A histogram feature extraction method for infrared polarization mosaic images

By directly extracting spatial polarization features from infrared polarization mosaic images and generating histogram descriptors, the problems of large information loss and calculation consumption in the prior art are solved, and efficient feature extraction and real-time video processing are achieved.

CN114359585BActive Publication Date: 2024-11-15NORTHWESTERN POLYTECHNICAL UNIV
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202111507158.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-10
Publication Date
2024-11-15
Estimated Expiration
2041-12-10

AI Technical Summary

Technical Problem

The existing infrared polarization mosaic image processing technology has information loss and high calculation consumption during the preprocessing process, which affects real-time performance and makes it difficult to directly extract features.

Method used

Using a method of extracting features directly from the original infrared polarization mosaic image, the first and second histograms are generated based on these graphs and cascaded to obtain histogram descriptors, avoiding Demosaic and polarization parameters solution steps.

Benefits of technology

It avoids information changes and loss, reduces calculation costs, improves the robustness and effectiveness of feature extraction, and is suitable for real-time video processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114359585B_ABST
    Figure CN114359585B_ABST
Patent Text Reader

Abstract

The invention discloses a histogram feature extraction method for an infrared polarization mosaic image. According to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image, a spatial polarization gradient map of the infrared polarization mosaic image is generated; according to the correlation between different polarization channels and a PPI image in the infrared polarization mosaic image, a polarization gradient map of the infrared polarization mosaic image is generated; based on the spatial polarization gradient map and the polarization gradient map, a first histogram and a second histogram are respectively generated; the first histogram and the second histogram are cascaded to obtain a histogram descriptor of the infrared polarization mosaic image; the method of the invention can directly extract spatial polarization features from the original mosaic image without preprocessing, thus avoiding information changes and information losses, while reducing calculation costs, improving the robustness and effectiveness of feature extraction, and being suitable for real-time video processing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of infrared polarization mosaic image processing, and in particular relates to a histogram feature extraction method for an infrared polarization mosaic image. Background Art

[0002] The infrared polarization mosaic image is collected by a sub-focal plane infrared polarization camera composed of a layer of micro-polarizer array processed on the infrared focal plane. Each 2×2 pixels in the micro-polarizer array is a group. Figure 3 As shown in (a), 2×2 pixels correspond to four different polarization filter angles, namely 0°, 45°, 90°, and 135°. The collected infrared polarization mosaic image is shown in Figure 3 (b) as shown.

[0003] Due to the characteristic mosaic structure of the acquired image, the traditional idea when extracting features is to first perform a series of preprocessing such as demosaicing and polarization parameter solution, and then further use the obtained polarization and spatial information to achieve target detection, recognition, tracking and other related tasks.

[0004] However, each step of the preprocessing process will lead to changes in information and inevitable information loss, and preprocessing will consume a lot of time, affecting the real-time performance of the method. Summary of the invention

[0005] The purpose of the present invention is to provide a histogram feature extraction method for infrared polarization mosaic images, which can directly extract features from the original infrared polarization mosaic images without the need for preprocessing steps such as demosaicing and polarization parameter calculation, thereby avoiding information changes and information losses.

[0006] The present invention adopts the following technical solution: a histogram feature extraction method for infrared polarization mosaic images, comprising the following steps:

[0007] Generate a spatial polarization gradient map of the infrared polarization mosaic image according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image;

[0008] According to the correlation between different polarization channels in the infrared polarization mosaic image and the PPI image, a polarization gradient map of the infrared polarization mosaic image is generated;

[0009] Generate a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map respectively;

[0010] The first histogram and the second histogram are cascaded to obtain the histogram descriptor of the infrared polarization mosaic image.

[0011] Further, generating a spatial polarization gradient map of an infrared polarization mosaic image includes:

[0012] Determine a spatial polarization gradient map of each pixel according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image;

[0013] The spatial polarization gradient map of the infrared polarization mosaic image is determined according to the spatial polarization gradient map of each pixel.

[0014] Further, determining the spatial polarization gradient map of each pixel includes:

[0015] According to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image, the spatial polarization gradient operator of each pixel is calculated;

[0016] The spatial polarization gradient operator of each pixel is convolved with the gray value of the pixel to obtain the spatial polarization gradient map of each pixel.

[0017] Furthermore, the spatial polarization gradient operator for each pixel is calculated by:

[0018] Generate a polarization correlation matrix of each pixel in the infrared polarization mosaic image according to the arrangement pattern of the micro-polarizer array used in the acquisition process of the infrared polarization mosaic image;

[0019] Based on the polarization correlation matrix and gradient operator of each pixel, the spatial polarization gradient operator of each pixel is calculated.

[0020] Further, generating a polarization gradient map of the infrared polarization mosaic image includes:

[0021] A polarization gradient operator of an infrared polarization mosaic image is generated according to the correlation between different polarization channels in the infrared polarization mosaic image and the PPI image;

[0022] The infrared polarization mosaic image is convolved with the polarization gradient operator to obtain the polarization gradient map of the infrared polarization mosaic image.

[0023] Further, generating a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map respectively includes:

[0024] Based on the spatial polarization gradient map, a first histogram is generated; wherein the first histogram includes a first gradient amplitude and a first angle direction.

[0025] Further, generating the first histogram and the second histogram based on the spatial polarization gradient map and the polarization gradient map respectively also includes:

[0026] A second histogram is generated based on the spatial polarization gradient map and the polarization gradient map; wherein the second histogram includes a second gradient amplitude and a second angle direction.

[0027] Another technical solution of the present invention: a histogram feature extraction device for infrared polarization mosaic images, comprising:

[0028] A first generating module generates a spatial polarization gradient map of the infrared polarization mosaic image according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image;

[0029] The second generation module generates a polarization gradient map of the infrared polarization mosaic image according to the correlation between different polarization channels in the infrared polarization mosaic image and the PPI image;

[0030] A third generating module, used to generate a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map respectively;

[0031] The cascade module is used to cascade the first histogram and the second histogram to obtain a histogram descriptor of the infrared polarization mosaic image.

[0032] Furthermore, the first generation module includes:

[0033] A first determination module is used to determine a spatial polarization gradient map of each pixel according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image;

[0034] The second determination module is used to determine the spatial polarization gradient map of the infrared polarization mosaic image according to the spatial polarization gradient map of each pixel.

[0035] Another technical solution of the present invention: a histogram feature extraction device for an infrared polarization mosaic image, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the above-mentioned histogram feature extraction method for an infrared polarization mosaic image when executing the computer program.

[0036] The beneficial effects of the present invention are as follows: the method of the present invention can directly extract spatial polarization features in the original mosaic image without the need for a series of preprocessing such as demosaicing and polarization parameter solution, thereby avoiding information changes and information losses, while reducing computational costs, improving the robustness and effectiveness of feature extraction, and being suitable for real-time video processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of a method for extracting histogram features of an infrared polarization mosaic image according to an embodiment of the present invention;

[0038] Figure 2 is a schematic diagram of the correlation of associated pixels in an embodiment of the present invention;

[0039] Figure 3 An example of an infrared polarization mosaic image collected in the prior art and a schematic diagram of polarization filtering angles of 2×2 pixels. DETAILED DESCRIPTION

[0040] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] The embodiment of the present invention discloses a method for extracting features of a histogram of an infrared polarization mosaic image. Figure 1 As shown, the method includes the following steps: generating a spatial polarization gradient map of the infrared polarization mosaic image according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image; generating a polarization gradient map of the infrared polarization mosaic image according to the correlation between different polarization channels and the PPI image in the infrared polarization mosaic image; generating a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map respectively; and cascading the first histogram and the second histogram to obtain a histogram descriptor of the infrared polarization mosaic image.

[0042] The method of the present invention can directly extract spatial polarization features from the original mosaic image without the need for a series of preprocessing such as demosaicing and polarization parameter calculation, thus avoiding information changes and information losses, while reducing computational costs, improving the robustness and effectiveness of feature extraction, and being suitable for real-time video processing.

[0043] In this embodiment, generating a spatial polarization gradient map of an infrared polarization mosaic image includes: generating a polarization correlation matrix for each pixel in the infrared polarization mosaic image according to the arrangement pattern of the micro-polarizer array used in the acquisition process of the infrared polarization mosaic image (i.e., according to the spatial correlation between pixels in the infrared polarization mosaic image and the correlation between different polarization channels); calculating a spatial polarization gradient operator for each pixel based on the polarization correlation matrix and gradient operator of each pixel; convolving the spatial polarization gradient operator of each pixel with the grayscale value of the pixel to obtain a spatial polarization gradient map of each pixel; and determining the spatial polarization gradient map of the infrared polarization mosaic image according to the spatial polarization gradient map of each pixel.

[0044] Gradient information has become an important feature in target detection, recognition, tracking and other related tasks. However, these gradient features only use spatial correlation. In polarization mosaic images, there is both spatial correlation and polarization correlation, which is the aliasing of spatial information and polarization information. Therefore, when constructing feature extraction operators, it is necessary to consider the aliasing of spatial and polarization information in the original mosaic image.

[0045] The present invention uses spatial distance to represent the spatial correlation between pixels. The spatial distance can be the actual distance between pixels or calculated by the Euclidean distance between pixels. Each pixel in the mosaic image has the same spatial correlation matrix, which is expressed as follows:

[0046]

[0047] Among them, f s represents the spatial correlation matrix of the center pixel, d i They respectively represent the spatial distance between the i-th neighborhood pixel and the central pixel.

[0048] In the present invention, the mutual projection coefficients between channels at different angles are used to represent the polarization correlation between different angles. In the vibration plane, the vibration of light waves is averagely symmetrical in all directions, and they have the same amplitude, so the vibration amplitude of non-polarized light in all directions in the vibration plane is the same. When non-polarized light passes through a polarizer in a certain direction, the vibration amplitude of the light in that direction will be greater than that in other directions, thus forming polarized light.

[0049] In this embodiment, the polarization correlation is described by using the component coefficients of the vibrating light waves in other directions in the direction of the polarizer. For example, the correlation between the pixel associated with 0° and the pixel associated with 45° can be described by the projection coefficient of the light in the 45° direction in the 0° direction, and the correlation between the pixel associated with 0° and the pixel associated with 135° can be described by the projection coefficient of the light in the 135° direction in the 0° direction. Because 0° and 90° are two orthogonal channels, the correlation between them is 0.

[0050] Specifically, the correlation between the pixel associated with 0° and the pixels associated with the other three angles 45°, 90° and 135° can be used Figure 2 To illustrate, Figure 2 (a) is a schematic diagram of the correlation between the 0° channel and the 45° channel. Figure 2 (b) is a schematic diagram of the correlation between the 0° channel and the 135° channel. From this figure, we can see that the polarization correlation matrix corresponding to the pixel associated with 0° can be expressed as follows:

[0051]

[0052] Among them, f p It represents the polarization correlation matrix between pixels associated with a certain angle and pixels associated with other angles, and r represents the amplitude of light.

[0053] Construct the correlation matrix f p , p = {0°, 45°, 90°, 135°}, then the polarization correlation matrix fp and the horizontal gradient operator f sx and the vertical gradient operator f sy By performing convolution, we can obtain the spatial polarization horizontal gradient operator and the spatial polarization vertical gradient operator, which are specifically expressed as follows:

[0054]

[0055]

[0056] Among them, f xp is the spatial polarization horizontal gradient operator, f yp is the spatial polarization vertical gradient operator.

[0057] In this embodiment of the present invention, the horizontal gradient operator f sx and the vertical gradient operator f sy It is expressed as follows:

[0058]

[0059]

[0060] After obtaining the spatial polarization horizontal gradient operator and the spatial polarization vertical gradient operator, they can be used to calculate the spatial polarization gradient map of the infrared polarization mosaic image. Specifically, the following formula can be used for calculation:

[0061]

[0062]

[0063] Among them, G xp is the spatial polarization horizontal gradient map of the infrared polarization mosaic image, G yp is the spatial polarization vertical gradient map of the infrared polarization mosaic image. Due to the arrangement pattern of the mosaic image, the final spatial polarization gradient map of the mosaic image includes the spatial polarization horizontal gradient map and the spatial polarization vertical gradient map, that is, G M = {G x ,G y In addition, since the above calculation process is performed on each pixel, it is necessary to combine the spatial polarization gradient maps of each pixel to obtain the spatial polarization gradient map G of the infrared polarization mosaic image. M .

[0064] In the process of combination, the method of taking the union can be used to achieve it, which is specifically expressed as follows:

[0065]

[0066]

[0067] Among them, G xp (p) is the spatial polarization horizontal gradient map of the pixel associated with angle p, G yp (p) is the spatial polarization vertical gradient map of the pixel associated with angle p.

[0068] In multispectral images, the spectral gradient is usually calculated by the difference between adjacent frequency bands. The pseudo-panchromatic image (PPI) represents the average image on all channels of the spectral image, and the PPI is closely related to all channels. Using this feature, it can be concluded that the difference between the mosaic spectral image and the PPI image can be used to represent the spectral gradient of the target. Since the polarization gradient is calculated by the difference between the pixels associated with adjacent polarization angles, and the PPI has a strong correlation with all channels, the conclusion of the PPI in the spectral mosaic image can be used for the polarization mosaic image, that is, the polarization gradient is represented by the difference between the polarization mosaic image and the PPI image.

[0069] In one embodiment of the present invention, a polarization gradient operator of the infrared polarization mosaic image is generated according to the correlation between different polarization channels in the infrared polarization mosaic image and the PPI image (i.e., according to the size of the filter used in the histogram feature extraction process); the infrared polarization mosaic image is convolved with the polarization gradient operator to obtain a polarization gradient map of the infrared polarization mosaic image.

[0070] Specifically, the difference between the polarization mosaic image and the PPI image is used to represent the polarization gradient. Since PPI is the average of the pixels corresponding to all channels, an average filter is used to estimate the PPI of the polarization mosaic image. The filter needs to consider all channels, and its size should not be less than the size of the minimum odd neighborhood window containing all channels. In this embodiment, the size of the filter M is 3×3, each value in M ​​is 1 / n, n is the number of times the channel appears in the filter window, and M is expressed as follows:

[0071]

[0072] Furthermore, based on the above conclusions, the polarization gradient map of the infrared polarization mosaic image is expressed as follows:

[0073]

[0074] Among them, f l =[1]-M represents the polarization gradient operator, and [1] represents a 3×3 matrix (the central element of the matrix is ​​1, and the other elements are 0). It can be seen from this that when calculating the polarization gradient map, the polarization gradient operator f can be directly used. l Convolve the original mosaic image to obtain the polarization gradient map G l .

[0075] After the gradient calculation, a histogram is constructed by the magnitude and direction of the gradient, which is then concatenated to form a mosaic spatial polarization gradient direction histogram descriptor for infrared polarization mosaic images.

[0076] Specifically, in the process of generating the first histogram and the second histogram, the specific method can be to generate the first histogram based on the spatial polarization gradient map, wherein the first histogram includes a first gradient amplitude and a first angle direction. Generate the second histogram based on the spatial polarization gradient map and the polarization gradient map, wherein the second histogram includes a second gradient amplitude and a second angle direction.

[0077] As a specific implementation method, the first histogram is obtained by using the spatial polarization horizontal gradient map G x and spatial polarization vertical gradient map G y To calculate the first gradient magnitude M xy and the first angle direction θ xy , whose expression is:

[0078]

[0079] θ xy =tan -1 (G x / G y ) (14)

[0080] The second histogram is the spatial polarization horizontal gradient map G x , spatial polarization vertical gradient map G y and the polarization gradient map G l To calculate the first gradient magnitude M xyl and the first angle direction Specifically:

[0081]

[0082]

[0083] In summary, the above-mentioned histogram is constructed in the local unit to represent the local spatial polarization information, similar to the classic HOG feature, and the histogram is constructed in 9 sensitive directions and 18 insensitive directions, and the histogram of each pair of amplitude and direction is calculated. Finally, the MSPHOG feature is obtained by connecting the histograms in series in the third dimension. The method of the present invention can directly extract the spatial polarization feature in the original mosaic image without a series of preprocessing such as demosaicing and polarization parameter solution, avoiding the change and loss of information, and reducing the computational cost. The robustness and effectiveness of feature extraction are improved, and it is suitable for real-time video processing.

[0084] In addition, another embodiment of the present invention also discloses a histogram feature extraction device for an infrared polarization mosaic image, including: a first generation module, used to generate a spatial polarization gradient map of the infrared polarization mosaic image according to the arrangement pattern of the micro-polarizer array used in the acquisition process of the infrared polarization mosaic image; a second generation module, used to generate a polarization gradient map of the infrared polarization mosaic image according to the size of the filter used in the histogram feature extraction process; a third generation module, used to generate a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map, respectively; and a cascade module, used to cascade the first histogram and the second histogram to obtain a histogram descriptor of the infrared polarization mosaic image.

[0085] Furthermore, the first generation module includes: a first determination module, used to determine the spatial polarization gradient map of each pixel based on the spatial correlation between pixels in the infrared polarization mosaic image and the correlation between different polarization channels; a second determination module, used to determine the spatial polarization gradient map of the infrared polarization mosaic image based on the spatial polarization gradient map of each pixel.

[0086] It should be noted that the information interaction, execution process and other contents between the modules of the above-mentioned device are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0087] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The functional modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0088] Another embodiment of the present invention also discloses a histogram feature extraction device for an infrared polarization mosaic image, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned histogram feature extraction method for an infrared polarization mosaic image when executing the computer program.

[0089] The device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the device may include more or fewer components, or a combination of certain components, or different components, such as an input / output device, a network access device, etc.

[0090] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0091] In some embodiments, the memory may be an internal storage unit of the device, such as a hard disk or memory of the device. In other embodiments, the memory may also be an external storage device of the device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the device. Further, the memory may also include both an internal storage unit of the device and an external storage device. The memory is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program, etc. The memory may also be used to temporarily store data that has been output or is to be output.

[0092] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. Those of ordinary skill in the art can appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented with electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

Claims

1. A histogram feature extraction method for infrared polarization mosaic images, It is characterized in that The following steps are involved: Generate a spatial polarization gradient map of the infrared polarization mosaic image according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image; generating a polarization gradient map of the infrared polarization mosaic image according to the correlation between different polarization channels in the infrared polarization mosaic image and the PPI image; Generate a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map, respectively; cascading the first histogram and the second histogram to obtain a histogram descriptor of the infrared polarization mosaic image; Generating the spatial polarization gradient map of the infrared polarization mosaic image includes: Determine a spatial polarization gradient map of each pixel according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image; Determine the spatial polarization gradient map of the infrared polarization mosaic image according to the spatial polarization gradient map of each pixel; Determining the spatial polarization gradient map of each pixel comprises: Calculating a spatial polarization gradient operator for each pixel according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image; Convolving the spatial polarization gradient operator of each pixel with the grayscale value of the pixel to obtain a spatial polarization gradient map of each pixel; Calculating the spatial polarization gradient operator for each pixel includes: Generate a polarization correlation matrix of each pixel in the infrared polarization mosaic image according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image; Calculating a spatial polarization gradient operator for each pixel based on the polarization correlation matrix and gradient operator of each pixel; Generating a polarization gradient map of the infrared polarization mosaic image includes: generating a polarization gradient operator of the infrared polarization mosaic image according to the correlation between different polarization channels in the infrared polarization mosaic image and the PPI image; The infrared polarization mosaic image is convolved with the polarization gradient operator to obtain a polarization gradient map of the infrared polarization mosaic image.

2. The method for extracting features from a histogram of an infrared polarization mosaic image according to claim 1, It is characterized in that Generating a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map respectively comprises: Based on the spatial polarization gradient map, a first histogram is generated; wherein the first histogram includes a first gradient amplitude and a first angle direction.

3. The histogram feature extraction method of infrared polarization mosaic image according to claim 2, It is characterized in that Generating a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map respectively also includes: A second histogram is generated based on the spatial polarization gradient map and the polarization gradient map; wherein the second histogram includes a second gradient amplitude and a second angle direction.

4. A histogram feature extraction device for infrared polarization mosaic images, It is characterized in that include: A first generating module, configured to generate a spatial polarization gradient map of the infrared polarization mosaic image according to the spatial correlation between pixels in the infrared polarization mosaic image and the correlation between different polarization channels; A second generating module generates a polarization gradient map of the infrared polarization mosaic image according to the correlation between different polarization channels in the infrared polarization mosaic image and the PPI image; A third generating module, configured to generate a first histogram and a second histogram based on the spatial polarization gradient map and the polarization gradient map, respectively; A cascading module, used for cascading the first histogram and the second histogram to obtain a histogram descriptor of the infrared polarization mosaic image; Generating the spatial polarization gradient map of the infrared polarization mosaic image includes: Determine a spatial polarization gradient map of each pixel according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image; Determine the spatial polarization gradient map of the infrared polarization mosaic image according to the spatial polarization gradient map of each pixel; Determining the spatial polarization gradient map of each pixel comprises: Calculating a spatial polarization gradient operator for each pixel according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image; Convolving the spatial polarization gradient operator of each pixel with the grayscale value of the pixel to obtain a spatial polarization gradient map of each pixel; Calculating the spatial polarization gradient operator for each pixel includes: Generate a polarization correlation matrix of each pixel in the infrared polarization mosaic image according to the spatial correlation between pixels and the correlation between different polarization channels in the infrared polarization mosaic image; Calculating a spatial polarization gradient operator for each pixel based on the polarization correlation matrix and gradient operator of each pixel; Generating a polarization gradient map of the infrared polarization mosaic image includes: generating a polarization gradient operator of the infrared polarization mosaic image according to the correlation between different polarization channels in the infrared polarization mosaic image and the PPI image; The infrared polarization mosaic image is convolved with the polarization gradient operator to obtain a polarization gradient map of the infrared polarization mosaic image.

5. A histogram feature extraction device for infrared polarization mosaic images, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the computer program, it implements the histogram feature extraction method of the infrared polarization mosaic image as described in any one of claims 1 to 3.

Citation Information

Patent Citations

  • A mosaic spatial spectrum gradient direction histogram extraction method for a snapshot type spectral image

    CN113486906A

  • Image processing apparatus

    US20120206581A1

  • Method and system of low-complexity histrogram of gradients generation for image processing

    US20170091575A1

  • Image processing device, image processing method, and imaging device

    WO2019082462A1