Frequency domain-space domain combined texture enhancement positioning method and system and storage medium

Abstract

The invention discloses a frequency domain-space domain combined texture enhancement positioning method and system and a storage medium, and the method comprises the steps: obtaining an original image, and carrying out the frequency domain enhancement and space domain enhancement, and obtaining a target image; performing dynamic illumination compensation based on the target image to obtain an optimized image; performing feature point detection and fusion based on the optimized image to obtain a fusion response graph; and performing feature point screening and positioning based on the fused response diagram to obtain a feature point set for positioning. According to the method, through frequency domain-space domain joint enhancement, dynamic illumination compensation and multi-scale feature fusion, the feature extraction problem in a low-texture scene is effectively solved, the positioning precision and the system robustness are remarkably improved, and the method has a wide industrial application prospect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and more particularly, to a frequency domain-space domain joint texture enhancement positioning method, system and storage medium. BACKGROUND

[0002] Under the background of the rapid development of artificial intelligence and intelligent perception systems, precise positioning and feature extraction technology plays an increasingly key role in many application scenarios. Especially in the fields of agricultural robot navigation, fine textile detection, mobile AR / VR perception systems, etc., traditional image positioning methods relying on texture distribution face major challenges. One of the most core problems is the lack of stable and repeatable key points and descriptors in "low texture scenes", which leads to a significant decline in the accuracy of subsequent image registration, target tracking, and three-dimensional reconstruction tasks. For example, in the farmland environment, the texture of the large-area grass surface tends to be consistent, the leaf structure changes weakly and fluctuates with the wind, making it difficult to extract stable feature points from visual images; for example, in high-density textile fabrics, due to uniform structure and repeated texture, traditional corner detection algorithms (such as Harris, FAST) frequently fall into the "pseudo-feature" trap, making it difficult to perform robust recognition.

[0003] Most of the current mainstream image feature extraction methods rely on spatial domain gradient changes or response strength analysis, and common methods include scale-invariant feature transform, speeded-up robust features, and direction binary pattern. However, in low-texture backgrounds, these methods cannot produce a sufficient number of high-response feature points, which in turn affects the system's ability to perceive targets. In addition, although some end-to-end feature learning methods based on deep learning have improved accuracy, they require a large amount of labeled data for training and high computational resources, making them unsuitable for direct deployment in real-time navigation, field robots, edge terminals, and other lightweight scenarios.

[0004] On the other hand, existing literature on frequency domain methods in texture enhancement mostly stays at the signal analysis level, lacking efficient fusion strategies with spatial domain positioning tasks. For example, Gabor wavelet transform can better simulate the sensitivity of the human eye to frequency and direction, and has good local texture response ability, but improper parameter selection and response aggregation strategy can introduce spectral redundancy and interference information. In terms of hardware, dynamic light control technology has been applied in industrial detection fields, but how to adjust the local lighting strategy in real time according to the environment light and improve the texture saliency in natural scenes is still a difficulty. SUMMARY

[0005] The application aims to provide a frequency domain-space domain combined texture enhancement positioning method, system and storage medium, which can enhance the texture response of a target region in multiple directions and multiple scales through a set of high-pass adaptive filter chains, and can cooperatively improve the density and stability of feature points in an image through a dynamic light compensation device based on a region reflection model.

[0006] The first aspect of the application provides a frequency domain-space domain combined texture enhancement positioning method, which comprises the following steps:

[0007] An original image is acquired, and frequency domain enhancement and space domain enhancement are performed to obtain a target image;

[0008] Dynamic light compensation is performed based on the target image to obtain an optimized image;

[0009] Feature point detection and fusion are performed based on the optimized image to obtain a fusion response map;

[0010] Feature point screening and positioning are performed based on the fusion response map to obtain a feature point set for positioning.

[0011] In the scheme, the method further comprises adjusting local illumination based on ambient light to perform dynamic light compensation on the original image to obtain a pretreated image.

[0012] In the scheme, the acquisition of the original image and the performance of frequency domain enhancement and space domain enhancement to obtain the target image specifically comprises:

[0013] Frequency domain enhancement is performed on the original image or the pretreated image based on a preset filter algorithm to obtain a frequency domain enhanced image, wherein the frequency domain enhanced image comprises directional texture features;

[0014] Space domain enhancement is performed on the frequency domain enhanced image based on a preset difference Gaussian algorithm to obtain a space domain enhanced image as the target image.

[0015] In the scheme, the dynamic light compensation based on the target image to obtain the optimized image specifically comprises:

[0016] The local brightness of the target image is calculated for partitioning;

[0017] The contrast is calculated based on the local brightness, and a key illumination region is acquired in combination with a set threshold value;

[0018] The array intensity of a light source is adjusted based on the key illumination region to obtain an optimized image after dynamic light compensation, wherein the light source comprises an LED.

[0019] In the scheme, the feature point detection and fusion based on the optimized image to obtain the fusion response map specifically comprises:

[0020] A preset FAST corner detection algorithm is run on the optimized image to obtain a response map;

[0021] The fused response map is obtained by weighting the frequency domain response and spatial domain response based on the response map.

[0022] In this solution, the step of filtering and locating feature points based on the fused response map to obtain a set of feature points for localization specifically includes:

[0023] The algorithm for nonmaximum suppression is combined with the fused response map to find local maxima in the local neighborhood, where the maxima satisfy the following conditions:

[0024]

[0025] Where, N r (x i ,y i ) indicates that (x i ,y i The neighborhood region centered at x and with radius r, (x i ,y i (x,y) represent the image coordinates, S final (x,y) represents the fusion response map;

[0026] The set of feature points used for localization is obtained by redundancy removal based on the found maxima.

[0027] A second aspect of the present invention also provides a frequency-spatial domain joint texture enhancement localization system, including a memory and a processor. The memory includes a frequency-spatial domain joint texture enhancement localization method program, which, when executed by the processor, performs the following steps:

[0028] The original image is acquired, and frequency domain enhancement and spatial domain enhancement are performed to obtain the target image;

[0029] Dynamic illumination compensation is performed on the target image to obtain an optimized image;

[0030] Based on the optimized image, feature point detection and fusion are performed to obtain a fused response map;

[0031] Feature points are filtered and located based on the fused response map to obtain a set of feature points for localization.

[0032] In this scheme, the method further includes performing dynamic illumination compensation on the original image based on adjusting local illumination according to ambient light to obtain a preprocessed image.

[0033] In this scheme, the process of acquiring the original image and performing frequency domain enhancement and spatial domain enhancement to obtain the target image specifically includes:

[0034] Frequency domain enhancement is performed on the original image or the preprocessed image using a preset filtering algorithm to obtain a frequency domain enhanced image, wherein the frequency domain enhanced image includes directional texture features;

[0035] Based on the frequency domain enhanced image, a preset differential Gaussian algorithm is used to perform spatial domain enhancement to obtain a spatial domain enhanced image as the target image.

[0036] In this solution, the step of performing dynamic illumination compensation based on the target image to obtain an optimized image specifically includes:

[0037] Calculate the local brightness of the target image to partition it;

[0038] The contrast is calculated based on the local brightness, and the key lighting area is obtained by combining the set threshold.

[0039] The array intensity of the light source is adjusted based on the key lighting area to obtain an optimized image after dynamic illumination compensation, wherein the light source includes LEDs.

[0040] In this scheme, the step of performing feature point detection and fusion based on the optimized image to obtain a fused response map specifically includes:

[0041] A preset FAST corner detection algorithm is run on the optimized image to obtain a response map;

[0042] The fused response map is obtained by weighting the frequency domain response and spatial domain response based on the response map.

[0043] In this solution, the step of filtering and locating feature points based on the fused response map to obtain a set of feature points for localization specifically includes:

[0044] The algorithm for nonmaximum suppression is combined with the fused response map to find local maxima in the local neighborhood, where the maxima satisfy the following conditions:

[0045]

[0046] Where, N r (x i ,y i ) indicates that (x i ,y i The neighborhood region centered at x and with radius r, (x i ,y i (x,y) represent the image coordinates, S final (x,y) represents the fusion response map;

[0047] The set of feature points used for localization is obtained by redundancy removal based on the found maxima.

[0048] A third aspect of the present invention provides a computer-readable storage medium comprising a machine program for a frequency-spatial domain joint texture enhancement localization method, wherein when the frequency-spatial domain joint texture enhancement localization method program is executed by a processor, it implements the steps of the frequency-spatial domain joint texture enhancement localization method as described in any of the preceding claims.

[0049] The present invention discloses a frequency-spatial domain joint texture enhancement and localization method, system, and storage medium, which have the following beneficial effects:

[0050] 1. Improved feature point performance: In low-texture scenes, the number of feature points increases by more than 5 times, with a more uniform distribution and more stable response.

[0051] 2. Advantages of technology integration: It overcomes the limitations of single-domain methods by synergistic enhancement in the frequency and spatial domains; and reduces shadow interference and improves texture saliency by dynamic illumination compensation.

[0052] 3. Enhanced application value: Suitable for lightweight real-time systems (such as agricultural robots and edge terminals), reducing reliance on labeled data and computing power; capable of providing more robust positioning solutions for industrial inspection, AR / VR and other fields.

[0053] 4. It has high industrialization potential and can be integrated into intelligent sorting equipment, industrial vision systems, etc., promoting the development of multispectral imaging and AI-optical fusion technology. Attached Figure Description

[0054] Figure 1 A flowchart of a frequency domain-spatial domain joint texture enhancement localization method according to the present invention is shown;

[0055] Figure 2 A block diagram of a frequency domain-spatial domain joint texture enhancement localization system of the present invention is shown. Detailed Implementation

[0056] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0057] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0058] The core of this invention lies in frequency domain filtering, spatial domain contrast enhancement, dynamic illumination compensation, and multi-scale feature point fusion. It enhances the directional texture response in images through Gabor wavelet transform, and improves the feature contrast in weak structure regions by combining it with an improved DOG (Difference-of-Gaussian) spatial enhancement operator. Furthermore, it employs an LED array partitioning control strategy to perform dynamic illumination compensation for different brightness areas, eliminating the texture weakening problem caused by changes in natural light. The goal is to extract a feature point set that is more than five times larger, more uniformly distributed, and more stable in low-texture images, thereby significantly improving the positioning accuracy and visual navigation robustness in agricultural and textile scenarios.

[0059] Figure 1 A flowchart of a frequency domain-spatial domain joint texture enhancement localization method according to this application is shown.

[0060] like Figure 1 As shown, this application discloses a frequency domain-spatial domain joint texture enhancement and localization method, including the following steps:

[0061] S102, acquire the original image, and perform frequency domain enhancement and spatial domain enhancement to obtain the target image;

[0062] S104, Perform dynamic illumination compensation based on the target image to obtain an optimized image;

[0063] S106, Based on the optimized image, feature point detection and fusion are performed to obtain a fused response map;

[0064] S108, based on the fused response map, feature points are filtered and located to obtain a set of feature points for localization.

[0065] It should be noted that the process of acquiring the original image and performing frequency domain enhancement and spatial domain enhancement to obtain the target image specifically includes:

[0066] Frequency domain enhancement is performed on the original image or the preprocessed image using a preset filtering algorithm to obtain a frequency domain enhanced image, wherein the frequency domain enhanced image includes directional texture features;

[0067] Based on the frequency domain enhanced image, a preset differential Gaussian algorithm is used to perform spatial domain enhancement to obtain a spatial domain enhanced image as the target image.

[0068] It should be noted that, in this embodiment, assuming the original image is I(x,y), the constructed two-dimensional Gabor filter kernel G... θ,λ (x,y), where the direction θ∈[0,π), and the wavelength λ controls the frequency response. Specifically, the Gabor kernel is defined as follows:

[0069]

[0070] x′=xcosθ+ysinθ, y′=-xsinθ+ycosθ;

[0071] Among them, G θ,λ (x,y) represents the 2D Gabor filter kernel, (x,y) represents the image coordinates, γ represents the aspect ratio, σ represents the standard deviation of the Gaussian envelope, φ represents the phase shift, λ represents the frequency, and θ represents the frequency in multiple directions. i Filtering is performed at ∈{0,π / 8,...,7π / 8} and multiple frequencies (i.e., different λ) to obtain a frequency domain enhanced image, which accordingly includes directional texture features.

[0072] Furthermore, in this embodiment, the frequency-enhanced image R after frequency-domain enhancement is... f Further spatial contrast enhancement is performed on (x,y) by introducing a multi-scale difference Gaussian algorithm for spatial domain enhancement, resulting in the spatial domain enhanced image R. s (x, y), where the multi-scale difference Gaussian algorithm is only used as an application in this embodiment, and the specific process will not be described in detail. Finally, the spatial domain enhanced image obtained is used as the target image. Accordingly, the target image can highlight edges and weak structures.

[0073] According to an embodiment of the present invention, the step of performing dynamic illumination compensation based on the target image to obtain an optimized image specifically includes:

[0074] Calculate the local brightness of the target image to partition it;

[0075] The contrast is calculated based on the local brightness, and the key lighting area is obtained by combining the set threshold.

[0076] The array intensity of the light source is adjusted based on the key lighting area to obtain an optimized image after dynamic illumination compensation, wherein the light source includes LEDs.

[0077] It should be noted that, in this embodiment, to enhance the brightness consistency of the textured area and avoid strong shadow interference, an LED array control system is designed to partition the image based on the local brightness gradient:

[0078]

[0079] Where L(x,y) represents the local brightness of a local region centered at coordinates (x,y), specifically reflecting the overall illumination level of that region, I(x+i,y+j) represents the pixel brightness value of the image at coordinates (x+i,y+j), and n represents the size of the local neighborhood window.

[0080] Furthermore, in this embodiment, the contrast ratio is then calculated. Where, σ L Let be the standard deviation of brightness, and ∈ be the numerical stability constant, based on a set threshold T. C Select high-contrast areas as key lighting areas, and adjust the corresponding LED light source intensity I accordingly. LED (x,y)~f(C(x,y)), where the light source intensity I LED It is negatively correlated with contrast ratio C.

[0081] According to an embodiment of the present invention, the method further includes performing dynamic illumination compensation on the original image based on adjusting local illumination according to ambient light to obtain a preprocessed image.

[0082] It should be noted that, in this embodiment, dynamic illumination compensation is not limited to target images that are spatially enhanced, but can also be applied to the original image.

[0083] According to an embodiment of the present invention, the step of performing feature point detection and fusion based on the optimized image to obtain a fused response map specifically includes:

[0084] A preset FAST corner detection algorithm is run on the optimized image to obtain a response map;

[0085] The fused response map is obtained by weighting the frequency domain response and spatial domain response based on the response map.

[0086] It should be noted that, in this embodiment, the target image is optimized after dynamic illumination compensation, and the FAST response function S(x,y) is introduced and the response is weighted to obtain the response map S. final (x,y)=αR s (x,y)+βS(x,y), where α and β are fusion weights used to control the contributions from the frequency and spatial domains, and S final (x,y) is the response graph, R s (x,y) represents the spatial domain augmented image, and S(x,y) represents the FAST response function.

[0087] According to an embodiment of the present invention, the step of filtering and locating feature points based on the fused response map to obtain a set of feature points for localization specifically includes:

[0088] The algorithm for nonmaximum suppression is combined with the fused response map to find local maxima in the local neighborhood, where the maxima satisfy the following conditions:

[0089]

[0090] Where, N r (x i ,y i) indicates that (x i ,y i The neighborhood region centered at x and with radius r, (x i y i (x, y) represent the image coordinates, S final (x,y) represents the fusion response map;

[0091] The set of feature points used for localization is obtained by redundancy removal based on the found maxima.

[0092] It should be noted that, in this embodiment, a non-maximum suppression algorithm is used for feature point selection and localization. Specifically, maxima are found within a local neighborhood based on the fused response map, and the maxima satisfy the following conditions: Accordingly, N r (x i ,y i ) indicates that (x i ,y i The neighborhood region centered at x and with radius r, (x i ,y i (x,y) represent the image coordinates, S final (x,y) represents the fused response map, and finally, the set of feature points used for localization is obtained by redundancy removal based on the found maxima.

[0093] Figure 2 A block diagram of a frequency domain-spatial domain joint texture enhancement localization system of the present invention is shown.

[0094] like Figure 2 As shown, this invention discloses a frequency-spatial domain joint texture enhancement localization system, including a memory and a processor. The memory includes a frequency-spatial domain joint texture enhancement localization method program. When the frequency-spatial domain joint texture enhancement localization method program is executed by the processor, it implements the following steps:

[0095] The original image is acquired, and frequency domain enhancement and spatial domain enhancement are performed to obtain the target image;

[0096] Dynamic illumination compensation is performed on the target image to obtain an optimized image;

[0097] Based on the optimized image, feature point detection and fusion are performed to obtain a fused response map;

[0098] Feature points are filtered and located based on the fused response map to obtain a set of feature points for localization.

[0099] It should be noted that the process of acquiring the original image and performing frequency domain enhancement and spatial domain enhancement to obtain the target image specifically includes:

[0100] Frequency domain enhancement is performed on the original image or the preprocessed image using a preset filtering algorithm to obtain a frequency domain enhanced image, wherein the frequency domain enhanced image includes directional texture features;

[0101] Based on the frequency domain enhanced image, a preset differential Gaussian algorithm is used to perform spatial domain enhancement to obtain a spatial domain enhanced image as the target image.

[0102] It should be noted that, in this embodiment, assuming the original image is I(x, y), the constructed two-dimensional Gabor filter kernel G... θ,λ (x,y), where the direction θ∈[0,π), and the wavelength λ controls the frequency response. Specifically, the Gabor kernel is defined as follows:

[0103]

[0104] x′=xcosθ+ysinθ, y′=-xsinθ+ycosθ;

[0105] Among them, G θ,λ (x,y) represents the 2D Gabor filter kernel, (x,y) represents the image coordinates, γ represents the aspect ratio, σ represents the standard deviation of the Gaussian envelope, φ represents the phase shift, λ represents the frequency, and θ represents the frequency in multiple directions. i Filtering is performed at ∈{0,π / 8,...,7π / 8} and multiple frequencies (i.e., different λ) to obtain a frequency domain enhanced image, which accordingly includes directional texture features.

[0106] Furthermore, in this embodiment, the frequency-enhanced image R after frequency-domain enhancement is... f Further spatial contrast enhancement is performed on (x,y) by introducing a multi-scale difference Gaussian algorithm for spatial domain enhancement, resulting in the spatial domain enhanced image R. s (x,y), where the multi-scale difference Gaussian algorithm is only used as an application in this embodiment, and the specific process will not be described in detail. Finally, the spatial domain enhanced image obtained is used as the target image. Accordingly, the target image can highlight edges and weak structures.

[0107] According to an embodiment of the present invention, the step of performing dynamic illumination compensation based on the target image to obtain an optimized image specifically includes:

[0108] Calculate the local brightness of the target image to partition it;

[0109] The contrast is calculated based on the local brightness, and the key lighting area is obtained by combining the set threshold.

[0110] The array intensity of the light source is adjusted based on the key lighting area to obtain an optimized image after dynamic illumination compensation, wherein the light source includes LEDs.

[0111] It should be noted that, in this embodiment, to enhance the brightness consistency of the textured area and avoid strong shadow interference, an LED array control system is designed to partition the image based on the local brightness gradient:

[0112]

[0113] Where L(x,y) represents the local brightness of a local region centered at coordinates (x,y), specifically reflecting the overall illumination level of that region, I(x+i,y+j) represents the pixel brightness value of the image at coordinates (x+i,y+j), and n represents the size of the local neighborhood window.

[0114] Furthermore, in this embodiment, the contrast ratio is then calculated. Where, σ L Let be the standard deviation of brightness, and ∈ be the numerical stability constant, based on a set threshold T. C Select high-contrast areas as key lighting areas, and adjust the corresponding LED light source intensity I accordingly. LED (x,y)~f(C(x,y)), where the light source intensity I LED It is negatively correlated with contrast ratio C.

[0115] According to an embodiment of the present invention, the method further includes performing dynamic illumination compensation on the original image based on adjusting local illumination according to ambient light to obtain a preprocessed image.

[0116] It should be noted that, in this embodiment, dynamic illumination compensation is not limited to target images that are spatially enhanced, but can also be applied to the original image.

[0117] According to an embodiment of the present invention, the step of performing feature point detection and fusion based on the optimized image to obtain a fused response map specifically includes:

[0118] A preset FAST corner detection algorithm is run on the optimized image to obtain a response map;

[0119] The fused response map is obtained by weighting the frequency domain response and spatial domain response based on the response map.

[0120] It should be noted that, in this embodiment, the target image is optimized after dynamic illumination compensation, and the FAST response function S(x,y) is introduced and the response is weighted to obtain the response map S. final (x,y)=αR s (x,y)+βS(x,y), where α and β are fusion weights used to control the contributions from the frequency and spatial domains, and Sfinal (x,y) is the response graph, R s (x,y) represents the spatial domain augmented image, and S(x,y) represents the FAST response function.

[0121] According to an embodiment of the present invention, the step of filtering and locating feature points based on the fused response map to obtain a set of feature points for localization specifically includes:

[0122] The algorithm for nonmaximum suppression is combined with the fused response map to find local maxima in the local neighborhood, where the maxima satisfy the following conditions:

[0123]

[0124] Where, N r (x i ,y i ) indicates that (x i ,y i The neighborhood region centered at x and with radius r, (x i ,y i (x,y) represent the image coordinates, S final (x,y) represents the fusion response map;

[0125] The set of feature points used for localization is obtained by redundancy removal based on the found maxima.

[0126] It should be noted that, in this embodiment, a non-maximum suppression algorithm is used for feature point selection and localization. Specifically, maxima are found within a local neighborhood based on the fused response map, and the maxima satisfy the following conditions: Accordingly, N r (x i ,y i ) indicates that (x i ,y i The neighborhood region centered at x and with radius r, (x i ,y i (x,y) represent the image coordinates, S final (x,y) represents the fused response map, and finally, the set of feature points used for localization is obtained by redundancy removal based on the found maxima.

[0127] A third aspect of the present invention provides a computer-readable storage medium comprising a frequency-spatial domain joint texture enhancement and localization method program, wherein when the frequency-spatial domain joint texture enhancement and localization method program is executed by a processor, it implements the steps of the frequency-spatial domain joint texture enhancement and localization method as described in any of the preceding claims.

[0128] This invention discloses a frequency-spatial domain joint texture enhancement localization method, system, and storage medium. Through frequency-spatial domain joint enhancement, dynamic illumination compensation, and multi-scale feature fusion, it effectively solves the feature extraction problem in low-texture scenes, significantly improves localization accuracy and system robustness, and has broad industrial application prospects.

[0129] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0130] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0131] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0132] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0133] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A frequency-spatial domain joint texture enhancement positioning method, characterized in that, The method comprises the following steps: obtaining an original image, and performing frequency domain enhancement and spatial domain enhancement to obtain a target image; performing dynamic light compensation based on the target image to obtain an optimized image; performing feature point detection and fusion based on the optimized image to obtain a fusion response map; performing feature point screening and positioning based on the fusion response map to obtain a feature point set for positioning.

2. The method of claim 1, wherein, The method further comprises adjusting local illumination according to ambient light to perform dynamic light compensation on the original image to obtain a preprocessed image.

3. The method of claim 2, wherein, The method of obtaining an original image, and performing frequency domain enhancement and spatial domain enhancement to obtain a target image comprises the following steps: performing frequency domain enhancement on the original image or the preprocessed image by using a preset filter algorithm to obtain a frequency domain enhanced image, wherein the frequency domain enhanced image comprises directional texture features; performing spatial domain enhancement on the frequency domain enhanced image by using a preset difference Gaussian algorithm to obtain a spatial domain enhanced image as the target image.

4. The method of claim 1, wherein, The method of performing dynamic light compensation based on the target image to obtain an optimized image comprises the following steps: calculating local brightness of the target image for partitioning; calculating contrast based on the local brightness, and obtaining a key illumination area in combination with a set threshold value; adjusting array intensity of a light source based on the key illumination area to obtain an optimized image after dynamic light compensation, wherein the light source comprises an LED.

5. The method of claim 4, wherein, The method of performing feature point detection and fusion based on the optimized image to obtain a fusion response map comprises the following steps: running a preset FAST corner detection on the optimized image to obtain a response map; performing weighted calculation of frequency domain response and spatial domain response based on the response map to obtain the fusion response map.

6. The method of claim 5, wherein, The method of performing feature point screening and positioning based on the fusion response map to obtain a feature point set for positioning comprises the following steps: finding a maximum value point in a local neighborhood based on a non-maximum suppression algorithm in combination with the fusion response map, wherein the maximum value point satisfies the following condition: where N r (x i ,y i ) represents a neighborhood region with (x i ,y i ) as the center and r as the radius, (x i ,y i ), (x,y) represent image coordinates, S final (x,y) represents a fusion response map; performing redundancy elimination based on the found maximum value point to obtain the feature point set for positioning.

7. A frequency-spatial domain joint texture enhancement positioning system, characterized in that, A device comprises a memory and a processor, and the memory comprises a frequency domain-spatial domain joint texture enhancement positioning method program, which is executed by the processor to implement the following steps: obtaining an original image, and performing frequency domain enhancement and spatial domain enhancement to obtain a target image; performing dynamic light compensation based on the target image to obtain an optimized image; performing feature point detection and fusion based on the optimized image to obtain a fusion response map; performing feature point screening and positioning based on the fusion response map to obtain a feature point set for positioning.

8. The frequency-space domain joint texture-enhanced positioning system of claim 7, wherein, The method of performing dynamic light compensation based on the target image to obtain an optimized image comprises the following steps: calculating local brightness of the target image for partitioning; calculating contrast based on the local brightness, and obtaining a key illumination area in combination with a set threshold value; adjusting array intensity of a light source based on the key illumination area to obtain an optimized image after dynamic light compensation, wherein the light source comprises an LED.

9. The frequency-space domain joint texture-enhanced positioning system of claim 8, wherein, The method of performing feature point detection and fusion based on the optimized image to obtain a fusion response map comprises the following steps: running a preset FAST corner point detection on the optimized image to obtain a response map; calculating a frequency domain response and a spatial domain response based on the response map to obtain the fusion response map.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises a frequency domain-spatial domain joint texture enhancement positioning method program, and the frequency domain-spatial domain joint texture enhancement positioning method program is executed by the processor to realize the steps of the frequency domain-spatial domain joint texture enhancement positioning method in any one of claims 1 to 6.

Images