Night imaging enhancement method and system for hunting camera

Abstract

The invention discloses a night imaging enhancement method and system for a hunting camera, and relates to the technical field of image processing. The method comprises the following steps: capturing multispectral sensing image information through a target hunting camera at night; performing weighted fusion processing on the multispectral sensing image information according to the spectral band characteristic information to obtain night fusion image information; constructing a hunting entity label library, and performing region division on the night fusion image information to obtain N night image regions; mining to obtain a hunting scene source image library and a hunting scene night image library, performing associated feature extraction and imaging enhancement analysis, and constructing hunting night imaging enhancement multiple channels; and mapping the N night image areas to hunting night imaging enhancement multiple channels for parallel image enhancement, and generating hunting night enhanced image information. The technical problem of poor night imaging quality of the hunting camera in the prior art is solved, and the technical effect of improving the night imaging quality of the hunting camera is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a nighttime imaging enhancement method and system for a hunting camera. Background Art

[0002] Hunting cameras can automatically capture images or video information of the target area without human supervision, providing users with intuitive on-site information. However, the existing hunting cameras have poor imaging quality in low-light environments, especially at night. Due to the lack of visible light and near-infrared light, the image clarity is poor and the details are not fully presented, which affects the accuracy of wildlife monitoring and capture. Although the infrared imaging technology used in the prior art performs well in low-light environments, it still has problems such as high image noise and loss of details. Therefore, there is an urgent need for a new imaging technology that can solve the technical bottleneck of poor image quality in low-light environments by optimizing image capture and processing technology. Summary of the invention

[0003] The present application provides a night imaging enhancement method and system for a hunting camera, which solves the technical problem of poor night imaging quality of hunting cameras in the prior art.

[0004] In a first aspect of the present application, a night imaging enhancement method for a hunting camera is provided, the method comprising: A target hunting camera is obtained, wherein the target hunting camera has a built-in multispectral sensor, and multispectral perception image information is obtained by capturing the target hunting camera at night, wherein the multispectral perception image information includes visible light, near-infrared light and thermal infrared light band images; weighted fusion processing is performed on the multispectral perception image information according to spectral band characteristic information to obtain night fusion image information; a hunting entity label library is constructed, and the night fusion image information is divided into regions based on the hunting entity label library to obtain N night image regions; according to the hunting entity label library, a hunting scene source image library and a hunting scene night image library are mined and obtained, and associated features are extracted and imaging enhancement analysis is performed on the hunting scene source image library and the hunting scene night image library, and a hunting night imaging enhancement multi-channel is constructed; the N night image regions are mapped to the hunting night imaging enhancement multi-channel for parallel image enhancement to generate hunting night enhanced image information.

[0005] In a second aspect of the present application, a night imaging enhancement system for a hunting camera is provided, the system comprising: An image acquisition module for obtaining a target hunting camera, the target hunting camera being built with a multispectral sensor, and obtaining multispectral perception image information through night capture by the target hunting camera, the multispectral perception image information including visible light, near-infrared light, and thermal infrared light band images; a fusion module for performing weighted fusion processing on the multispectral perception image information according to spectral band characteristic information to obtain night fusion image information; a region division module for constructing a hunting entity tag library and dividing the night fusion image information based on the hunting entity tag library to obtain N night image regions; an analysis module for mining and obtaining a hunting scene source image library and a hunting scene night image library according to the hunting entity tag library, and performing correlation feature extraction and imaging enhancement analysis on the hunting scene source image library and the hunting scene night image library to build a multi-channel for hunting night imaging enhancement; an image enhancement module for mapping the N night image regions to the multi-channel for hunting night imaging enhancement for parallel image enhancement to generate hunting night enhanced image information.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, obtain a target hunting camera. The target hunting camera is built with a multispectral sensor, and obtain multispectral perception image information through night capture by the target hunting camera. The multispectral perception image information includes visible light, near-infrared light, and thermal infrared light band images. Next, perform weighted fusion processing on the multispectral perception image information according to spectral band characteristic information to obtain night fusion image information. Further, construct a hunting entity tag library and divide the night fusion image information based on the hunting entity tag library to obtain N night image regions. Then, according to the hunting entity tag library, mine and obtain a hunting scene source image library and a hunting scene night image library, and perform correlation feature extraction and imaging enhancement analysis on the hunting scene source image library and the hunting scene night image library to build a multi-channel for hunting night imaging enhancement. Finally, map the N night image regions to the multi-channel for hunting night imaging enhancement for parallel image enhancement to generate hunting night enhanced image information. This solves the technical problem of poor night imaging quality of hunting cameras in the prior art and achieves the technical effect of improving the night imaging quality of hunting cameras. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0008] Figure 1Schematic flow chart of the night imaging enhancement method for a hunting camera provided by an embodiment of the present application; Figure 2 Schematic structural diagram of the night imaging enhancement system for a hunting camera provided by an embodiment of the present application.

[0009] Explanation of reference numerals: Image acquisition module 11, fusion module 12, region division module 13, analysis module 14, image enhancement module 15. Detailed implementation manners

[0010] The present application provides a night imaging enhancement method and system for a hunting camera, and solves the technical problem of poor night imaging quality of hunting cameras in the prior art.

[0011] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0012] It should be noted that the terms "include" and "have" are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0013] Embodiment 1, as Figure 1 shown, the present application provides a night imaging enhancement method for a hunting camera, wherein the method includes: Obtain a target hunting camera, the target hunting camera is built with a multispectral sensor, and multispectral perception image information is captured at night by the target hunting camera, and the multispectral perception image information includes visible light, near-infrared light and thermal infrared light band images.

[0014] In the embodiments of the present application, the target hunting camera is built with a multispectral sensor. When working at night, the multispectral sensor can collect and obtain multispectral perception image information including visible light, near-infrared light, and thermal infrared light bands. Specifically, the visible light band image provides image information under traditional visible light shooting conditions and is suitable for target recognition in bright environments; the near-infrared light band image can provide additional image information in low-light environments, enhancing the visibility of night images; the thermal infrared light band image can effectively identify heat sources existing in the environment by detecting temperature differences, especially providing important support for target recognition of wild animals and the like in extremely poor lighting conditions. Through the perception of these different bands, the target hunting camera can perform high-quality image acquisition under various complex lighting conditions, thereby providing multi-dimensional image information for subsequent image processing and analysis.

[0015] Furthermore, the multispectral perception image information obtained by the target hunting camera through night capture includes: An ambient light sensor, a ranging sensor, an angle sensor, and a fill light control module are arranged in the target hunting camera. The ambient light intensity is monitored and obtained in real time through the ambient light sensor, and the position information of the hunting target is obtained through the ranging sensor and the angle sensor; based on the fill light control module, a dataset of object spectral reflectivity is called, and the band reflectivity information of the hunting target is obtained through matching the dataset of object spectral reflectivity; based on the ambient light intensity, the band reflectivity information, and the position information, fill light parameter analysis is performed to obtain camera fill light parameters, and the camera fill light parameters include fill light spectral distribution and target fill light intensity; the camera fill light parameters are used to perform fill light control and night information capture on the target hunting camera to obtain the multispectral perception image information.

[0016] A plurality of sensor modules are set in the target hunting camera, including an ambient light sensor, a ranging sensor, an angle sensor, and a fill light control module. The ambient light sensor is used to monitor and obtain the light intensity of the surrounding environment in real time to evaluate the brightness conditions of the current environment; the ranging sensor and the angle sensor are used to obtain the position information of the hunting target, that is, the position and angle information of the target. The fill light control module calls a preset object spectral reflectivity data set for matching to obtain the band reflectivity information of the hunting target. The object spectral reflectivity data set contains the reflectivity characteristics of different objects in different spectral bands and can provide accurate data for the reflection performance of each object in each spectral band. Combining the ambient light intensity, the band reflectivity information, and the position information of the target, fill light parameter analysis is performed to obtain the camera fill light parameters applicable to the current environment and target. The camera fill light parameters include the fill light spectral distribution and the target fill light intensity; the fill light spectral distribution is used to determine the fill light intensity distribution in different bands, and the target fill light intensity is the optimal fill light intensity calculated based on the distance of the target, the ambient light conditions, and the reflectivity information. The obtained camera fill light parameters are used to control the fill light of the target hunting camera and perform night information capture to obtain multi-spectral perception image information.

[0017] Furthermore, obtaining the camera fill light parameters includes: According to the fill light control module, obtaining a set of fill light sources, and at the same time, performing band identification on the set of fill light sources according to the ambient light intensity to obtain the band spectral intensity; constructing a fill light spectral weight function: Characterizing the fill light weight of the band spectrum, ρλ is the band reflectivity information, ρmax is the maximum band reflectivity, Ieλ is the band spectral intensity, and δ is to prevent zero terms; using the fill light spectral weight function to calculate the fill light spectral weight of the band spectral intensity and the band reflectivity information to obtain the fill light spectral distribution; performing fill light intensity calculation based on the ambient light intensity, the band reflectivity information, and the position information to obtain the target fill light intensity, and obtaining the camera fill light parameters according to the fill light spectral distribution and the target fill light intensity.

[0018] The fill light source set includes multiple fill light sources, and each light source provides different spectral intensities in different bands; obtain the fill light source set from the fill light control module, and based on the acquired ambient light intensity, perform band identification on these fill light sources to determine the band spectral intensity corresponding to each fill light source, that is, according to the measured ambient light intensity, identify and classify the bands corresponding to each fill light source, so as to be able to select an appropriate spectral band for fill light according to the environmental conditions. Through the fill light spectral weight function, perform fill light spectral weight calculation on the acquired band spectral intensity and band reflectance information, so as to obtain the fill light spectral distribution. The fill light spectral distribution indicates the intensity distribution of the fill light source in different bands and provides a quantitative basis for subsequent fill light control. Based on the ambient light intensity, band reflectance information, and target position information, calculate the fill light intensity to obtain the target fill light intensity. By combining the fill light spectral distribution and the target fill light intensity, complete camera fill light parameters are obtained, covering the spectral distribution and intensity information of the fill light.

[0019] Furthermore, obtaining the target fill light intensity includes: Construct a fill light intensity calculation function: where Is represents the fill light intensity, K is the environmental coefficient, usually taken as 1.5 at night, Id is the desired image brightness, ρλ is the band reflectance information, Ieλ is the band spectral intensity, and ε is to prevent zero terms; use the fill light intensity calculation function to perform fill light calculation on the ambient light intensity and the band reflectance information to obtain the initial fill light intensity; according to the position information, introduce an attenuation factor, and based on the attenuation factor, perform dynamic product adjustment on the initial fill light intensity to obtain the target fill light intensity.

[0020] Construct a fill light intensity calculation function for calculating the required fill light intensity according to the ambient light intensity and band reflectance information. Where Is represents the fill light intensity, K is the environmental coefficient, usually taken as 1.5 at night, Id is the desired image brightness, representing the image brightness value expected to be obtained in the target environment, ρλ is the band reflectance information, Ieλ is the band spectral intensity, representing the light intensity of the environmental light source in this band, and ε is to prevent zero terms.

[0021] Using a fill light intensity calculation function, perform fill light calculation on the acquired ambient light intensity and band reflectance information to obtain the initial fill light intensity. The initial fill light intensity reflects the fill light requirement calculated based on the expected image brightness and band reflectance information under the current lighting environment. According to the position information of the target, introduce an attenuation factor, which is used to simulate the attenuation phenomenon of light during propagation, and its value usually depends on the distance between the target and the camera; according to the attenuation factor, dynamically adjust the initial fill light intensity by multiplication, that is, adjust the fill light intensity based on the relative position between the target and the camera to make it adapt to different shooting distances and angles, so as to obtain the final target fill light intensity.

[0022] Perform weighted fusion processing on the multi-spectral perception image information according to the spectral band characteristic information to obtain the night fusion image information.

[0023] The spectral band characteristic information refers to the performance of the reflection, absorption, and transmission of different spectral bands (such as visible light, near-infrared light, thermal infrared light, etc.) on objects. The multi-spectral perception image information contains image data of multiple bands, and the image data of each band has different spectral characteristics and response capabilities in a specific environment. The visible light band is usually effective in a relatively bright environment, but its contribution is small in a low-light environment; the near-infrared band has a good response to a low-light environment and can provide certain depth information; the thermal infrared band can effectively identify heat sources through temperature differences, especially showing superiority under extremely low light conditions. According to the characteristics of different bands, assign a weighting factor to each band, and the weighting factor is dynamically calculated based on the spectral characteristics of each band and its contribution degree in a specific environment. For example, in a night environment, the thermal infrared band may be given a higher weighting factor to make full use of its adaptability to a low-light environment, while the visible light band may be adjusted according to the ambient light intensity. Use the above weighting factors to perform weighted processing on the images of each band. Specifically, the pixel value of each band's image will be multiplied by the weighting factor of that band to obtain the weighted band image; then, the weighted images of all bands are synthesized to obtain the final fused image.

[0024] Furthermore, obtaining the night fusion image information includes: Initialize a multi-band filter according to the spectral band characteristic information; perform filtering processing on the multi-spectral perception image information according to the multi-band filter to obtain standard multi-spectral perception image information; evaluate the night imaging contribution degree of the spectral band information to obtain the band night imaging contribution degree, and determine the band fusion weight factor information according to the band night imaging contribution degree; perform weighted fusion processing on the multi-spectral perception image information based on the band fusion weight factor information to obtain the night fusion image information.

[0025] Specifically, according to the spectral band characteristic information, a multi-band filter is initialized. The multi-band filter is used to preprocess the images of each spectral band by filtering out noise, enhancing image details, and ensuring the consistency of the image data of each band during subsequent processing. The multi-band filter is designed according to the spectral characteristics of each band and the reflection characteristics of the target object to optimize the effective information of the images of each band and avoid unnecessary interference.

[0026] Using the multi-band filter, filter the multi-spectral perception image information, including operations such as denoising, contrast enhancement, and image smoothing, to ensure that the images of each band have a high signal-to-noise ratio and detailed performance, thereby obtaining standard multi-spectral perception image information. Evaluate the contribution of each spectral band information to night imaging to obtain the contribution of each band to night imaging. Optionally, analyze the characteristics of each spectral band, such as the visible light band, near-infrared band, and thermal infrared band; the visible light band has a small contribution in low-light environments, while the near-infrared band and thermal infrared band perform excellently in low-light or no-light conditions at night. Especially the thermal infrared band can effectively capture the thermal radiation information of the object and is not affected by ambient light; evaluate the lighting conditions of the night environment, including ambient light intensity, thermal radiation of the target object, etc., to determine the performance of each band in this environment; by calculating the reflectivity and spectral intensity of each band, evaluate its contribution to the imaging quality. For example, the near-infrared band enhances the details of the target in low-light environments, while the thermal infrared band provides information related to temperature differences; quantify the contribution of each band to night imaging according to factors such as the reflectivity, imaging clarity, contrast, and noise suppression ability of each band. This contribution reflects the effectiveness of each band in the night image, and the bands with higher contributions will have a greater weight in the final image fusion.

[0027] Determine the band fusion weight factor information according to the contribution of each band to night imaging; the weight factor is allocated according to the contribution of each band, and the bands with higher contributions will be given greater weights. The weight factor determines the influence of each band in the final image, ensuring that the bands that contribute more to the image quality can occupy a more important position in the fused image. Based on the band fusion weight factor information, perform weighted fusion processing on the multi-spectral perception image information. Specifically, the image of each band will be multiplied by its corresponding weight factor, and then the weighted images of each band will be synthesized to obtain the final night fusion image information. Through weighted fusion, the advantages of each band can be maximized, and the brightness, contrast, clarity, and detailed performance of the image can be optimized to ensure high-quality images in low-light night environments.

[0028] Construct a hunting entity tag library, and based on the hunting entity tag library, divide the night fusion image information into N night image regions.

[0029] The pre-stored hunting entity tag library includes image feature information of various typical hunting targets, such as environmental elements like animals (deer, wild boars, foxes, etc.), trees, shrubs, rocks, water bodies, etc. The hunting entity tag library can be constructed by manually annotating training sample images and combining image recognition algorithms. The tag information includes the shape features, edge contours, thermal imaging features, near-infrared response features, and contrast characteristics in multi-band images of the targets. Match the characteristics of the night fusion image information with various hunting entities in the tag library, and based on the matching results, perform intelligent partitioning of the image to identify the regions in the image that conform to the tag characteristics. The identified regions are classified and divided according to the corresponding entities to form N night image regions, and each image region corresponds to a potential hunting target or environmental feature region.

[0030] Furthermore, obtaining N night image regions includes: Identify the demand type and imaging clarity level of the hunting entity tag library according to the hunting imaging requirements to obtain imaging clarity level information; classify and summarize the hunting entity tag library according to the imaging clarity level information to obtain a multi-level hunting entity tag set; use the multi-level hunting entity tag set to classify and identify the night fusion image information to obtain multi-level night tag image information; perform region connectivity and integration division on the multi-level night tag image information to obtain the N night image regions.

[0031] Preferably, identify the demand type and imaging clarity level of the tags in the hunting entity tag library according to the hunting imaging requirements; the demand type identification includes functional classification of the target categories, such as prey targets (e.g., wild deer, wild boars), human activity targets (e.g., patrol personnel), background interference targets (e.g., trees, rocks), etc.; the imaging clarity level is based on the resolution requirements, contrast level, and recognition reliability of night imaging, and the imaging quality requirements of different targets are classified to form imaging clarity level information. According to the imaging clarity level information, classify and summarize the tags in the hunting entity tag library to construct a multi-level hunting entity tag set. The multi-level hunting entity tag set aggregates the target entity features under similar imaging requirements with the clarity level as the dimension. Apply the multi-level hunting entity tag set to the night fusion image information, perform area-by-area scanning and matching on the image to identify the target types and imaging levels corresponding to different regions in the image, so as to form multi-level night tag image information. Perform region connectivity analysis on the same-type regions in the tag image, and use image segmentation and boundary recognition techniques to integrate and divide the connected tag regions, and finally obtain N night image regions with clear semantics and feature attribution.

[0032] According to the hunting entity tag library, excavate and obtain the hunting scene source image library and the hunting scene night image library, and perform associated feature extraction and imaging enhancement analysis on the hunting scene source image library and the hunting scene night image library to build a multi-channel for hunting night imaging enhancement.

[0033] Based on the information such as target categories, morphological features, and spectral response attributes included in the constructed hunting entity tag library, perform image sample matching and screening, and excavate and establish two independent image resource libraries from a preset image dataset: one is the hunting scene source image library, which contains high-quality, multi-band imaging samples taken in natural light or daytime environments, and is used to characterize the image features of various hunting entities under ideal imaging conditions; the other is the hunting scene night image library, which contains typical hunting environment image samples taken at night or under low-light conditions, and is used to record the change features of various entities in the night imaging environment.

[0034] Perform associated feature extraction on the hunting scene source image library and the hunting scene night image library. Specifically, through methods such as object detection, edge recognition, contour analysis, and thermal infrared distribution modeling, extract the key imaging features of the corresponding entities in each pair of source images and night images, and perform comparison and feature mapping. After completing the associated feature extraction, based on the extracted mapping relationship, build an imaging enhancement model, and accordingly build a multi-channel for hunting night imaging enhancement.

[0035] Furthermore, building a multi-channel for hunting night imaging enhancement includes: Perform associated feature extraction on the hunting scene source image library and the hunting scene night image library according to the hunting entity tag library to obtain the hunting entity source associated feature set and the hunting entity night associated feature set; perform mapping association processing on the hunting entity source associated feature set and the hunting entity night associated feature set to determine the hunting image sample dataset; perform image enhancement training on the hunting image sample dataset based on the imaging clarity level information to build a multi-channel for hunting night imaging enhancement.

[0036] Based on the target categories, outer contours, texture features, spectral responses, and thermal characteristics of various hunting entities in the tag library, perform feature analysis on the target areas in the source image and the night image, and extract the hunting entity source associated feature set formed under natural light conditions and the hunting entity night associated feature set formed under low-light or infrared conditions at night respectively. Perform mapping association processing on the hunting entity source associated feature set and the hunting entity night associated feature set. The feature correspondence relationship between the source image and the night image can be constructed by means of feature point matching, morphological structure alignment, distribution density pairing, etc., and then the hunting image sample dataset is determined. This dataset is used to characterize the feature conversion law of hunting entities under different imaging conditions and serves as an important basis for enhancing model training.

[0037] Based on the image clarity level information, the constructed hunting image sample data set is trained for image enhancement. According to the clarity requirement level of different targets in night imaging, the hierarchical enhancement strategy and model structure are called, and the samples are trained through the image enhancement network (such as a multi-branch convolutional network, a feature compensation network, or an attention guidance network) to improve the image quality in terms of target clarity, edge discernibility, texture restoration, etc. After the training is completed, a hunting night imaging enhancement multi-channel system is built, where each channel corresponds to an imaging clarity level and target type, and can be called in parallel in the subsequent image processing stage to achieve personalized night image enhancement output for specific targets, significantly improving target recognition accuracy and visual effects.

[0038] Furthermore, we build a hunting night imaging enhancement multi-channel system, including: According to the imaging clarity level information, an image enhancement multi-channel architecture is constructed, in which each channel architecture corresponds to an imaging clarity level one by one; the image enhancement multi-channel architecture is used to perform image enhancement training on the hunting image sample data set to obtain a night imaging enhancement channel set; the night imaging enhancement channel set is parallel labeled and integrated to construct the hunting night imaging enhancement multi-channel.

[0039] Specifically, the hunting entities are divided into multiple levels according to the imaging clarity requirements, such as high-precision recognition level, medium-definition recognition level, and background perception level, and multiple corresponding image enhancement channels are designed accordingly. Each channel architecture is configured exclusively according to the imaging characteristics of the level to which it belongs, covering the depth of the convolution structure, feature extraction strategy, contrast enhancement method, edge compensation mechanism, etc., to ensure that the channel enhancement logic matches the target clarity level and realize the directional enhancement processing capability of layered and sub-target. The above-mentioned hunting image sample data set is trained using the image enhancement multi-channel architecture constructed above. During the training process, the system automatically assigns image samples under different clarity labels to the corresponding channels, and performs enhancement model training through each channel, including but not limited to image brightness compensation, detail reconstruction, infrared heat source extraction, low-contrast area enhancement, etc., so that each channel has the image enhancement capability to adapt to its own level target. After the training is completed, a set of night imaging enhancement channels is output, that is, a set of multiple imaging channel models with independent enhancement strategies. Finally, the set of night imaging enhancement channels is parallel labeled and integrated to form the final hunting night imaging enhancement multi-channel.

[0040] Furthermore, a set of night imaging enhancement channels is obtained, including: Map and classify the hunting image sample dataset according to the image enhancement multi-channel architecture to obtain a multi-channel image sample dataset; use a convolutional neural network structure to perform image enhancement training and iterative verification tuning on the multi-channel image sample dataset respectively to obtain the night imaging enhancement channel set.

[0041] Based on the imaging clarity level information, map each target area in the image sample to the corresponding channel in the image enhancement multi-channel architecture according to its imaging requirements such as the required clarity level, contrast characteristics, infrared performance, or edge integrity, etc., to achieve hierarchical organization of the hunting image sample on the channel architecture. Through mapping, a multi-channel image sample dataset is formed. The multi-channel image sample dataset has the characteristics of clear structure, strong label consistency, and high training adaptability, providing high-quality input for subsequent deep learning model training. Use a convolutional neural network (CNN) structure to perform image enhancement training on the multi-channel image sample dataset respectively. An independent neural network model is constructed for each channel, and the network structure includes multiple convolutional layers, a feature extraction module, a residual connection unit, and an image reconstruction module to adapt to the image enhancement requirements of various clear level targets. During the training process, through supervised learning, use the corresponding features of the source image and the night image for contrast training, and use a loss function to guide the network to gradually optimize key performance indicators such as image brightness restoration, texture detail recovery, and thermal imaging enhancement. At the same time, perform iterative verification and tuning, evaluate the network output effect through the validation set, and dynamically adjust the learning rate, network parameters, and enhancement strategy between channels to ensure the optimal performance of each channel training model in terms of generalization ability and enhancement effect. After training is completed, obtain the night imaging enhancement channel set. This channel set consists of multiple image enhancement models dedicated to different imaging clarity levels, and can be called in parallel as needed during subsequent processing to achieve partitioned and hierarchical image enhancement processing, significantly improving the clarity, detail restoration ability, and target recognition accuracy of night images.

[0042] Map the N night image regions to the hunting night imaging enhancement multi-channel for parallel image enhancement to generate hunting night enhanced image information.

[0043] Based on the corresponding target entity labels and imaging clarity level information in each of the N night image areas, determine the image enhancement channel that should be matched and called for each image area. Specifically, the system maps the area to the corresponding enhancement channel in the hunting night imaging enhancement multi-channel architecture according to the type of target in each image area (such as animals, humans, environmental objects, etc.), regional feature complexity, lighting conditions and clarity level. For the N night image areas that have completed mapping and matching, the system activates the channel models corresponding to them in the night imaging enhancement channel set in parallel and performs image enhancement operations. In the parallel processing process, each channel independently performs directional image enhancement operations such as brightness optimization, edge sharpening, thermal texture restoration, and low-contrast area enhancement on the image area it receives, ensuring that different areas obtain the best image enhancement effect in their respective adaptation channels. Finally, the image area results output by each enhancement channel are reconstructed and spliced ​​according to the original image space coordinates to generate a unified and complete hunting night enhanced image information.

[0044] In summary, the embodiments of the present application have at least the following technical effects: First, a target hunting camera is obtained. The target hunting camera has a built-in multispectral sensor. Multispectral perception image information is obtained by capturing the target hunting camera at night. The multispectral perception image information includes visible light, near-infrared light and thermal infrared light band images. Then, the multispectral perception image information is weighted and fused according to the spectral band characteristic information to obtain night fusion image information. Further, a hunting entity label library is constructed, and the night fusion image information is divided into regions based on the hunting entity label library to obtain N night image regions. Then, according to the hunting entity label library, the hunting scene source image library and the hunting scene night image library are mined and obtained, and the hunting scene source image library and the hunting scene night image library are subjected to correlation feature extraction and imaging enhancement analysis, and a hunting night imaging enhancement multi-channel is constructed. Finally, the N night image regions are mapped to the hunting night imaging enhancement multi-channel for parallel image enhancement to generate hunting night enhanced image information. The technical problem of poor night imaging quality of hunting cameras in the prior art is solved, and the technical effect of improving the night imaging quality of hunting cameras is achieved.

[0045] Embodiment 2 is based on the same inventive concept as the night imaging enhancement method for hunting cameras in the aforementioned embodiment. Figure 2 As shown, the present application provides a night imaging enhancement system for a hunting camera, wherein the system includes: The image acquisition module 11 is used to obtain a target hunting camera. The target hunting camera is built with a multispectral sensor. The multispectral perception image information is obtained by the target hunting camera through night capture. The multispectral perception image information includes visible light, near-infrared light, and thermal infrared light band images. The fusion module 12 is used to perform weighted fusion processing on the multispectral perception image information according to the spectral band characteristic information to obtain night fusion image information. The region division module 13 is used to build a hunting entity tag library and divide the night fusion image information based on the hunting entity tag library to obtain N night image regions. The analysis module 14 is used to, according to the hunting entity tag library, excavate and obtain a hunting scene source image library and a hunting scene night image library, and perform associated feature extraction and imaging enhancement analysis on the hunting scene source image library and the hunting scene night image library to build a multi-channel for hunting night imaging enhancement. The image enhancement module 15 is used to map the N night image regions to the multi-channel for hunting night imaging enhancement for parallel image enhancement to generate hunting night enhanced image information.

[0046] Further, the image acquisition module 11 is used to execute the following method: An ambient light sensor, a ranging sensor, an angle sensor, and a fill light control module are set in the target hunting camera. The ambient light intensity is monitored and obtained in real time through the ambient light sensor. The position information of the hunting target is obtained through the ranging sensor and the angle sensor. Based on the fill light control module, a dataset of object spectral reflectance is called, and the band reflectance information of the hunting target is obtained through matching of the dataset of object spectral reflectance. Based on the ambient light intensity, the band reflectance information, and the position information, fill light parameter analysis is performed to obtain camera fill light parameters. The camera fill light parameters include fill light spectral distribution and target fill light intensity. The target hunting camera is controlled for fill light and night information capture by using the camera fill light parameters to obtain the multispectral perception image information.

[0047] Further, the image acquisition module 11 is used to execute the following method: According to the fill light control module, a fill light source set is obtained, and at the same time, the fill light source set is band-identified according to the ambient light intensity to obtain band spectral intensity. A fill light spectral weight function is constructed: Characterize the spectral supplementary lighting weight. ρλ is the band reflectance information, ρmax is the maximum band reflectance, Ieλ is the spectral intensity of the band, and δ is to prevent zero terms. Use the supplementary lighting spectral weight function to calculate the supplementary lighting spectral weight for the spectral intensity of the band and the band reflectance information to obtain the supplementary lighting spectral distribution. Calculate the supplementary lighting intensity based on the ambient light intensity, the band reflectance information, and the position information to obtain the target supplementary lighting intensity, and obtain the camera supplementary lighting parameters according to the supplementary lighting spectral distribution and the target supplementary lighting intensity.

[0048] Further, the image acquisition module 11 is used to execute the following method: Construct a supplementary lighting intensity calculation function: Where Is represents the supplementary lighting intensity, K is the environmental coefficient, usually taken as 1.5 at night, Id is the desired image brightness, ρλ is the band reflectance information, Ieλ is the spectral intensity of the band, and ε is to prevent zero terms. Use the supplementary lighting intensity calculation function to perform supplementary lighting calculation on the ambient light intensity and the band reflectance information to obtain the initial supplementary lighting intensity. According to the position information, introduce an attenuation factor, and perform dynamic product adjustment on the initial supplementary lighting intensity based on the attenuation factor to obtain the target supplementary lighting intensity.

[0049] Further, the fusion module 12 is used to execute the following method: Initialize a multi-band filter according to the spectral band characteristic information; perform filtering processing on the multi-spectral perception image information according to the multi-band filter to obtain standard multi-spectral perception image information; evaluate the contribution degree of the spectral band information to night imaging to obtain the contribution degree of the spectral band to night imaging, and determine the band fusion weight factor information according to the contribution degree of the spectral band to night imaging; perform weighted fusion processing on the multi-spectral perception image information based on the band fusion weight factor information to obtain night fusion image information.

[0050] Further, the region division module 13 is used to execute the following method: Identify the demand type and imaging clarity level of the hunting entity tag library according to the hunting imaging requirements to obtain the imaging clarity level information; classify and summarize the hunting entity tag library according to the imaging clarity level information to obtain a multi-level hunting entity tag set; use the multi-level hunting entity tag set to classify and identify the night fusion image information to obtain multi-level night tag image information; perform regional connectivity and integration division on the multi-level night tag image information to obtain the N night image regions.

[0051] Further, the analysis module 14 is used to execute the following method: Extract the associated features of the hunting scene source image library and the hunting scene night image library according to the hunting entity tag library to obtain the hunting entity source associated feature set and the hunting entity night associated feature set; perform mapping association processing on the hunting entity source associated feature set and the hunting entity night associated feature set to determine the hunting image sample data set; perform image enhancement training on the hunting image sample data set based on the imaging clarity level information to build a multi-channel for hunting night imaging enhancement.

[0052] Further, the analysis module 14 is used to execute the following method: Build an image enhancement multi-channel architecture according to the imaging clarity level information, and each channel architecture in the image enhancement multi-channel architecture corresponds to the imaging clarity level one by one; use the image enhancement multi-channel architecture to perform image enhancement training on the hunting image sample data set to obtain a set of night imaging enhancement channels; perform parallel identification integration on the set of night imaging enhancement channels to build the multi-channel for hunting night imaging enhancement.

[0053] Further, the analysis module 14 is used to execute the following method: Map and classify the hunting image sample data set according to the image enhancement multi-channel architecture to obtain a multi-channel image sample data set; use the convolutional neural network structure to perform image enhancement training and iterative verification and tuning on the multi-channel image sample data set respectively to obtain the set of night imaging enhancement channels.

[0054] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above description of specific embodiments of this specification is made. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0055] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0056] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A night imaging enhancement method for a hunting camera, characterized in that: The method comprises: Acquire a target hunting camera, wherein the target hunting camera has a built-in multispectral sensor, and obtain multispectral perception image information by capturing the target hunting camera at night, wherein the multispectral perception image information includes visible light, near infrared light, and thermal infrared light band images; Performing weighted fusion processing on the multi-spectral perception image information according to the spectral band characteristic information to obtain night fusion image information; Constructing a hunting entity tag library, and dividing the night fusion image information into regions based on the hunting entity tag library to obtain N night image regions; According to the hunting entity tag library, a hunting scene source image library and a hunting scene night image library are mined and acquired, and correlation feature extraction and imaging enhancement analysis are performed on the hunting scene source image library and the hunting scene night image library to build a hunting night imaging enhancement multi-channel; The N night image areas are mapped to the hunting night imaging enhancement multi-channel for parallel image enhancement to generate hunting night enhanced image information.

2. The night imaging enhancement method for a hunting camera as claimed in claim 1, characterized in that: The multi-spectral perception image information is obtained by capturing the target hunting camera at night, including: An ambient light sensor, a distance sensor, an angle sensor and a fill light control module are arranged in the target hunting camera, and the ambient light intensity is acquired by real-time monitoring through the ambient light sensor, and the position information of the hunting target is acquired through the distance sensor and the angle sensor; Based on the fill light control module, calling the object spectral reflectance data set, and obtaining the band reflectance information of the hunting target through matching the object spectral reflectance data set; Perform fill light parameter analysis based on the ambient light intensity, the band reflectivity information and the position information to obtain camera fill light parameters, wherein the camera fill light parameters include fill light spectrum distribution and target fill light intensity; The camera fill light parameters are used to perform fill light control and nighttime information capture on the target hunting camera to obtain the multi-spectral perception image information.

3. The nighttime imaging enhancement method for a hunting camera as claimed in claim 2, characterized in that: The obtaining of camera fill light parameters includes: According to the fill light control module, a fill light source set is obtained, and at the same time, a band identification is performed on the fill light source set according to the ambient light intensity to obtain a band spectrum intensity; Construct fill light spectral weight function: in, Characterizes the band spectral fill-in light weight, ρλ is the band reflectance information, ρmax is the band maximum reflectance, Ieλ is the band spectral intensity, and δ is the zero elimination prevention; The fill light spectrum weight function is used to calculate the fill light spectrum weight of the band spectrum intensity and the band reflectivity information to obtain the fill light spectrum distribution; The fill light intensity is calculated based on the ambient light intensity, the band reflectivity information and the position information to obtain the target fill light intensity, and the camera fill light parameters are obtained according to the fill light spectrum distribution and the target fill light intensity.

4. The nighttime imaging enhancement method for a hunting camera as claimed in claim 3, characterized in that: The step of obtaining the target fill light intensity includes: Construct fill light intensity calculation function: Among them, Is represents the fill light intensity, K is the environmental coefficient, which is usually 1.5 at night, Id is the desired image brightness, ρλ is the band reflectance information, Ieλ is the band spectral intensity, and ε is the zero elimination prevention term; Using the fill light intensity calculation function to perform fill light calculation on the ambient light intensity and the band reflectivity information to obtain an initial fill light intensity; According to the position information, an attenuation factor is introduced, and the initial fill light intensity is dynamically multiplied and adjusted based on the attenuation factor to obtain the target fill light intensity.

5. The nighttime imaging enhancement method for a hunting camera as claimed in claim 1, characterized in that: The obtaining of night-time fusion image information includes: Initializing a multi-band filter according to the spectral band characteristic information; Performing filtering processing on the multi-spectral perception image information according to the multi-band filter to obtain standard multi-spectral perception image information; Evaluate the night imaging contribution of the spectral band information to obtain the night imaging contribution of the band, and determine the band fusion weight factor information according to the night imaging contribution of the band; The multispectral perception image information is subjected to weighted fusion processing based on the band fusion weight factor information to obtain nighttime fusion image information.

6. The nighttime imaging enhancement method for a hunting camera as claimed in claim 1, characterized in that: The N nighttime image regions are obtained, including: According to the hunting imaging requirements, the hunting entity tag library is identified by the type of requirement and the imaging clarity is graded to obtain imaging clarity level information; Classify and summarize the hunting entity tag library according to the imaging clarity level information to obtain a multi-level hunting entity tag set; Using the multi-level hunting entity label set to classify and label the night fusion image information, to obtain multi-level night label image information; The multi-level nighttime label image information is connected and integrated to obtain the N nighttime image regions.

7. The nighttime imaging enhancement method for a hunting camera as claimed in claim 6, characterized in that: The hunting night imaging enhancement multi-channel setup includes: Extracting associated features from the hunting scene source image library and the hunting scene night image library according to the hunting entity tag library to obtain a hunting entity source associated feature set and a hunting entity night associated feature set; Mapping and associating the hunting entity source associated feature set and the hunting entity night associated feature set to determine a hunting image sample data set; Based on the imaging clarity level information, image enhancement training is performed on the hunting image sample data set to build a hunting night imaging enhancement multi-channel.

8. The nighttime imaging enhancement method for a hunting camera as claimed in claim 7, characterized in that: The hunting night imaging enhancement multi-channel setup includes: According to the imaging clarity level information, an image enhancement multi-channel architecture is constructed, wherein each channel architecture in the image enhancement multi-channel architecture corresponds to the imaging clarity level one by one; Using the image enhancement multi-channel architecture to perform image enhancement training on the hunting image sample dataset to obtain a night imaging enhancement channel set; The night imaging enhancement channel set is integrated in parallel to build the hunting night imaging enhancement multi-channel.

9. The nighttime imaging enhancement method for a hunting camera as claimed in claim 8, characterized in that: The step of obtaining a nighttime imaging enhancement channel set comprises: Mapping and classifying the hunting image sample dataset according to the image enhancement multi-channel architecture to obtain a multi-channel image sample dataset; A convolutional neural network structure is used to perform image enhancement training and iterative verification tuning on the multi-channel image sample data set to obtain the night imaging enhancement channel set.

10. A night imaging enhancement system for a hunting camera, characterized in that: A system for implementing the night imaging enhancement method for a hunting camera according to any one of claims 1 to 9, the system comprising: An image acquisition module is used to acquire a target hunting camera, wherein the target hunting camera has a built-in multispectral sensor, and multispectral perception image information is obtained by capturing the target hunting camera at night, wherein the multispectral perception image information includes visible light, near infrared light, and thermal infrared light band images; A fusion module, used for performing weighted fusion processing on the multi-spectral perception image information according to the spectral band characteristic information to obtain night fusion image information; A region division module, used for building a hunting entity label library, and performing region division on the night fusion image information based on the hunting entity label library to obtain N night image regions; An analysis module is used to mine and obtain a hunting scene source image library and a hunting scene night image library according to the hunting entity tag library, and to perform associated feature extraction and imaging enhancement analysis on the hunting scene source image library and the hunting scene night image library, and to build a hunting night imaging enhancement multi-channel; The image enhancement module is used to map the N night image areas to the hunting night imaging enhancement multi-channel for parallel image enhancement to generate hunting night enhanced image information.

Images