Self-adaptive scene switching target identification method and system fusing visible light and infrared imaging

Abstract

The invention discloses an adaptive scene switching target identification method and system fusing visible light and infrared imaging, and relates to the technical field of data processing, and the method comprises the steps: obtaining a visible light image and an infrared image of a target region, obtaining the illumination intensity corresponding to the visible light image, and obtaining the temperature difference corresponding to the infrared image; acquiring a coupling coefficient based on the coupling model, the illumination intensity and the temperature difference quantity, and acquiring an illumination weight of the visible light image and an infrared weight of the infrared image according to the coupling coefficient; obtaining a target feature image according to the illumination weight, the infrared weight, the visible light image and the infrared image; and obtaining a target recognition result according to the target feature image. The method has the advantages of dynamic adaptability, multi-mode cooperation and key feature enhancement.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an adaptive scene switching target recognition method and system that integrates visible light and infrared imaging. Background Art

[0002] In complex and ever-changing real-world scenarios, target recognition technology faces multiple challenges, including drastic fluctuations in illumination, meteorological interference, and dynamic changes in thermal characteristics. Existing fusion methods have the following defects: First, fusion strategies based on fixed weights or a single threshold cannot adapt to the dynamic changes in light intensity and thermal radiation characteristics. Especially under critical environmental conditions such as the alternation of dawn and dusk, fog, haze, rain and snow, the effective information of visible light and infrared imaging exhibits a nonlinear complementary relationship. The existing methods lack fine perception of environmental parameters and a dynamic coupling mechanism, resulting in inaccurate fusion weight distribution. Furthermore, existing technologies lack the ability to adaptively process local pixel interference (such as headlight glare and heat source noise). In strong backlight or complex thermal background scenes, the noise components of visible light and infrared images are linearly superimposed, causing the signal-to-noise ratio of the fused target feature image to drop sharply. Most importantly, existing methods have not established a mechanism to associate modal weights with feature enhancement, resulting in the weakening or masking of key target features during the fusion process. This is essentially due to the lack of pixel-level dynamic perception and nonlinear coupling capabilities of light intensity and temperature differences, and the failure to map changes to feature enhancement mechanisms. Ultimately, the target recognition rate, positioning accuracy and environmental adaptability in complex scenes are difficult to meet actual application requirements. Summary of the Invention

[0003] In response to the deficiencies in the prior art, the present invention provides an adaptive scene switching target recognition method and system that integrates visible light and infrared imaging.

[0004] An adaptive scene switching target recognition method that integrates visible light and infrared imaging includes: acquiring a visible light image and an infrared image of a target area, acquiring the light intensity corresponding to the visible light image, and acquiring the temperature difference corresponding to the infrared image; acquiring a coupling coefficient based on a coupling model, the light intensity, and the temperature difference, and acquiring a light weight of the visible light image and an infrared weight of the infrared image according to the coupling coefficient; acquiring a target feature image according to the light weight, the infrared weight, the visible light image, and the infrared image; and acquiring a target recognition result according to the target feature image.

[0005] Optionally, obtaining the light intensity corresponding to the visible light image includes: collecting light information of the target area according to the photosensitive sensor array, and outputting a light intensity distribution map corresponding to the spatial position of the visible light image pixel according to the light information; and obtaining the light intensity corresponding to the visible light image according to the light intensity distribution map.

[0006] Optionally, obtaining the temperature difference corresponding to the infrared image includes: collecting the radiation temperature information of the target area according to the infrared sensor array, and outputting a temperature distribution map corresponding to the spatial position of the infrared image pixel based on the radiation temperature information; presetting a basic background temperature, and obtaining the temperature difference corresponding to the infrared image based on the basic background temperature and the temperature distribution map.

[0007] Optionally, obtaining the illumination weight of the visible light image and the infrared weight of the infrared image according to the coupling coefficient includes: directly mapping the coupling coefficient to the illumination weight of the visible light image; and generating the infrared weight of the infrared image according to the illumination weight and a preset weight constraint relationship.

[0008] Optionally, obtaining a target recognition result based on a target feature image includes: matching and comparing the target feature image with a pre-stored target feature database, and determining a plurality of candidate target categories based on matching similarity; screening out a target category that meets preset confidence conditions from the plurality of candidate target categories as a target recognition result, and outputting the target position and category information.

[0009] Optionally, the coupling model in obtaining the coupling coefficient based on the coupling model, light intensity, and temperature difference is expressed as: , ;in, Pixel The corresponding coupling coefficient, is the minimum coupling coefficient, is the pixel point in the visible light image The corresponding light intensity, is the critical light intensity threshold, is the pixel point in the infrared image The corresponding temperature difference, is the adjustment coefficient.

[0010] Optionally, the target feature image is obtained based on the illumination weight, infrared weight, visible light image and infrared image as follows: ;in, is the pixel point in the target feature image The corresponding eigenvalues, is the lighting weight, is the pixel point in the visible light image The corresponding gradient amplitude is is the infrared weight, is the pixel point in the infrared image The corresponding gradient magnitude.

[0011] An adaptive scene switching target recognition system that integrates visible light and infrared imaging is also provided. The system includes: a data acquisition module for acquiring visible light images and infrared images of the target area, and acquiring the light intensity corresponding to the visible light image, and acquiring the temperature difference corresponding to the infrared image; a data adaptive coupling module for acquiring a coupling coefficient based on a coupling model, light intensity and temperature difference, and acquiring the light weight of the visible light image and the infrared weight of the infrared image according to the coupling coefficient; a data feature processing module for acquiring a target feature image according to the light weight, infrared weight, visible light image and infrared image; and a comparison and recognition module for acquiring a target recognition result according to the target feature image.

[0012] Optionally, the data acquisition module is also used to: collect lighting information of the target area according to the photosensor array, and output a light intensity distribution map corresponding to the spatial position of the visible light image pixel according to the lighting information; and obtain the light intensity corresponding to the visible light image according to the light intensity distribution map.

[0013] Optionally, the data acquisition module is also used to: collect the radiation temperature information of the target area according to the infrared sensor array, and output a temperature distribution map corresponding to the spatial position of the infrared image pixel based on the radiation temperature information; preset the basic background temperature, and obtain the temperature difference corresponding to the infrared image based on the basic background temperature and the temperature distribution map.

[0014] The beneficial effects of the present invention are embodied in: In the adaptive scene switching target recognition method that integrates visible light and infrared imaging, first, based on the nonlinear coupling model of pixel-level light intensity and temperature difference, adaptive cross-modal weight allocation is achieved. In areas with drastic light fluctuations (such as strong backlight or the alternation of dawn and dusk), the visible light weight is dynamically suppressed by the temperature difference to retain the texture details of the highlight area. In low-light or meteorological interference scenes (such as haze, rain and snow), the infrared weight is enhanced based on the thermal radiation penetration characteristics. At the same time, the residual light information is used to correct the edge blur of the thermal image, significantly improving the signal-to-noise ratio of the fused image and the clarity of the target outline. Furthermore, the combination of the gradient-guided feature competition mechanism and the local interference suppression strategy solves the noise superposition problem. For visible light interference such as headlight glare, the dominant mode is dynamically switched through the stability analysis of the thermal radiation spectrum to eliminate pseudo edges. In complex thermal backgrounds, the spatial gradient of the temperature difference is used to constrain isolated thermal noise points and enhance the thermal conduction characteristics of the real target. Furthermore, the database joint analysis and confidence-driven decision-making mechanism further improve the recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0016] Figure 1 A schematic diagram of the steps of an adaptive scene switching target recognition method integrating visible light and infrared imaging according to an embodiment of the present invention; Figure 2 Schematic diagram of part of the steps S1 in the adaptive scene switching target recognition method integrating visible light and infrared imaging of the present invention; Figure 3 Schematic diagram of another part of the steps of S1 in the adaptive scene switching target recognition method integrating visible light and infrared imaging of the present invention; Figure 4 Schematic diagram of part of the steps S2 in the adaptive scene switching target recognition method integrating visible light and infrared imaging of the present invention; Figure 5 This is a schematic diagram of part of the steps in S4 of the adaptive scene switching target recognition method integrating visible light and infrared imaging of the present invention. DETAILED DESCRIPTION

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0018] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0019] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. In addition, the terms "first," "second," etc. are used only to distinguish the descriptions and are not to be understood as indicating or implying relative importance.

[0020] like Figure 1 As shown, a method for adaptive scene switching target recognition by integrating visible light and infrared imaging is provided, comprising: S1. Obtain a visible light image and an infrared image of the target area, obtain the light intensity corresponding to the visible light image, and obtain the temperature difference corresponding to the infrared image; S2. Obtaining a coupling coefficient based on the coupling model, the light intensity, and the temperature difference, and obtaining a light weight of the visible light image and an infrared weight of the infrared image according to the coupling coefficient; S3, obtaining a target feature image according to the illumination weight, infrared weight, visible light image and infrared image; S4. Obtain target recognition results based on the target feature image.

[0021] In this embodiment, it should be noted that in S1, multimodal sensors are used to collaboratively collect environmental information to achieve refined perception of light intensity and thermal radiation characteristics. For the acquisition of light intensity in visible light images, the photosensitive sensor array is directly used to obtain regional block sampling, and then combined with adaptive interpolation to infer the global light distribution. For example, in scenes with drastic changes in lighting, the brightness extreme points in the high dynamic range area of ​​the image (such as the center of a strong light source and the edge of a shadow) are preferentially extracted, and a light intensity mapping model of sparse sampling points is established in combination with local contrast analysis. Then, the full-image light intensity distribution is generated through gradient-guided interpolation. It can capture the gradual change of light intensity over a large range during the transition from dawn to dusk, and avoid the computing power redundancy of independent pixel-by-pixel acquisition, significantly reducing the computing load while ensuring accuracy.

[0022] Furthermore, for infrared images, a temperature field mapping model is constructed based on the physical properties of thermal radiation. An infrared sensor array scans the target area. Based on the temporal characteristics of infrared images, a sliding window is used to compile historical temperature data. This generates a real-time background temperature baseline (such as the persistently low temperature background of snow in a snowy scene). Inter-frame differencing is then used to extract localized temperature anomalies of stable heat source targets (such as the human body or vehicle engines). For example, under the interference of headlight glare, the thermal radiation spectrum characteristics are analyzed to distinguish between transient thermal noise (such as light reflections) and true heat sources. Temperature differences are calculated only for pixels that meet the thermal characteristics, avoiding the waste of point-by-point calculations across the entire image. Furthermore, temperature field sparse coding technology is used to compress redundant thermal data, reducing the amount of data to be processed in real time while preserving the edge gradients of key targets, achieving a balance between energy efficiency and accuracy.

[0023] In S2, a dynamic coupling model is used to adaptively allocate visible light and infrared modal weights, addressing the problem of fusion misalignment caused by nonlinear changes in environmental parameters. Specifically, the coupling model takes pixel-level light intensity and temperature difference as inputs and simulates the interaction between the two under critical conditions using a nonlinear function: when the light intensity is above a critical threshold, the model prioritizes enhancing the visible light weight to utilize its rich texture details. If the area also has a significant temperature difference (such as a high-temperature target), the visible light weight is suppressed by the temperature difference. Conversely, in low-light scenarios, the model activates the infrared weight based on the temperature difference to supplement the thermal features and imposes a smoothing constraint on the infrared weight through the gradient change of light intensity to avoid edge blurring of low-temperature differential targets. For example, in a hazy environment, by analyzing the coupling relationship between light attenuation and thermal radiation penetration characteristics, the infrared weight is automatically increased in foggy areas. At the same time, residual light information is used to correct the blurred edges of the thermal image, achieving cross-modal feature compensation.

[0024] Furthermore, the coupled model incorporates a local interference suppression mechanism, distinguishing true targets from transient noise through temporal stability analysis of temperature differences. When headlight glare is detected, resulting in a localized overbrightness of visible light, the model dynamically reduces the visible light weight and activates the infrared mode's noise suppression coefficient, taking into account the abrupt temperature difference characteristics in that area (such as the spectral difference between transient thermal reflections and stable heat sources). This prevents glare noise from propagating into the fused image. Furthermore, in complex thermal backgrounds, the model applies edge-guided constraints to the infrared weight based on the spatial distribution of temperature differences (such as the transition gradient between the heat source and the background), suppressing thermal diffusion noise while enhancing the target outline. For example, in snowy scenes, the model uses the spatial distribution of the temperature difference between the low-temperature background of the snow and the heat source of the human body to reduce the visible light weight in reflective areas of the snow while enhancing the infrared gradient weight at the edges of the human outline, preventing the target in the fused image from being overwhelmed by snow reflections or low-temperature noise.

[0025] In S3, a gradient-guided cross-modal feature competition mechanism is used to achieve complementarity in key target features. Based on the spatial gradient distribution of visible and infrared images, the two types of gradients are nonlinearly modulated using dynamic weights. When the combined effect of the visible light gradient amplitude and its weight at a pixel is significantly higher than that of the infrared modality, visible light edge features are amplified and infrared noise is suppressed. Conversely, when the infrared gradient exhibits a stronger target contour response under low illumination or complex thermal backgrounds, infrared features are prioritized and visible light interference components are weakened. For example, in a dense foggy environment, the road marking gradient in the visible light image is attenuated due to scattering, while the thermal radiation gradient of the vehicle engine in the infrared image remains clear. At this time, the contribution of the infrared gradient to the fusion feature is automatically enhanced through weighted comparison, achieving cross-modal feature complementarity.

[0026] Furthermore, a gradient competition threshold adaptive noise suppression strategy is introduced to address feature confusion caused by local interference. When an abnormal increase in gradient amplitude (e.g., headlight glare) is detected in a visible light highlight region and the corresponding infrared temperature difference fluctuates abnormally, the fusion ratio of the visible light gradient in that region is reduced through dynamic weighting, and gradient features with verified thermal stability in the infrared image are used instead. For regions with thermal background clutter, the gradient response of isolated thermal noise points is suppressed through spatial continuity constraints on the infrared weights. For example, in a scene with strong snow reflections, the false visible light edges caused by snow reflections are filtered out due to dynamic weight reduction, while the continuous temperature gradient of human targets in the infrared image is enhanced. The resulting fused image eliminates reflection artifacts while fully preserving the true edge of the interface between the human thermal signature and the snow.

[0027] In S4, based on the fused target feature image, multi-level features are first extracted to construct a cross-modal feature descriptor, which is then matched against a pre-stored database for multi-dimensional similarity. For example, in a haze scene, for a vehicle target obscured by fog, the vehicle type and location of a truck or car can be effectively identified by comparing the vehicle features retained in the fused features with the features of different models in the database. A spatiotemporal continuity verification mechanism is also introduced to correlate the target's motion trajectory and thermal signature changes in adjacent frames, eliminating mismatches caused by transient noise.

[0028] Furthermore, a confidence-driven dynamic decision-making mechanism is used to optimize recognition accuracy. For the set of candidate targets generated by matching, a multi-dimensional confidence assessment model is constructed by integrating local feature matching, environmental interference suppression coefficient, and target motion consistency. In strong backlight scenarios, for targets whose visible light features are missing due to glare, the infrared modal confidence weight is improved through thermal radiation stability analysis, and trajectory prediction and completion are performed in combination with historical recognition results. For targets with low temperature differences in the snow, the recognition threshold is dynamically adjusted through the combined calculation of edge gradient continuity detection and background thermal noise suppression coefficient. For example, when identifying a stationary human body in the snow, the thermal conduction gradient characteristics of the contact surface between the target and the snow are analyzed to distinguish between a real human body and low-temperature interference objects such as rocks. At the same time, historical motion data is used to eliminate false triggers caused by snow sliding, ultimately outputting a recognition result that is both real-time and reliable.

[0029] In summary, the adaptive scene switching target recognition method that integrates visible light and infrared imaging first implements adaptive cross-modal weight allocation based on a nonlinear coupling model of pixel-level light intensity and temperature difference. In areas with drastic light fluctuations (such as strong backlight or the transition between dawn and dusk), the visible light weight is dynamically suppressed by the temperature difference, preserving the texture details of the highlight area. In low-light or meteorological interference scenes (such as haze, rain and snow), the infrared weight is enhanced based on the thermal radiation penetration characteristics. At the same time, the residual light information is used to correct the edge blur of the thermal image, significantly improving the signal-to-noise ratio of the fused image and the clarity of the target outline. Furthermore, the combination of the gradient-guided feature competition mechanism and the local interference suppression strategy solves the noise superposition problem. For visible light interference such as headlight glare, the dominant mode is dynamically switched through the stability analysis of the thermal radiation spectrum to eliminate pseudo edges. In complex thermal backgrounds, the spatial gradient of the temperature difference is used to constrain isolated thermal noise points and enhance the thermal conduction characteristics of the real target. Furthermore, the joint database analysis and the confidence-driven decision-making mechanism further improve the recognition accuracy.

[0030] like Figure 2 As shown, in one embodiment, obtaining the light intensity corresponding to the visible light image in S1 includes: S11, collecting illumination information of the target area according to the light sensor array, and outputting an illumination intensity distribution map corresponding to the spatial position of the visible light image pixel according to the illumination information; S12. Obtain the light intensity corresponding to the visible light image according to the light intensity distribution map.

[0031] In this embodiment, it should be noted that in S11, the light intensity of the target area is precisely sensed using a photosensor array. To address the issue of dramatic light fluctuations in complex scenes (such as global gradients during the transition from dawn to dusk or localized high dynamic range in strong backlight), the sensor employs a strategy that combines sparse sampling with prioritized capture of key areas. First, the sampling density is dynamically divided based on scene characteristics. Dense sampling points are deployed in areas of sudden changes in illumination (such as the center of headlight glare or the edge of building shadows). Local contrast extremes are used to detect key nodes of brightness changes. For areas of moderate illumination, sparse sampling is used to avoid redundant data collection. For example, in a hazy environment, the sensor prioritizes capturing the extreme points of residual light intensity that penetrates the fog. Using an atmospheric scattering model, the sensor infers the effect of fog thickness on light attenuation, thereby constructing a sparse sampling network covering the entire area. This process is processed in real time by edge computing devices, completing preliminary light intensity feature extraction at the sensor end, reducing data transmission pressure.

[0032] In S12 , the illumination intensity distribution map is quickly matched with the pixel level of the visible light image, so that each pixel of the visible light image is directly associated with the illumination intensity value of the corresponding position.

[0033] like Figure 3 As shown, in one embodiment, obtaining the temperature difference corresponding to the infrared image in S1 includes: S13, collecting radiation temperature information of the target area according to the infrared sensor array, and outputting a temperature distribution map corresponding to the spatial position of the infrared image pixel according to the radiation temperature information; S14. Preset a basic background temperature, and obtain a temperature difference corresponding to the infrared image according to the basic background temperature and the temperature distribution map.

[0034] In this embodiment, it should be noted that in S13, the infrared sensor array achieves precise capture and spatial mapping of thermal radiation signatures. To address complex thermal backgrounds (such as interference from snow reflections or noise from moving heat sources), the sensor array employs an adaptive scanning strategy: The sampling frequency is increased in areas with densely populated high-temperature targets (e.g., vehicle engine clusters) to capture subtle temperature fluctuations, while the sampling density is reduced in areas with low-temperature backgrounds (e.g., snow or water) to conserve energy. For example, in headlight glare scenarios, by analyzing the spectral characteristics of the thermal radiation signal (e.g., the transient nature of reflected heat pulses and the persistence of engine heat sources), the pseudo-temperature difference signals caused by light reflections are dynamically filtered out, retaining only the true target temperature data that conforms to the laws of heat conduction. Simultaneously, the sensor integrates the texture features of the infrared image and performs edge-guided spatial correction on the temperature distribution map, eliminating temperature field distortion caused by sensor parallax and ensuring pixel-level alignment of the thermal radiation data with the infrared image.

[0035] In S14, dynamic background temperature modeling and abnormal temperature difference extraction techniques are used to accurately separate target thermal features. First, based on the spatiotemporal continuity of the thermal scene, a sliding window is used to statistically analyze the temperature distribution patterns of historical frames and construct an adaptive unified background temperature baseline. Then, a temperature difference distribution map is obtained based on the temperature distribution map and the unified background temperature baseline. This temperature difference distribution map is quickly matched to the infrared image at the pixel level, directly associating each pixel in the infrared image with the temperature difference at the corresponding location.

[0036] like Figure 4 As shown, in one embodiment, obtaining the illumination weight of the visible light image and the infrared weight of the infrared image according to the coupling coefficient in S2 includes: S21. Directly mapping the coupling coefficient to an illumination weight of a visible light image; S22. Generate an infrared weight of the infrared image according to the illumination weight and a preset weight constraint relationship.

[0037] In this embodiment, it should be noted that in S21, the dynamic response of the modal weight is achieved through direct mapping of the coupling coefficient and the illumination weight. For example, the coupling coefficient is directly used as the illumination weight (coupling coefficient = illumination weight). This mapping mechanism retains the coupling model's ability to analyze the nonlinear relationship between illumination intensity and temperature difference, ensuring that the weight distribution can reflect the microscopic changes in environmental parameters in real time. For example, in a scene of alternating dawn and dusk, when the illumination intensity in a certain area is near the critical threshold and there is a high-temperature target (such as vehicle exhaust), the coupling coefficient will dynamically fine-tune the visible light weight according to the temperature difference - both utilizing the residual texture of visible light under critical illumination and avoiding overexposure caused by reflection of metal surfaces through temperature difference suppression. This direct mapping method avoids the information attenuation caused by secondary conversion in existing methods, so that the weight adjustment in high dynamic range scenes has a sub-pixel response speed.

[0038] In S22, global cross-modal information balance is achieved through weight complementarity constraints. For example, the infrared weight is calculated by subtracting the illumination weight from 1 and adding the minimum base weight (1-illumination weight + minimum base weight = infrared weight). This ensures a strict inverse correlation between the infrared weight and the visible light weight, forcing the fusion process to retain the effective components of bimodal information at any pixel. For example, in a dense foggy environment, when the fog causes an overall decrease in the visible light weight, the infrared weight automatically rises to a dominant position. However, the residual illumination gradient constraint in the coupling model (such as the outline of a directional light source penetrating the fog) ensures that the infrared weight retains the ability to correct for visible light edges during the increase. This constraint mechanism not only prevents the complete failure of a single modality but also naturally suppresses noise superposition through the weight competition relationship. When the weight of a certain area of ​​visible light is abnormally increased due to glare, the corresponding decrease in the infrared weight prevents the noise from being transmitted to the fusion result, while compensating for it by leveraging the stable thermal characteristics of the infrared modality in that area.

[0039] like Figure 5 As shown, in one embodiment, obtaining a target recognition result according to the target feature image in S4 includes: S41, matching and comparing the target feature image with a pre-stored target feature database, and determining a plurality of candidate target categories based on matching similarities; S42: Filter out target categories that meet preset confidence conditions from multiple candidate target categories as target recognition results, and output target location and category information.

[0040] In this embodiment, it should be noted that in S41, precise matching of target features is achieved through a cross-modal feature association model. Based on multi-level features extracted from the fused image (such as edge topology and thermal radiation distribution patterns in the prior art), a database-compatible composite descriptor is constructed. A prior art adaptive similarity measurement strategy is employed to address environmental interference. In haze scenes, the infrared modality's penetrating features are weighted to compensate for texture detail lost due to visible light scattering. Furthermore, a prior art local occlusion analysis mechanism is introduced to dynamically partition the feature matching area to avoid global similarity distortion caused by fog occlusion. For example, when identifying vehicles partially covered by snow, the heat conduction contours of the unobstructed area are prioritized, along with the vehicle's geometric features remaining in visible light. Motion trajectory prediction is then used to complete the feature vectors of the snow-covered areas, effectively improving the matching accuracy of partially visible targets. A spatiotemporal continuity verification mechanism further filters transient mismatches. By analyzing the target's motion inertia and thermal signature variations (such as the slow-changing engine temperature) across multiple frames, false candidate classes caused by falling snow or flickering light and shadows are eliminated.

[0041] In S41, a conventional confidence decision model is employed to address target classification ambiguity in complex environments. This model integrates feature matching, environmental interference suppression coefficient, and motion consistency metrics, dynamically adjusting the weights of each dimension. In strong backlight scenarios, the thermal radiation stability metric is weighted higher, verifying matching reliability by analyzing the thermal inertia characteristics of high-temperature targets (e.g., the temperature change rate of a vehicle's exhaust pipe). In snowy, low-temperature gradient scenarios, the combined effect of edge gradient continuity and background noise suppression coefficient is enhanced, distinguishing true targets from static thermal artifacts by analyzing the thermal diffusion characteristics of the target contour (e.g., the thermal conduction gradient between a person and clothing). For example, when identifying stationary targets in the snow, the model combines thermal conduction pattern verification (e.g., the temperature attenuation difference between a person and a rock) with motion history data (e.g., the temporal plausibility of the target's appearance). This multi-dimensional confidence assessment allows for accurate target classification, even when the target's temperature difference approaches the ambient noise level, and outputs a probabilistically weighted recognition result and spatial positioning information.

[0042] In one embodiment, the coupling coefficient obtained in S2 based on the coupling model, light intensity and temperature difference is expressed as: , ;in, Pixel The corresponding coupling coefficient, is the minimum coupling coefficient, is the pixel point in the visible light image The corresponding light intensity, is the critical light intensity threshold (light intensity at dusk, such as 50 lux), is the pixel point in the infrared image The corresponding temperature difference, is the adjustment coefficient.

[0043] In this embodiment, it should be noted that, in the entire expression, It is used to distinguish whether the light intensity exceeds the critical threshold and solve the weight mutation problem of critical light intensity scenes such as the transition from dawn to dusk. hour, =1, allows the temperature difference to adjust the visible light weight, enters the visible light dominant mode, and retains texture details; when hour, , forced to enter low light mode, trigger infrared enhancement, and force visible light weight to be no less than , since the exponential term is always non-negative, , so through the outer layer Constraints, ultimately , ensuring that visible light still retains basic contributions in extremely low illumination (such as metal reflections in moonlight).

[0044] Further, is the fusion correction term; middle, It is used to normalize the offset of the light intensity relative to the critical value to eliminate the dimension effect, and the absolute value is used to eliminate sign interference to ensure that the change in light intensity is always positively involved in the calculation. Used to regulate the temperature difference, is the adjustment coefficient, used to ensure and sensitivity (for example, k = 0.05 increases the impact of double-digit temperature differences).

[0045] Furthermore, when ,(like =1000lux, =50lux, =10 degrees Celsius, Take 0.05), =19, ,at this time Approaching 0.76, the coupling coefficient represents the extent to which visible light exceeds the infrared influence. (like =50lux, =50lux) 0 ， at this time equal , the visible light weight is relatively low, and the temperature difference does not affect the coupling coefficient, thus avoiding texture loss. Slightly larger than ,at this time 1, 1, Can easily influence The value of =10 degrees Celsius, Take 0.05, then ,at this time Approaching 0.42, the coupling coefficient represents the extent to which infrared exceeds the influence of visible light.

[0046] In one embodiment, the target feature image obtained in S3 according to the illumination weight, infrared weight, visible light image and infrared image is expressed as: ;in, is the pixel point in the target feature image The corresponding eigenvalues, is the lighting weight, is the pixel point in the visible light image The corresponding gradient amplitude is is the infrared weight, is the pixel point in the infrared image The corresponding gradient magnitude.

[0047] In this embodiment, it should be noted that in the entire expression, the gradient amplitude and They are all obtained through the Sobel operator in the existing technology. The Sobel operator is a classic image edge detection algorithm that identifies edges by calculating the spatial gradient of the image grayscale value. Its core idea is to enhance the contrast of the object boundary in the image through directional differentiation. It is widely used in target recognition, image fusion and other fields.

[0048] Further, and The product of gradient amplitude and weight strengthens the effective features of the dominant mode. The higher the weight, the greater the gradient contribution of the corresponding mode. This solves the problem of feature weakening caused by fixed weights in existing methods. For example: in strong light areas ( = 0.9), even if the visible light gradient is small (e.g. =10), the product is 0.9 =9 can still retain details; in low temperature difference areas ( = 0.2), if the infrared gradient is significant (e.g. =30), 0.2*30=6 also cannot dominate the fusion result.

[0049] Further, and are all correction items, reflecting the competition mechanism; When , in the visible light gradient amplitude, =1, the visible light coefficient is 1.5, and in the infrared gradient amplitude, the infrared coefficient is 0.5; when When , the sign is reversed, the infrared term is enhanced by 1.5 times, and the visible light term is weakened to 0.5 times. This suppresses the superposition of dual-modal noise. For example, the visible light weight in the headlight glare area is artificially high due to the glare ( =0.7), but its Abnormal, showing noise; infrared weight =0.3, but the gradient is stable, reflecting the real edge; the noise term is filtered out by the max function, and the infrared is finally dominant.

[0050] Furthermore, the maximum function The choice of is equivalent to retaining the most significant characteristic mode in the current environment, avoiding the blurring of details caused by weighted averaging. It solves the problem of blurred edges of low temperature target, for example, human body in snow: low visible light weight ( = 0.3), the gradient is disturbed by snow reflection ( =15 pseudo edge); infrared weight is high ( = 0.7), the gradient is stable ( = 10 true contour); the infrared term is equal to 10.5, and the visible light term is equal to 2.25; as a result, the infrared features are retained and the visible light noise is excluded.

[0051] An adaptive scene switching target recognition system that integrates visible light and infrared imaging is also provided. The system includes: a data acquisition module for acquiring visible light images and infrared images of the target area, and acquiring the light intensity corresponding to the visible light image (each pixel corresponds to a light intensity), and acquiring the temperature difference corresponding to the infrared image (each pixel corresponds to a temperature difference); a data adaptive coupling module for acquiring a coupling coefficient based on a coupling model, light intensity and temperature difference, and acquiring the light weight of the visible light image and the infrared weight of the infrared image according to the coupling coefficient; a data feature processing module for acquiring a target feature image based on the light weight, infrared weight, visible light image and infrared image; and a comparison and recognition module for acquiring a target recognition result based on the target feature image.

[0052] In one embodiment, the data acquisition module is further used to: collect lighting information of the target area according to the photosensor array, and output a light intensity distribution map corresponding to the spatial position of the visible light image pixel according to the lighting information; and obtain the light intensity corresponding to the visible light image according to the light intensity distribution map.

[0053] In one embodiment, the data acquisition module is also used to: collect the radiation temperature information of the target area according to the infrared sensor array, and output a temperature distribution map corresponding to the spatial position of the infrared image pixel based on the radiation temperature information; preset the basic background temperature, and obtain the temperature difference corresponding to the infrared image based on the basic background temperature and the temperature distribution map.

[0054] In this embodiment, it should be noted that, regarding the above-mentioned adaptive scene switching target recognition system that integrates visible light and infrared imaging, the specific method of performing the operation has been described in detail in the implementation of the adaptive scene switching target recognition method that integrates visible light and infrared imaging, and will not be elaborated here.

[0055] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings. However, the present disclosure is not limited to the specific details of the above embodiments. Within the technical concept of the present disclosure, various simple modifications can be made to the technical solutions of the present disclosure, and these simple modifications all fall within the scope of protection of the present disclosure.

[0056] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.

[0057] In addition, the various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. An adaptive scene switching target recognition method integrating visible light and infrared imaging, characterized in that: include: Obtain visible light images and infrared images of the target area, obtain the light intensity corresponding to the visible light image, and obtain the temperature difference corresponding to the infrared image; A coupling coefficient is obtained based on the coupling model, the light intensity and the temperature difference, and the light weight of the visible light image and the infrared weight of the infrared image are obtained according to the coupling coefficient; Acquire a target feature image according to the illumination weight, the infrared weight, the visible light image and the infrared image; Obtain target recognition results based on target feature images.

2. The adaptive scene switching target recognition method integrating visible light and infrared imaging according to claim 1 is characterized in that: Acquiring the light intensity corresponding to the visible light image includes: Collecting illumination information of the target area according to the light sensor array, and outputting a light intensity distribution map corresponding to the spatial position of the visible light image pixel according to the illumination information; Obtain the light intensity corresponding to the visible light image according to the light intensity distribution map.

3. The adaptive scene switching target recognition method integrating visible light and infrared imaging according to claim 1 is characterized in that: The temperature difference corresponding to the infrared image is obtained as follows: Collecting radiation temperature information of the target area according to the infrared sensor array, and outputting a temperature distribution map corresponding to the spatial position of the infrared image pixel according to the radiation temperature information; The basic background temperature is preset, and the temperature difference corresponding to the infrared image is obtained according to the basic background temperature and the temperature distribution map.

4. The adaptive scene switching target recognition method integrating visible light and infrared imaging according to claim 1 is characterized in that: The step of obtaining the illumination weight of the visible light image and the infrared weight of the infrared image according to the coupling coefficient includes: Directly mapping the coupling coefficient to an illumination weight of a visible light image; The infrared weight of the infrared image is generated according to the illumination weight and the preset weight constraint relationship.

5. The adaptive scene switching target recognition method integrating visible light and infrared imaging according to claim 1 is characterized in that: The obtaining of the target recognition result according to the target feature image comprises: Matching and comparing the target feature image with a pre-stored target feature database, and determining a plurality of candidate target categories based on matching similarities; Target categories that meet the preset confidence conditions are selected from multiple candidate target categories as target recognition results, and the target location and category information are output.

6. The adaptive scene switching target recognition method integrating visible light and infrared imaging according to claim 1 is characterized in that: The coupling model in obtaining the coupling coefficient based on the coupling model, light intensity and temperature difference is expressed as: , ;in, Pixel The corresponding coupling coefficient, is the minimum coupling coefficient, is the pixel point in the visible light image The corresponding light intensity, is the critical light intensity threshold, is the pixel point in the infrared image The corresponding temperature difference, is the adjustment coefficient.

7. The adaptive scene switching target recognition method integrating visible light and infrared imaging according to claim 1 is characterized in that: The target feature image obtained according to the illumination weight, infrared weight, visible light image and infrared image is expressed as: ;in, is the pixel point in the target feature image The corresponding eigenvalues, is the lighting weight, is the pixel point in the visible light image The corresponding gradient amplitude is is the infrared weight, is the pixel point in the infrared image The corresponding gradient magnitude.

8. An adaptive scene switching target recognition system integrating visible light and infrared imaging, characterized in that: The system comprises: A data acquisition module is used to acquire visible light images and infrared images of the target area, obtain the light intensity corresponding to the visible light image, and obtain the temperature difference corresponding to the infrared image; A data adaptive coupling module is used to obtain a coupling coefficient based on a coupling model, light intensity, and temperature difference, and to obtain a light weight of a visible light image and an infrared weight of an infrared image according to the coupling coefficient; A data feature processing module is used to obtain a target feature image based on the illumination weight, infrared weight, visible light image and infrared image; The comparison and recognition module obtains the target recognition result based on the target feature image.

9. The adaptive scene switching target recognition system integrating visible light and infrared imaging according to claim 8, characterized in that: The data acquisition module is also used for: Collecting illumination information of the target area according to the light sensor array, and outputting a light intensity distribution map corresponding to the spatial position of the visible light image pixel according to the illumination information; Obtain the light intensity corresponding to the visible light image according to the light intensity distribution map.

10. The adaptive scene switching target recognition system integrating visible light and infrared imaging according to claim 8, characterized in that: The data acquisition module is also used for: Collecting radiation temperature information of the target area according to the infrared sensor array, and outputting a temperature distribution map corresponding to the spatial position of the infrared image pixel according to the radiation temperature information; The basic background temperature is preset, and the temperature difference corresponding to the infrared image is obtained according to the basic background temperature and the temperature distribution map.

Images