] Figure 8 This is a schematic block diagram of an image decomposition and fusion system with adaptive filtering parameters in one embodiment of this application.

[0135] Figure 9 This is a schematic diagram of the hardware structure of an electronic device in one embodiment of this application. Detailed Implementation

[0136] To make the purpose, features, and advantages of this application more apparent and understandable, specific embodiments and accompanying drawings will be used to clearly and completely describe the technical solution protected by this application. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0137] The image decomposition and fusion method involved in this application will be described in detail below. Specific details such as particular system structures and techniques are presented for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details.

[0138] In the image decomposition and fusion method disclosed in this application, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or sets thereof. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0139] To facilitate a clear description of the technical solutions of this application, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" do not necessarily imply that they are different.

[0140] The terms "one embodiment" or "some embodiments" used in this application mean that one or more embodiments of this application include the specific features, structures, or characteristics described in that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this application do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized.

[0141] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0142] The image decomposition and fusion method provided in this application embodiment is executed by a computer device, and correspondingly, the image decomposition and fusion system with adaptive filtering parameters runs in the computer device.

[0143] Figure 1 This is a flowchart of an image decomposition and fusion method with adaptive filtering parameters according to an embodiment of this application. Figure 1 The executing entity can be an image decomposition and fusion system. Depending on the specific requirements, the order of the steps in this flowchart can be changed, and some steps can be omitted.

[0144] like Figure 1 As shown, the image decomposition and fusion method with adaptive filtering parameters includes:

[0145] Step S1: Obtain the source images to be fused, all source images having the same size. In this embodiment, image decomposition and fusion are performed on two sets of source images from two different scenes, as follows:

[0146] right Figure 2 and Figure 3 Image decomposition and fusion are performed, where Figure 2 The first source image of the first scene is a visible light image; Figure 3 The second source image of the first scene is an infrared image;

[0147] right Figure 5 and Figure 6 Image decomposition and fusion are performed, where Figure 5 The first source image of the second scene is a visible light image; Figure 6The second source image for the second scene is an infrared image.

[0148] Step S2: Calculate the pixel intensity quartile skewness of each source image. The formula is:

[0149]

[0150] in, This indicates the current source image, whose pixel intensity values ​​are in the range [0, 255].

[0151] The truncated mean of pixel intensity values ​​of the remaining pixels in the current source image after removing the top 10% and bottom 10% of pixels with the highest and lowest pixel intensity values.

[0152] This represents the average pixel intensity of all pixels in the current source image;

[0153] The interquartile range of the current source image is represented by the following formula:

[0154]

[0155] in, This represents the first quartile in the distribution of pixel intensity values ​​in the current source image;

[0156] This represents the third quartile in the distribution of pixel intensity values ​​of the current source image.

[0157] After calculation Figure 2 of =2.3694;

[0158] Figure 3 of =3.8292;

[0159] Figure 5 of =1.2899;

[0160] Figure 6 of =1.8980;

[0161] Step S3: For each source image, calculate the corresponding Gaussian filter standard deviation, spatial domain standard deviation of the co-occurrence filter, and Gaussian standard deviation of the co-occurrence matrix value domain based on the pixel intensity quartile skewness. Spatial domain standard deviation of co-occurrence filtering Co-occurrence matrix range Gaussian standard deviation The calculation formula is:

[0162]

[0163]

[0164]

[0165] in, The standard deviation of Gaussians is the preset coefficient. ;

[0166] The standard deviation coefficients of the co-occurrence filter spatial domain are preset. ;

[0167] The co-occurrence matrix range is defined by the Gaussian standard deviation coefficients. .

[0168] After calculation Figure 2 of =2.3694;

[0169] Figure 3 of =3.8292;

[0170] Figure 5 of =1.2899;

[0171] Figure 6 of =1.8980;

[0172] Step S4: Gaussian filtering is performed on the source images based on the standard deviation of Gaussian filtering corresponding to each source image to obtain the Gaussian filtering result; co-occurrence filtering is performed on the source images based on the spatial domain standard deviation of co-occurrence filtering and the Gaussian standard deviation of co-occurrence matrix value domain corresponding to each source image to obtain the co-occurrence filtering result; then, the difference operation is performed on the Gaussian filtering result and the co-occurrence filtering result to obtain the basal layer, feature layer and edge layer corresponding to each source image.

[0173] Among them, Gaussian filtering is performed using Gaussian convolution kernels. The expression is:

[0174]

[0175] in, This represents any point in the Gaussian convolution kernel relative to the center point. coordinates The x-axis is... The vertical axis is used as the coordinate.

[0176] This represents the standard deviation of the Gaussian filter.

[0177] A co-occurrence filter is used for co-occurrence filtering. The expression for the co-occurrence filter is:

[0178]

[0179] in, Indicates the center pixel;

[0180] Represents the center pixel The new pixel intensity value obtained after co-occurrence filtering;

[0181] The center pixel of the preset size ;

[0182] Indicates in the neighborhood Inside, except for the center pixel Any pixel other than the one mentioned above;

[0183] Represents based on the center pixel With pixels The one-dimensional Gaussian function of the Euclidean distance between them is expressed as:

[0184]

[0185] This represents the spatial domain standard deviation of the co-occurrence filter;

[0186] Represents the center pixel With pixels The Euclidean distance between them;

[0187] Indicates the current source image In the middle, the center pixel Pixel intensity value;

[0188] Indicates the current source image In the middle, pixels Pixel intensity value;

[0189] This represents the normalized co-occurrence matrix, used to measure the center pixel. pixel intensity value with neighboring pixels pixel intensity value The co-occurrence relationship between them is expressed as:

[0190]

[0191] Represents pixel value In the current source image The number of occurrences in the expression is:

[0192]

[0193] Represents pixel value In the current source image The number of occurrences in is expressed as:

[0194]

[0195] The co-occurrence matrix is ​​expressed as:

[0196]

[0197] Represents pixels With pixels The Euclidean distance between them;

[0198] The range of the co-occurrence matrix represents the Gaussian standard deviation.

[0199] This is an indicator function that executes when the condition within the curly braces is true. ,otherwise ;

[0200] Step S4 specifically includes:

[0201] S401. Based on the standard deviation of the Gaussian filter corresponding to the source image, apply Gaussian filtering to each source image. conduct The first-order iterative decomposition yields the th Gaussian filtering results ;

[0202] Based on the spatial domain standard deviation and Gaussian standard deviation of the co-occurrence matrix corresponding to the source image, co-occurrence filtering is used to analyze the source image. conduct The first-order iterative decomposition yields the th Level co-occurrence filtering results ;

[0203] Among them, Gaussian filtering is used on the source image Conduct the first The first-order iterative decomposition yields the th Gaussian filtering results The formula is:

[0204]

[0205] Use co-occurrence filtering on the source image Conduct the first The first-order iterative decomposition yields the th Level co-occurrence filtering results The formula is:

[0206]

[0207] in, Represents the standard deviation of the Gaussian filter corresponding to the source image. For the first Gaussian filtering results Perform Gaussian filtering;

[0208] Represents the spatial domain standard deviation of the co-occurrence filter based on the source image. Co-occurrence matrix range Gaussian standard deviation For the first Level co-occurrence filtering results Perform co-occurrence filtering;

[0209] Indicates the total decomposition series. ;

[0210] Indicates the decomposition level index. ;

[0211] ;

[0212] S402. The first Gaussian filtering results As the base layer :

[0213] ;

[0214] S403. Differential Separation Feature Layer and edge layer :

[0215]

[0216]

[0217] in, Indicates the first Gaussian filtering results.

[0218] Step S5: A weighted fusion strategy is used to fuse the base layers corresponding to each source image of the same scene to obtain a fused base layer for the scene. In the fused base layer, the coordinate points... pixel intensity value The expression is:

[0219]

[0220] in, This represents the coordinates of the base layer corresponding to the first source image. Pixel intensity value;

[0221] This indicates the coordinates of the basal layer corresponding to the second source image. Pixel intensity value;

[0222] The fusion weight coefficients for the basal layer corresponding to the first source image are expressed as follows:

[0223] ;

[0224] A fusion strategy combining average gradient and spatial frequency is employed to fuse the feature layers corresponding to each source image of the same scene, resulting in a fused feature layer for that scene. In this fused feature layer, coordinate points... pixel intensity value The expression is:

[0225]

[0226] in, This represents the coordinates of the feature layer corresponding to the first source image. Pixel intensity value;

[0227] This represents the coordinates of the feature layer corresponding to the second source image. Pixel intensity value;

[0228] This indicates that in the feature layer corresponding to the first source image, the coordinate points... The average gradient value within a window of a specified size centered on the target;

[0229] This indicates that in the feature layer corresponding to the first source image, the coordinate points... Spatial frequency values ​​within a specified-size window centered on the target;

[0230] This indicates that in the feature layer corresponding to the second source image, the coordinate points... The average gradient value within a window of a specified size centered on the target;

[0231] This indicates that in the feature layer corresponding to the second source image, the coordinate points... Spatial frequency values ​​within a specified-size window centered on the target;

[0232] An absolute value maximum fusion strategy is used to fuse the edge layers corresponding to each source image of the same scene, resulting in a fused edge layer for that scene. Within this fused edge layer, coordinate points... pixel intensity value The expression is:

[0233]

[0234] in, This represents the coordinates of the edge layer corresponding to the first source image. Pixel intensity value;

[0235] This indicates the coordinates of the edge layer corresponding to the second source image. The pixel intensity value.

[0236] Step S6: Linearly superimpose the fusion base layer, fusion feature layer, and fusion edge layer of the same scene to obtain the fused image of the scene.

[0237] Among them, the fused image of the first scene is as follows Figure 4 As shown, Figure 4 For use Figure 2 and Figure 3 Image decomposition and fusion are performed to obtain the result.

[0238] The fused image of the second scene is as follows Figure 7 As shown, Figure 7 For use Figure 5 and Figure 6 The image is obtained by decomposing and fusing it.

[0239] The following are embodiments of the image decomposition and fusion system with adaptive filtering parameters provided in this application. This image decomposition and fusion system with adaptive filtering parameters belongs to the same inventive concept as the image decomposition and fusion methods in the above embodiments. For details not described in detail in the embodiments of the image decomposition and fusion system, please refer to the embodiments of the image decomposition and fusion method with adaptive filtering parameters described above.

[0240] like Figure 8 As shown, the image decomposition and fusion system with adaptive filtering parameters includes:

[0241] The source image acquisition module is used to acquire the source images to be fused.

[0242] The pixel intensity quartile skewness calculation module is used to calculate the pixel intensity quartile skewness of each source image.

[0243] The filter parameter calculation module is used to calculate the corresponding Gaussian filter standard deviation, spatial domain standard deviation of co-occurrence filter, and Gaussian standard deviation of co-occurrence matrix value domain for each source image based on the pixel intensity quartile skewness.

[0244] The hierarchical decomposition module is used to perform Gaussian filtering on the source images based on the Gaussian filter standard deviation corresponding to each source image to obtain the Gaussian filtering result, and to perform co-occurrence filtering on the source images based on the spatial domain standard deviation and the Gaussian standard deviation of the co-occurrence matrix value domain corresponding to each source image to obtain the co-occurrence filtering result. Then, the difference operation is performed on the Gaussian filtering result and the co-occurrence filtering result to obtain the basal layer, feature layer and edge layer corresponding to each source image.

[0245] The layered fusion module is used to: fuse the base layers corresponding to each source image to obtain a fused base layer; fuse the feature layers corresponding to each source image to obtain a fused feature layer; and fuse the edge layers corresponding to each source image to obtain a fused edge layer.

[0246] The fused image generation module is used to linearly superimpose the fused base layer, fused feature layer, and fused edge layer to obtain a fused image.

[0247] The image decomposition and fusion system in this embodiment is used to implement an image decomposition and fusion method with adaptive filtering parameters.

[0248] This application also provides an electronic device for implementing the various embodiments of this application. Figure 9 To illustrate the hardware structure of an electronic device according to various embodiments of this application, as shown in the following diagram... Figure 9 As shown, the electronic device includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor.

[0249] Those skilled in the art will understand that the electronic device structure involved in the embodiments of this application does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0250] In embodiments of this application, electronic devices include, but are not limited to, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.

[0251] In this application embodiment, the processor can be implemented using at least one of an Application-Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a processor, a controller, a microcontroller, a microprocessor, or an electronic unit designed to perform the functions described herein. In some cases, such implementations can be implemented within a controller. For software implementations, implementations such as processes or functions can be implemented with separate software modules that allow the performance of at least one function or operation. The software code can be implemented by a software application (or program) written in any suitable programming language, and the software code can be stored in memory and executed by the controller.

[0252] In addition, the electronic device includes some functional modules not shown, which will not be described in detail here.

[0253] Those skilled in the art will understand that the various aspects of the electronic device provided in this application can be implemented as a system, method, or program product. Therefore, the various aspects of this application can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."

[0254] This application also provides a storage medium storing a program product capable of implementing an image decomposition and fusion method with adaptive filtering parameters. In some possible implementations, various aspects of this application can also be implemented as a program product comprising program code that, when run on a terminal device, causes the terminal device to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this application.

[0255] The storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0256] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An image decomposition and fusion method with adaptive filtering parameters, characterized in that, include: S1. Obtain the source images to be fused, all of which have the same size; S2. Calculate the pixel intensity quartile skewness for each source image. The formula is: in, This indicates the current source image, whose pixel intensity values ​​are in the range [0, 255]. The truncated mean of pixel intensity values ​​of the remaining pixels in the current source image after removing the top 10% and bottom 10% of pixels with the highest and lowest pixel intensity values. This represents the average pixel intensity of all pixels in the current source image; This represents the interquartile range of the current source image; S3. For each source image, calculate the corresponding Gaussian filter standard deviation, spatial domain standard deviation of co-occurrence filter, and Gaussian standard deviation of co-occurrence matrix value domain based on pixel intensity quartile skewness; S4. Gaussian filtering is performed on the source images based on the standard deviation of the Gaussian filter corresponding to each source image to obtain the Gaussian filtering result. Co-occurrence filtering is performed on the source images based on the spatial domain standard deviation and the Gaussian standard deviation of the co-occurrence matrix value domain corresponding to each source image to obtain the co-occurrence filtering result. Then, the difference operation is performed on the Gaussian filtering result and the co-occurrence filtering result to obtain the basal layer, feature layer and edge layer corresponding to each source image. S5. Fuse the base layers corresponding to each source image to obtain a fused base layer; The feature layers corresponding to each source image are fused to obtain a fused feature layer; The edge layers corresponding to each source image are fused to obtain a fused edge layer; S6. Linearly stack the fused base layer, fused feature layer, and fused edge layer to obtain the fused image.

2. The image decomposition and fusion method as described in claim 1, characterized in that, In step S2, the interquartile range of the current source image The calculation formula is: in, This represents the first quartile in the distribution of pixel intensity values ​​in the current source image; This represents the third quartile in the distribution of pixel intensity values ​​of the current source image.

3. The image decomposition and fusion method as described in claim 1, characterized in that, In step S3, the standard deviation of the Gaussian filter Spatial domain standard deviation of co-occurrence filtering Co-occurrence matrix range Gaussian standard deviation The calculation formula is: in, The standard deviation coefficient of Gaussians is the preset value. The standard deviation coefficients of the pre-defined co-occurrence filter spatial domain; The standard deviation coefficients of the Gaussian coefficients are the range of the predefined co-occurrence matrix.

4. The image decomposition and fusion method as described in claim 3, characterized in that, 。 5. The image decomposition and fusion method as described in claim 1, characterized in that, Step S4 specifically includes: S401. Based on the standard deviation of the Gaussian filter corresponding to the source image, apply Gaussian filtering to each source image. conduct The first-order iterative decomposition yields the th Gaussian filtering results ; Based on the spatial domain standard deviation and Gaussian standard deviation of the co-occurrence matrix corresponding to the source image, co-occurrence filtering is used to analyze the source image. conduct The first-order iterative decomposition yields the th Level co-occurrence filtering results ; Among them, Gaussian filtering is used on the source image Proceed to the first The first-order iterative decomposition yields the th Gaussian filtering results The formula is: Use co-occurrence filtering on the source image Proceed to the first The first-order iterative decomposition yields the th Level co-occurrence filtering results The formula is: in, Represents the standard deviation of the Gaussian filter corresponding to the source image. For the first Gaussian filtering results Perform Gaussian filtering; Represents the spatial domain standard deviation of the co-occurrence filter based on the source image. Co-occurrence matrix range Gaussian standard deviation For the first Level co-occurrence filtering results Perform co-occurrence filtering; Indicates the total decomposition series; Indicates the decomposition level index. ; ； S402. The first Gaussian filtering results As the base layer : ； S403. Differential Separation Feature Layer and edge layer : in, Indicates the first Gaussian filtering results.

6. The image decomposition and fusion method as described in claim 1, characterized in that, Total decomposition series .

7. The image decomposition and fusion method as described in claim 1, characterized in that, In step S5, for the case of fusing two source images, a weighted fusion strategy is used to fuse the base layers corresponding to each source image to obtain a fused base layer. In the fusion base layer, coordinate points pixel intensity value The expression is: in, This represents the coordinates of the base layer corresponding to the first source image. Pixel intensity value; This indicates the coordinates of the basal layer corresponding to the second source image. Pixel intensity value; The fusion weight coefficients for the basal layer corresponding to the first source image are expressed as follows: 。 8. The image decomposition and fusion method as described in claim 1, characterized in that, In step S5, for the case of fusing two source images, the absolute value maximum fusion strategy is used to fuse the edge layers corresponding to each source image to obtain the fused edge layer; In the blended edge layer, coordinate points pixel intensity value The expression is: in, This represents the coordinates of the edge layer corresponding to the first source image. Pixel intensity value; This indicates the coordinates of the edge layer corresponding to the second source image. The pixel intensity value.

9. The image decomposition and fusion method as described in claim 1, characterized in that, In step S5, for the case of fusing two source images, a product fusion strategy of average gradient and spatial frequency is used to fuse the feature layers corresponding to each source image to obtain a fused feature layer. In the fusion feature layer, coordinate points pixel intensity value The expression is: in, This represents the coordinates of the feature layer corresponding to the first source image. Pixel intensity value; This represents the coordinates of the feature layer corresponding to the second source image. Pixel intensity value; This indicates that in the feature layer corresponding to the first source image, the coordinate points... The average gradient value within a window of a specified size centered on the target; This indicates that in the feature layer corresponding to the first source image, the coordinate points... Spatial frequency values ​​within a specified-size window centered on the target; This indicates that in the feature layer corresponding to the second source image, the coordinate points... The average gradient value within a window of a specified size centered on the target; This indicates that in the feature layer corresponding to the second source image, the coordinate points... Spatial frequency values ​​within a specified sized window centered on the center.

10. An image decomposition and fusion system with adaptive filtering parameters, characterized in that, To implement the image decomposition and fusion method as described in any one of claims 1-9, comprising: The source image acquisition module is used to acquire the source images to be fused. The pixel intensity quartile skewness calculation module is used to calculate the pixel intensity quartile skewness of each source image. The filter parameter calculation module is used to calculate the corresponding Gaussian filter standard deviation, spatial domain standard deviation of co-occurrence filter, and Gaussian standard deviation of co-occurrence matrix value domain for each source image based on the pixel intensity quartile skewness. The hierarchical decomposition module is used to perform Gaussian filtering on the source images based on the Gaussian filter standard deviation corresponding to each source image to obtain the Gaussian filtering result, and to perform co-occurrence filtering on the source images based on the spatial domain standard deviation and the Gaussian standard deviation of the co-occurrence matrix value domain corresponding to each source image to obtain the co-occurrence filtering result. Then, the difference operation is performed on the Gaussian filtering result and the co-occurrence filtering result to obtain the basal layer, feature layer and edge layer corresponding to each source image. The layered fusion module is used to: fuse the base layers corresponding to each source image to obtain a fused base layer; fuse the feature layers corresponding to each source image to obtain a fused feature layer; and fuse the edge layers corresponding to each source image to obtain a fused edge layer. The fused image generation module is used to linearly superimpose the fused base layer, fused feature layer, and fused edge layer to obtain a fused image.

Images