Power consumption control method, system and equipment of thermal infrared imager and medium

Abstract

The invention provides a power consumption control method, system and device of a thermal infrared imager and a medium. The method comprises the following steps: acquiring infrared image data of the thermal infrared imager and power consumption data of a target imaging assembly of the thermal infrared imager in real time; performing feature analysis on the infrared image data through a preset target detection model to obtain a preset target in the image; grading the importance of the current detection scene based on a preset target to obtain an attention level, and converting the attention level into a quantitative attention parameter according to a preset conversion rule; determining a target performance configuration file according to the quantitative attention parameter; obtaining a lowest target power consumption value meeting the operation parameters based on a mapping relation between the operation parameters and preset performance power consumption; and generating a power consumption adjusting signal by combining the lowest target power consumption value and the power consumption data, thereby adjusting the power consumption of the target imaging component. By implementing the technical scheme provided by the invention, dynamic and accurate matching of the power consumption and the imaging performance of the thermal infrared imager is realized.

Description

Technical Field

[0001] This invention relates to the field of infrared thermal imager control technology, and in particular to a power consumption control method, system, device and medium for infrared thermal imagers. Background Technology

[0002] Infrared thermal imagers are widely used in security, industrial, and portable monitoring fields due to their unique imaging capabilities. Especially in battery-powered scenarios, the devices need to operate for extended periods. Therefore, minimizing power consumption and extending battery life while ensuring detection performance is a crucial technical goal in this field.

[0003] Existing technologies mostly operate using fixed, preset mechanisms, such as timed sleep mode or manual low-power mode. Their power consumption regulation mechanisms operate independently of the real-time image data stream. Specifically, timed sleep mode operates based on a preset time sequence, meaning the device's operating state is not directly related to whether critical events occur in the scene. Switching between operating modes relies on manual commands. This interaction method is limited by operating conditions in applications requiring long-term, fully autonomous operation. This separation of power consumption regulation mechanisms from real-time image content restricts the device's ability to autonomously and intelligently adjust power consumption according to scene changes. Summary of the Invention

[0004] In view of this, this application provides a power consumption control method, system, device and medium for an infrared thermal imager to solve the above problems.

[0005] Firstly, a power consumption control method for an infrared thermal imager is provided, applied to a power consumption control platform. This method includes:

[0006] Real-time acquisition of infrared image data from infrared thermal imagers, as well as power consumption data of target imaging components in infrared thermal imagers;

[0007] The infrared image data is analyzed by a preset target detection model to obtain the preset target in the infrared image data.

[0008] Based on the preset target, the importance of the current detection scene corresponding to the infrared image data is classified into levels of attention, and the attention levels are converted into corresponding quantitative attention parameters through preset conversion rules.

[0009] Based on the quantified attention parameters, the target performance profile is determined. The target performance profile includes runtime parameters, including the target frame rate and the target resolution.

[0010] Based on the operating parameters, the minimum target power consumption value required to meet the operating parameters is obtained through a preset performance-power consumption mapping relationship;

[0011] Based on the minimum target power consumption value and power consumption data, a power consumption adjustment signal is generated, and the power consumption of the target imaging component is adjusted according to the power consumption adjustment signal.

[0012] The above technical solution analyzes real-time infrared image data using a pre-defined target detection model to obtain preset targets, fundamentally establishing a direct perception link between the power consumption control system and the image content stream. This breaks down the barrier of existing technology adjustment mechanisms operating independently of the data stream. Based on the acquired preset targets, the importance of the scene is quantified, and the required target performance configuration file is determined accordingly. This successfully transforms the abstract "scene content" into a concrete "performance decision," achieving intelligent correlation. The optimal power consumption value is calculated based on the determined performance configuration, and closed-loop adjustment is executed. This constructs a complete power consumption control closed loop driven by real-time image content. Power consumption adjustment is no longer a timed or manual command unrelated to scene events, but rather dynamically driven by the image content, thus giving the device the ability to autonomously and intelligently adjust power consumption according to scene changes.

[0013] Optionally, a preset target detection model is used to perform feature analysis on the infrared image data to obtain preset targets in the infrared image data, specifically including:

[0014] Infrared image data is processed by multiple convolutional layers of the backbone network in the target detection model. Each stage of convolutional processing outputs a feature map of the corresponding resolution. The feature map output by the current stage is downsampled and then input into the next adjacent stage to obtain a set of feature maps consisting of all feature maps output by at least two stages. This set of feature maps is used as the multi-scale feature of the preset target. The feature map includes thermal contour and shape information.

[0015] Through the connection layer in the object detection model, the first feature map generated by the later stage of any two adjacent stages in the multi-scale feature is upsampled, and the upsampled first feature map is fused with the second feature map to obtain a fused feature map. The second feature map is generated by the earlier stage of any two adjacent stages. The fused feature map is processed by multi-branch convolution through a preset receptive field enhancement module to generate enhanced fused features.

[0016] The enhanced fusion features are classified using the detection head in the target detection model to determine the target category. The enhanced fusion features are then regressed using the regression head in the detection head to determine the bounding box. By combining the target category and the bounding box, the preset target is obtained.

[0017] The above technical solution, through a three-stage processing flow of "backbone network - connection layer - detection head" in the target detection model, provides a highly efficient and accurate target extraction path specifically tailored to the characteristics of infrared images. This ensures that the system can not only perform general analysis, but also precisely extract, fuse, and decode specific information (such as category and location) of preset targets from infrared images lacking texture and with low signal-to-noise ratios through a structured, well-defined deep learning model.

[0018] Optionally, based on a preset target, the importance of the current detection scene corresponding to the infrared image data is classified to obtain an attention level, and the attention level is converted into a corresponding quantitative attention parameter through a preset conversion rule, specifically including:

[0019] Acquire environmental parameters, including temperature, from sensors pre-configured in the infrared thermal imager.

[0020] Based on the target type and environmental parameters of the preset objectives, the attention level of the current detection scenario is determined through a preset scenario classification and grading system;

[0021] Through a pre-defined mapping relationship, the level of attention is converted into a quantitative parameter of attention.

[0022] The above technical solutions improve the accuracy and reliability of scene importance assessment by introducing environmental parameters and combining them with preset targets for comprehensive evaluation. They avoid the limitations of simply judging based on the presence or absence of a target, instead establishing a more comprehensive scene classification and grading system by integrating multi-dimensional information such as temperature and humidity. This makes the determination of "attention level" more closely reflect the needs of real-world complex environments.

[0023] Optionally, based on the target type and environmental parameters of the preset objectives, the attention level of the current detection scenario is determined through a preset scenario classification and grading system, specifically including:

[0024] If there is a first target among the preset targets, the first target belongs to the preset high importance category, or the first target's movement speed exceeds the first speed threshold, or the first target's thermal anomaly change rate exceeds the first change rate threshold, or the temperature exceeds the first temperature threshold, then the attention level is determined to be the first level.

[0025] If the first target is not among the preset targets, then determine whether there is a second target among the preset targets. The second target belongs to the preset medium importance category, or the movement speed of the second target exceeds the second speed threshold, or the thermal anomaly change rate of the second target exceeds the second change rate threshold. The second speed threshold is less than the first speed threshold, and the second change rate threshold is less than the first change rate threshold. If they exist, then the attention level is determined to be the second level.

[0026] When there are no first or second objectives among the preset goals, the attention level will be set to the third level, where the first level is higher than the second level, and the second level is higher than the third level.

[0027] The above technical solution provides a highly flexible, configurable, and widely applicable attention level determination mechanism by defining a hierarchical judgment rule based on "importance category" and "dynamic characteristics." This solution is no longer limited to a specific target (such as personnel) but, through abstract rules, can easily adapt to various application scenarios such as industrial inspection (judging thermal anomalies) and security monitoring (judging personnel intrusion), greatly enhancing the universality of the invention.

[0028] Optionally, based on the quantified attention parameters, a target performance profile is determined, which specifically includes:

[0029] Acquire the device status data of the infrared thermal imager, which includes the remaining battery power and the temperature of the target imaging components;

[0030] Based on the preset conservative power consumption mapping relationship and the quantized attention parameters, the first candidate performance profile is calculated. The first candidate performance profile includes the first operating parameters and the first imaging quality threshold.

[0031] Based on the preset priority performance mapping relationship and the quantified attention parameters, the second candidate performance profile is calculated. The second candidate performance profile includes the second operating parameters and the second imaging quality threshold.

[0032] When the remaining power is lower than the preset power threshold and / or the temperature is higher than the preset temperature threshold, the first candidate performance profile is selected as the target performance profile.

[0033] When the remaining battery power is higher than the battery power threshold and the temperature is lower than the temperature threshold, the second candidate performance profile is selected as the target performance profile. The total number of pixels processed per unit time corresponding to the running parameters in the first candidate performance profile is lower than the total number of pixels processed per unit time corresponding to the running parameters in the second candidate performance profile. The total number of pixels processed per unit time is the product of the target frame rate and the target resolution.

[0034] The above technical solutions, by introducing the device's own status (power level, temperature) as a second decision-making dimension, add "self-protection" and "resource awareness" intelligence to the power consumption control strategy. This allows the device to automatically switch to a more conservative power consumption strategy when resources are scarce, and to more actively protect performance when resources are abundant. This achieves robustness and reliability of device operation over a longer timescale, avoiding sudden shutdowns or damage caused by performance overload.

[0035] Optionally, based on the lowest target power consumption value and power consumption data, a power consumption adjustment signal is generated, and the power consumption of the target imaging component is adjusted according to the power consumption adjustment signal, specifically including:

[0036] The imaging quality index of the target imaging component is determined based on the signal-to-noise ratio of the infrared image data;

[0037] Obtain the target imaging quality threshold from the target performance configuration file;

[0038] When the imaging quality index is lower than the target imaging quality threshold, the quality difference between the imaging quality index and the target imaging quality threshold is calculated, and the quality difference is converted into a power consumption enhancement signal through a preset mapping relationship.

[0039] When the imaging quality index is not lower than the target imaging quality threshold, the power consumption error is calculated based on the lowest target power consumption value and power consumption data, and a power consumption reduction signal is generated based on the power consumption error.

[0040] The power consumption of the target imaging component is adjusted based on the power consumption enhancement signal or the power consumption reduction signal.

[0041] The above technical solution, by introducing an "imaging quality index" as feedback, elevates power consumption control from a "target tracking" logic to a "result guarantee" logic. It no longer simply approximates power consumption to a theoretically calculated value, but establishes a closed-loop optimization system with the hard constraint of "ensuring the final imaging quality is not lower than the standard." This ensures that under any operating condition, the user receives an image of acceptable quality, while simultaneously achieving true minimization of power consumption.

[0042] Optionally, the imaging quality index of the target imaging component is determined based on the signal-to-noise ratio of the infrared image data, specifically including:

[0043] In infrared image data, signal regions and noise regions are identified and divided. Signal regions are image regions that contain preset targets, while noise regions are image regions that do not contain preset targets.

[0044] Calculate the arithmetic mean of the pixel values ​​of all pixels within the signal region to obtain the signal average.

[0045] Calculate the standard deviation of pixel values ​​for all pixels within the noise region to obtain the noise standard deviation;

[0046] The signal-to-noise ratio (SNR) is obtained by dividing the average signal value by the noise standard deviation, and is used as an indicator of image quality.

[0047] The above technical solution, by defining a specific calculation method for the signal-to-noise ratio, provides an objective, stable, and repeatable quantitative standard for the "imaging quality index." By clearly defining the signal and noise regions and employing standard statistical methods, it ensures that the "imaging quality index," as the core feedback of the control system, is reliable and irrefutable, thereby guaranteeing the stable and accurate operation of the entire closed-loop optimization system.

[0048] Secondly, a power consumption control system for an infrared thermal imager is provided. The system includes a data acquisition module, a target detection module, a scene transformation module, a performance decision module, a power consumption calculation module, and a power consumption adjustment module, wherein:

[0049] The data acquisition module is configured to acquire infrared image data from the infrared thermal imager in real time, as well as power consumption data of the target imaging components in the infrared thermal imager.

[0050] The target detection module is configured to perform feature analysis on infrared image data using a preset target detection model to obtain preset targets in the infrared image data.

[0051] The scene conversion module is configured to classify the importance of the current detection scene corresponding to the infrared image data based on a preset target, obtain the attention level, and convert the attention level into the corresponding quantitative attention parameter through a preset conversion rule.

[0052] The performance decision module is configured to determine the target performance profile based on the quantified attention parameters. The target performance profile includes runtime parameters, including the target frame rate and the target resolution.

[0053] The power consumption calculation module is configured to obtain the minimum target power consumption value required to meet the operating parameters based on the preset performance-power consumption mapping relationship.

[0054] The power consumption adjustment module is configured to generate a power consumption adjustment signal based on the lowest target power consumption value and power consumption data, and to adjust the power consumption of the target imaging component according to the power consumption adjustment signal.

[0055] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the foregoing.

[0056] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the preceding descriptions.

[0057] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0058] By analyzing real-time infrared image data using a pre-defined target detection model to obtain preset targets, a direct perception link is fundamentally established between the power consumption control system and the image content stream, breaking down the barrier of existing technology adjustment mechanisms operating independently of the data stream. Based on the acquired preset targets, the importance of the scene is quantified, and the required target performance profile is determined accordingly, successfully transforming the abstract "scene content" into a concrete "performance decision," achieving intelligent correlation. The optimal power consumption value is calculated based on the determined performance profile, and closed-loop adjustment is executed. A complete power consumption control closed loop driven by real-time image content is constructed, making power consumption adjustment no longer a timed or manual command unrelated to scene events, but dynamically dominated by the image content, thus giving the device the ability to autonomously and intelligently adjust power consumption according to scene changes. Attached Figure Description

[0059] Figure 1 This is an exemplary system architecture diagram of an infrared thermal imager power consumption control method or a power consumption control system of an infrared thermal imager that applies the present application.

[0060] Figure 2 This is a flowchart illustrating a power consumption control method for an infrared thermal imager according to an embodiment of this application.

[0061] Figure 3 This is a schematic diagram of the power consumption control system of an infrared thermal imager according to an embodiment of this application;

[0062] Figure 4 This is a schematic diagram of the structure of an electronic device disclosed in the application embodiment.

[0063] Explanation of reference numerals in the attached figures: 100, System architecture; 101, First terminal device; 102, Second terminal device; 103, Third terminal device; 104, Network; 105, Server; 301, Data acquisition module; 302, Target detection module; 303, Scene conversion module; 304, Performance decision module; 305, Power consumption calculation module; 306, Power consumption adjustment module; 401, Processor; 402, Communication bus; 403, User interface; 404, Network interface; 405, Memory. Detailed Implementation

[0064] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0065] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0066] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0067] Figure 1 An exemplary system architecture diagram is shown, illustrating an embodiment of a power consumption control method or a power consumption control system for an infrared thermal imager to which this application can be applied.

[0068] like Figure 1 As shown, the system architecture 100 may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0069] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as model training applications, video recognition applications, web browser applications, social platform software, etc.

[0070] Terminal devices 101, 102, and 103 can be either hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with displays, including but not limited to smartphones, tablets, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptops, and desktop computers, etc. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices. They can be implemented as multiple software programs or software modules (e.g., multiple software programs or software modules used to provide distributed services) or as a single software program or software module. No specific limitations are imposed here.

[0071] When terminals 101, 102, and 103 are hardware devices, video capture devices can also be installed on them. These video capture devices can be various devices capable of capturing video, such as cameras, sensors, etc. Users can use the video capture devices on terminals 101, 102, and 103 to capture video.

[0072] Server 105 can be a server that provides various services, such as a backend server for processing data displayed on terminal devices 101, 102, and 103. The backend server can analyze and process the received data and can feed back the processing results (such as recognition results) to the terminal devices.

[0073] It should be noted that a server can be either hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software programs or software modules (e.g., multiple software programs or software modules used to provide distributed services), or as a single software program or software module. No specific limitations are made here.

[0074] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included. In particular, if the target data does not need to be obtained remotely, the above system architecture may exclude the network and include only terminal devices or servers.

[0075] Figure 2This is a flowchart illustrating a power consumption control method for an infrared thermal imager according to an embodiment of this application. This method can be implemented using a computer program, a microcontroller, or run on a power consumption control system for the infrared thermal imager. The computer program can be integrated into an application or run as a standalone utility application. The specific steps of the power consumption control method for an infrared thermal imager are described in detail below.

[0076] S201: Real-time acquisition of infrared image data from the infrared thermal imager, as well as power consumption data of the target imaging components in the infrared thermal imager.

[0077] In the embodiments of this application, the target imaging component refers to one or more core hardware modules inside the infrared thermal imager that are directly related to image capture, processing and transmission and are the object of power consumption adjustment, such as image sensors, signal processors, storage modules and communication modules.

[0078] For example, this step is a continuous, parallel, multi-source data acquisition process. On one hand, the system continuously acquires a series of image frames characterizing the thermal radiation distribution within the monitoring field of view from the infrared lens of the infrared thermal imager, and uses this image frame sequence as the infrared image data; on the other hand, the system uses high-precision current and voltage sensors pre-deployed on the one or more target imaging components to measure the operating current and voltage of each component in real time, and calculates a value characterizing its instantaneous energy consumption based on the measured values, and uses this value as the power consumption data.

[0079] S202: Perform feature analysis on infrared image data using a preset target detection model to obtain preset targets in the infrared image data.

[0080] For example, the input infrared image data is processed by a backbone network through staged, progressively deeper convolutional processing, and multi-scale features containing target thermal contour and shape information are generated through downsampling. A top-down information fusion process in the connection layer upsamples the feature maps with lower resolution and strong semantic information in the multi-scale features and fuses them with feature maps with higher resolution and strong spatial information. Then, a receptive field enhancement module performs multi-branch convolutional processing to aggregate contextual information, thereby generating enhanced fused features. These enhanced fused features are then sent to a detection head for parsing. This parsing process performs two tasks in parallel: using its classification head, the enhanced fused features are matched with features of a set of pre-learned standard target categories within the model to identify the target category of the object; simultaneously, its regression head performs regression processing to determine the precise bounding box of the object. Combining the identified target category with the corresponding bounding box yields a preset target that conforms to the standard target category and is successfully located in the image.

[0081] In a preferred embodiment, the backbone network employs an advanced convolutional neural network architecture incorporating ConvNeXt blocks. Specifically, the backbone network includes an initial convolutional layer and four consecutive processing stages. The initial convolutional layer performs preliminary downsampling and feature extraction on the input infrared image data. Each processing stage consists of multiple stacked ConvNeXt blocks used for depth-level feature extraction. Between adjacent processing stages, a downsampling layer is placed to halve the resolution of the feature map output from the previous stage to construct a multi-scale feature pyramid. This structure can efficiently extract contour and shape information from infrared images, laying the foundation for subsequent accurate detection.

[0082] In one possible implementation, feature analysis is performed on infrared image data using a preset target detection model to obtain a preset target from the infrared image data. Specifically, this includes: performing multi-stage convolution processing on the infrared image data using multiple convolutional layers of the backbone network in the target detection model. Each stage of convolution processing outputs a feature map at a corresponding resolution. The feature map output from the current stage is then downsampled and input into the next adjacent stage, resulting in a set of feature maps composed of all feature maps output from at least two stages. This set of feature maps is used as the multi-scale features of the preset target, and the feature maps include thermal contours and shape information. The target detection model then performs feature analysis on the infrared image data. The connection layer upsamples the first feature map generated by the later stage of any two adjacent stages in the multi-scale feature model, and fuses the upsampled first feature map with the second feature map to obtain a fused feature map. The second feature map is generated by the earlier stage of any two adjacent stages. The fused feature map is processed by multi-branch convolution through a preset receptive field enhancement module to generate enhanced fused features. The enhanced fused features are classified by the detection head in the target detection model to determine the target category, and the enhanced fused features are regressed by the regression head in the detection head to determine the bounding box. The preset target is obtained by combining the target category and the bounding box.

[0083] In the embodiments of this application, multi-scale features refer to a set of feature maps with different spatial resolutions obtained by processing the same infrared image data at different levels. This set can simultaneously characterize the overview contour information and local detail information in the image, and can be used for subsequent detection of targets of different sizes.

[0084] Specifically, this step is a complete target detection process. The backbone network processes the input infrared image data through a staged, progressively deeper convolutional operation. Each stage outputs a feature map at that resolution, and downsampling reduces the resolution before moving to the next stage. Since infrared images lack rich texture, this process primarily extracts the target's thermal outline and shape information. The feature maps output from different stages, each with different resolutions, together constitute multi-scale features. Next, the connection layer processes these multi-scale features. This process is a top-down information fusion process, upsampling the feature maps generated by later stages (lower resolution, stronger semantic information). Here, "later stage" refers to the next stage in the process, which is deeper and has a lower output feature map resolution. This later stage is then fused with the feature map generated by the adjacent previous stage (higher resolution, stronger spatial information) to transfer deeper semantic information to shallower layers. Finally, a receptive field enhancement module performs multi-branch convolutional processing on the fused feature map to aggregate contextual information, thereby generating enhanced fused features. It's important to note that the receptive field enhancement module is a module that utilizes multiple convolutional branches with different receptive fields to process feature maps in parallel, and then aggregates the results from these branches to enhance the network's ability to perceive contextual information. This enhanced and fused feature is fed into the detection head for parsing. This parsing process performs two tasks in parallel: classification by its classification head to determine the target category of the identified object; and regression by its regression head to determine the precise bounding box of the object. Finally, the identified target category is combined with the corresponding bounding box to obtain the preset target present in the image.

[0085] For example, taking the identification of a person in an image as a preset target, the specific execution process of the target detection model in this application is explained as follows: When any frame of infrared image data is input into the backbone network, the backbone network generates a set of multi-scale features through its phased convolution and downsampling processing. In the lower-resolution deep feature maps, the backbone network can extract abstract semantic features that are highly sensitive to the overall contour and posture of the human body; while in the higher-resolution shallow feature maps, it retains the precise spatial details of the human body's thermal signal edges. Subsequently, the connection layer performs top-down information fusion on the multi-scale features, that is, the lower-resolution feature maps generated in later stages of the backbone network that contain high-level semantic information are combined with the precise edge details in the shallow feature maps layer by layer through upsampling. After processing by the receptive field enhancement module, a set of enhanced fusion features is finally generated. At this point, the feature map corresponding to the area where the person is located in the original image contains both high-level abstract features that can distinguish the target category and precise spatial features that can indicate the target contour. Finally, this enhanced fusion feature is sent to the detection head. After analyzing the features of the region, the classification head in the detection head matches them with the pre-learned personnel category criteria within the model and outputs a high-confidence personnel target category. Simultaneously, the regression head precisely calculates the bounding box coordinates (e.g., x, y, w, h) that tightly surround the person's thermal signal. In the bounding box coordinates, x represents the x-coordinate of the top-left corner vertex of the bounding box, defining the target's starting position in the horizontal direction in the image; y represents the y-coordinate of the top-left corner vertex of the bounding box, defining the target's starting position in the vertical direction in the image; w represents the width of the bounding box, defining the size of the area occupied by the target in the horizontal direction; and h represents the height of the bounding box, defining the size of the area occupied by the target in the vertical direction. By combining the personnel category label with these bounding box coordinates, the system obtains a pre-defined target that has been successfully identified and located.

[0086] S203: Based on the preset target, the importance of the current detection scene corresponding to the infrared image data is classified to obtain the attention level, and the attention level is converted into the corresponding quantitative attention parameter through the preset conversion rule.

[0087] For example, environmental parameters, including temperature, are acquired through sensors. Then, a pre-defined scene classification and grading system is used to perform hierarchical rule-based judgments on the identified pre-defined targets and their environments to determine a qualitative level of attention. This judgment process prioritizes checking whether there are targets belonging to the high-importance category, or whether their movement speed or thermal anomaly change rate exceeds a high-level threshold, or whether the ambient temperature exceeds the limit, etc. If these conditions are met, the target is determined to be at the first level. If not, the target is further judged under more lenient conditions to determine whether it meets the second level. If none of these conditions are met, the target is determined to be at the third level. Through a pre-defined mapping relationship (such as a lookup table), the determined first, second, or third level of attention is directly converted into a clear quantitative attention parameter that can be used for subsequent calculations.

[0088] In one possible implementation, based on a preset target, the importance of the current detection scene corresponding to the infrared image data is classified to obtain an attention level, and the attention level is converted into a corresponding quantitative attention parameter through a preset conversion rule. Specifically, this includes: acquiring environmental parameters collected by sensors pre-configured in the infrared thermal imager, including temperature; determining the attention level of the current detection scene based on the target type of the preset target and the environmental parameters through a preset scene classification and grading system; and converting the attention level into a quantitative attention parameter through a preset mapping relationship.

[0089] In the embodiments of this application, the scene classification and grading system refers to a set of pre-set logical rules for comprehensively judging the importance of the current scene. The rule set can give a graded and qualitative evaluation conclusion of the scene based on the attributes of the identified preset target and the environmental parameters in which it is located, such as judging the scene as high, medium and low attention levels.

[0090] Specifically, this step is a comprehensive evaluation process that integrates target detection results with environmental information and ultimately outputs a quantified control parameter. The current environmental parameters are collected in real time using a temperature sensor pre-configured in the infrared thermal imager. Then, the pre-detected targets (including their type and other information) from the previous step, along with the collected environmental parameters, are fed into a scene classification and grading system for processing. This system determines the importance of the current scene based on pre-defined rules (e.g., whether there are high-importance targets or whether the ambient temperature exceeds a safety threshold) and outputs a qualitative attention level (e.g., level one, level two). Finally, through a pre-defined mapping relationship (usually a lookup table), this qualitative attention level is directly converted into a specific numerical value that can be used for subsequent calculations—the quantified attention parameter; for example, level one is converted to the value 3.0.

[0091] In one possible implementation, based on the target type and environmental parameters of the preset targets, the attention level of the current detection scene is determined through a preset scene classification and grading system. Specifically, if a first target exists among the preset targets, and the first target belongs to a preset high importance category, or the movement speed of the first target exceeds a first speed threshold, or the thermal anomaly change rate of the first target exceeds a first change rate threshold, or the temperature exceeds a first temperature threshold, then the attention level is determined to be level one. If a first target does not exist among the preset targets, then it is determined whether a second target exists among the preset targets, and the second target belongs to a preset medium importance category, or the movement speed of the second target exceeds a second speed threshold, or the thermal anomaly change rate of the second target exceeds a second change rate threshold, where the second speed threshold is less than the first speed threshold and the second change rate threshold is less than the first change rate threshold. If such a target exists, then the attention level is determined to be level two. When neither the first nor the second target exists among the preset targets, the attention level is determined to be level three, wherein level one is higher than level two, and level two is higher than level three.

[0092] In the embodiments of this application, the high importance category refers to a set of categories that are pre-set according to specific application scenarios and used to classify preset targets with the highest priority. The existence of a target belonging to this category itself represents a high-attention event. For example, in security monitoring applications, the target type "personnel" can be preset as a high importance category; while in industrial inspection applications, "key circuit components" can be preset as a high importance category.

[0093] Specifically, this step is a hierarchical, rule-based comprehensive judgment process to determine the final attention level. The system performs a first round of highest-priority condition screening on all currently acquired preset targets, checking one by one whether there are preset targets that meet the first-level conditions. This judgment includes multiple parallel "OR" conditions: determining whether a target's target type belongs to a preset high-importance category; or calculating whether its movement speed exceeds a first speed threshold; or calculating whether its thermal anomaly change rate exceeds a first change rate threshold; in addition, it also checks whether the temperature in the environmental parameters exceeds a first temperature threshold. As long as the current scene meets any of the above conditions, the screening process immediately stops, and the final attention level is determined to be the first level. If the first round of screening does not meet any conditions, the system enters a second round of second-highest-priority condition screening, checking whether there are preset targets that meet the second-level conditions. The judgment rules are similar to the first round, but more lenient conditions are used, namely, belonging to a medium-importance category, or a movement speed exceeding a smaller second speed threshold, or a thermal anomaly change rate exceeding a smaller second change rate threshold. If any condition in the second round is met, the attention level is determined to be the second level. If the conditions in the first two rounds are not met, the system will set the attention level to the default third level. In this way, the importance of the scene can be accurately classified.

[0094] S204: Determine the target performance profile based on the quantified attention parameters. The target performance profile includes runtime parameters, including the target frame rate and the target resolution.

[0095] For example, device status data, including remaining battery power and temperature, is acquired through sensors. Simultaneously, a set of preset conservative power consumption mapping relationships and a set of priority performance mapping relationships are used to calculate two candidate performance profiles in parallel, prioritizing power saving and performance assurance, respectively, based on the same quantified attention parameter. The system judges based on the acquired device status data. When the device is in a resource-scarce state (such as low battery or high temperature), the aforementioned conservative candidate performance profile is selected as the final output; otherwise, the performance-priority candidate profile is selected. The total number of pixels processed per unit time corresponding to the conservative candidate profile (defined by the product of frame rate and resolution) is necessarily lower than that of the performance-priority candidate profile, thus ensuring the objectivity and effectiveness of the two strategy selections.

[0096] In one possible implementation, determining the target performance profile based on quantized attention parameters specifically includes: acquiring device status data of the infrared thermal imager, including remaining battery power and the temperature of the target imaging component; calculating a first candidate performance profile based on the quantized attention parameters using a preset conservative power consumption mapping relationship, the first candidate performance profile including first operating parameters and a first imaging quality threshold; calculating a second candidate performance profile based on the quantized attention parameters using a preset priority performance mapping relationship, the second candidate performance profile including second operating parameters and a second imaging quality threshold; selecting the first candidate performance profile as the target performance profile when the remaining battery power is lower than a preset battery power threshold and / or the temperature is higher than a preset temperature threshold; selecting the second candidate performance profile as the target performance profile when the remaining battery power is higher than the battery power threshold and the temperature is lower than the temperature threshold, wherein the total number of pixels processed per unit time corresponding to the operating parameters in the first candidate performance profile is lower than the total number of pixels processed per unit time corresponding to the operating parameters in the second candidate performance profile, and the total number of pixels processed per unit time is the product of the target frame rate and the target resolution.

[0097] In the embodiments of this application, the conservative power consumption mapping relationship refers to a set of preset decision rules aimed at prioritizing the device's battery life and operational stability. It is used to transform the input quantitative attention parameters into a performance profile with relatively conservative performance output. For example, even if the input quantitative attention parameters are high, the operating parameters (such as frame rate) in the performance profile output by the mapping relationship will be limited to a low level that can ensure the device's safe operation for a long time.

[0098] Specifically, this step is a performance decision-making process with dual-strategy optimization, characterized by device self-awareness of its state. By reading data from the battery management chip and temperature sensor, device state data representing the current resource status of the device is obtained, namely, the remaining battery power and the temperature of the target imaging component. The system executes two strategy calculations in parallel: on the one hand, using a conservative power consumption mapping relationship, a first candidate performance profile prioritizing power saving is calculated based on the current quantified attention parameters; on the other hand, simultaneously using a priority performance mapping relationship, a second candidate performance profile prioritizing performance is calculated based on the same quantified attention parameters. After the calculations are completed, the system enters the decision selection phase: determining whether the current device state data triggers a resource stress condition, i.e., whether the remaining battery power is below a power threshold and / or whether the temperature is above a temperature threshold. If triggered, the first candidate performance profile is selected as the final output target performance profile; otherwise, if the device state is normal, the second candidate performance profile is selected as the final output.

[0099] To more clearly illustrate the difference between the conservative power consumption mapping relationship and the priority performance mapping relationship, an exemplary lookup table is used below. This lookup table outputs the corresponding target performance profile (which includes operating parameters and target imaging quality thresholds) based on the input quantization attention parameters:

[0100] Quantitative attention parameters Conservative power consumption mapping relationship (first candidate) Priority performance mapping (second candidate) 1.0 (Low attention) {10fps, 320x240, SNR≥25dB} {15fps, 320x240, SNR≥28dB} 2.0 (Medium Attention Level) {15fps, 640x480, SNR≥30dB} {25fps, 640x480, SNR≥32dB} 3.0 (High Attention) {20fps, 640x480, SNR≥35dB} {30fps, 640x480, SNR≥38dB}

[0101] The quantified attention parameter serves as the input index for querying this table. It quantifies an abstract scene importance level (e.g., high, medium, low attention) into a specific numerical value (e.g., 3.0, 2.0, 1.0). The conservative power consumption mapping relationship defines the performance profile to be adopted for different attention parameters when the device is under resource pressure (e.g., low battery or high temperature). Its core strategy is to prioritize ensuring the device's battery life and stable operation at the expense of some performance. The priority performance mapping relationship defines the performance profile to be adopted for different attention parameters when the device is under resource pressure. Its core strategy is to fully utilize available resources and prioritize ensuring the best imaging performance. Each row represents the different output results given by the two mapping relationships for a specific "quantified attention parameter" input. The "target performance profile" in each cell is a composite data set containing two parts: runtime parameters: {target frame rate, target resolution}, which are the specific operating parameters to be configured on the hardware. Target imaging quality threshold: i.e. SNR (SIGNAL-NOISE RATIO) range, which is the minimum image quality that should be achieved for the above operating parameters, and it will be used in the closed-loop feedback control of the subsequent step S206.

[0102] Taking a "Quantitative Attention Parameter" of 3.0 (High Attention) as an example: If the device's battery is low, the system will choose the "Conservative" configuration in the second column, limiting performance to {20fps, 640x480, SNR≥35dB} to conserve power. If the device has sufficient battery power, the system will choose the "Priority Performance" configuration in the third column, boosting performance to {30fps, 640x480, SNR≥38dB} to obtain the best image. By comparing the "Total Pixels Processed Per Unit Time" (frame rate × resolution) and "Target Imaging Quality Threshold" in the two columns, it is clear that for any identical input, the performance and quality requirements of the second column (first candidate) are always lower than those of the third column (second candidate). This objectively reflects the inherent difference and design intent of the two mapping relationships.

[0103] S205: Based on the operating parameters, the minimum target power consumption value required to meet the operating parameters is obtained through a preset performance-power consumption mapping relationship.

[0104] In the embodiments of this application, the performance power consumption mapping relationship refers to a dataset or mathematical model that is established in advance through hardware calibration tests. It is used to accurately describe the actual power consumption value of the target imaging component when it works stably under different operating parameters (such as different combinations of frame rate and resolution). For example, the relationship records that when the target imaging component works with the set of operating parameters of "30fps frame rate and 640x480 resolution", the power consumption value required for its stable operation is 500mW.

[0105] Specifically, this step is a calculation process that queries the optimal power consumption target based on known performance requirements. Previous steps have already determined a set of specific operating parameters for the current scenario (e.g., the target frame rate is determined to be 30fps, and the target resolution is determined to be 640x480). The system uses this set of specific operating parameters as a query index to search or calculate within a preset performance-power consumption mapping relationship (usually implemented as a multi-dimensional lookup table or fitting function). This mapping relationship stores a large number of "performance-power consumption" corresponding data points obtained from actual testing. The query process aims to find the data point that perfectly matches or is closest to the currently required operating parameters and extract the corresponding power consumption value. This extracted power consumption value is the theoretical minimum power consumption required to stably meet the required operating parameters under the current hardware and environment. This value is then determined as the minimum target power consumption value and serves as the benchmark target for the next step of power consumption closed-loop adjustment.

[0106] S206: Based on the lowest target power consumption value and power consumption data, generate a power consumption adjustment signal, and adjust the power consumption of the target imaging component according to the power consumption adjustment signal.

[0107] For example, by dividing infrared image data into a signal region containing a preset target and a noise region not containing the target, and calculating the ratio of the average pixel value of the signal region to the standard deviation of the pixels in the noise region, a signal-to-noise ratio (SNR) is determined as a real-time imaging quality indicator. This imaging quality indicator is compared with a target imaging quality threshold obtained from a target performance profile: if the indicator is lower than the threshold, the system enters a performance-priority mode, generating a power-enhancing signal based on the quality difference between the two through a preset mapping relationship; conversely, if the indicator is not lower than the threshold, the system enters a power-optimization mode, generating a power-reducing signal based on the power error between the lowest target power consumption value and the real-time power consumption data; the power consumption of the target imaging component is precisely adjusted according to the generated enhanced or reduced signal.

[0108] In one possible implementation, a power consumption adjustment signal is generated based on the minimum target power consumption value and power consumption data, and the power consumption of the target imaging component is adjusted according to the power consumption adjustment signal. Specifically, this includes: determining the imaging quality index of the target imaging component based on the signal-to-noise ratio of the infrared image data; obtaining the target imaging quality threshold in the target performance configuration file; when the imaging quality index is lower than the target imaging quality threshold, calculating the quality difference between the imaging quality index and the target imaging quality threshold, and converting the quality difference into a power consumption enhancement signal through a preset mapping relationship; when the imaging quality index is not lower than the target imaging quality threshold, calculating the power consumption error based on the minimum target power consumption value and power consumption data, and generating a power consumption reduction signal based on the power consumption error; and adjusting the power consumption of the target imaging component according to the power consumption enhancement signal or the power consumption reduction signal.

[0109] In the embodiments of this application, the power consumption enhancement signal refers to a control command generated by the system when the real-time imaging quality does not meet the preset standard. This command is used to forcibly increase the actual operating power consumption of the target imaging component. Its strength is usually related to the difference in imaging quality. For example, when the calculated quality difference is large, the signal will instruct the power management unit to increase the component power consumption by a large margin in order to quickly improve the imaging quality.

[0110] Specifically, the system needs to acquire two key real-time feedback data: first, by calculating parameters such as the signal-to-noise ratio of the current infrared image data, a quantified imaging quality index is obtained; second, the target imaging quality threshold corresponding to the current operating parameters is directly read from the target performance configuration file determined in the previous step. The system then enters the core conditional judgment stage: comparing whether the current "imaging quality index" is lower than the "target imaging quality threshold." If it is lower, the system enters performance-priority mode, calculating the quality difference between the two and converting it into a power consumption enhancement signal of corresponding strength through a preset mapping relationship (such as a lookup table or function). Conversely, if the imaging quality meets or exceeds the standard, the system enters power consumption optimization mode. In this mode, the "minimum target power consumption value" is used as a benchmark, compared with the real-time "power consumption data," calculating the "power consumption error," and generating a power consumption reduction signal based on this error to reduce or maintain the current power consumption. Regardless of whether a power consumption enhancement signal or a power consumption reduction signal is generated, this signal will be used to precisely adjust the power consumption of the target imaging component.

[0111] In one possible implementation, the imaging quality index of the target imaging component is determined based on the signal-to-noise ratio (SNR) of the infrared image data. Specifically, this includes: identifying and dividing the infrared image data into signal regions and noise regions, where the signal region is the image region containing the preset target and the noise region is the image region not containing the preset target; calculating the arithmetic mean of the pixel values ​​of all pixels in the signal region to obtain the signal average; calculating the standard deviation of the pixel values ​​of all pixels in the noise region to obtain the noise standard deviation; dividing the signal average by the noise standard deviation to obtain the SNR, and using the SNR as the imaging quality index.

[0112] In the embodiments of this application, signal-to-noise ratio (SNR) refers to a standard indicator used to quantitatively measure image quality. It represents the ratio between the intensity of the effective signal (i.e., the thermal features of the preset target) in the image and the intensity of the background noise. For example, a high SNR value indicates that the thermal features of the target are clearly highlighted above the background noise, and the image quality is good; while a low SNR value indicates that the target is almost submerged in the background noise, and the image quality is poor.

[0113] Specifically, based on the location information (such as bounding boxes) of the preset targets identified in the preceding steps, the pixel regions in the current infrared image data are divided. Specifically, the region containing the preset targets is defined as the signal region, and a representative region in the background area that does not contain any targets is selected as the noise region. The system performs two parallel statistical calculations: first, it calculates the arithmetic mean of the pixel values ​​(representing thermal intensity) of all pixels within the signal region to obtain the signal average value characterizing the average intensity of the target signal; second, it calculates the standard deviation of the pixel values ​​of all pixels within the noise region to obtain the noise standard deviation characterizing the fluctuation amplitude of the background noise. By dividing the calculated signal average value by the noise standard deviation, the signal-to-noise ratio (SNR) of the current image is obtained, and this SNR calculation result is used as the final output, quantified imaging quality indicator.

[0114] In one possible implementation, the physical execution process of adjusting the power consumption of the target imaging component based on the power consumption adjustment signal is as follows: A processing unit within the power control platform (such as a Central Processing Unit (CPU) or Digital Signal Processing (DSP) technology) generates the power consumption enhancement signal or power consumption reduction signal, which is a digitized instruction. This instruction is sent to a power management unit (PMU). Based on this instruction, the PMU uses Dynamic Voltage and Frequency Scaling (DVFS) technology to change the core operating voltage and / or clock frequency supplied to the target imaging component (such as a signal processor) in real time. For example, the power consumption enhancement signal instructs the power management unit (PMU) in the infrared thermal imager to increase the voltage or frequency, while the power consumption reduction signal instructs the PMU to decrease the voltage or frequency, thereby achieving precise, physical adjustment of the actual operating power consumption.

[0115] Figure 3 This is a schematic diagram of a power consumption control system for an infrared thermal imager according to an embodiment of this application. This system can be implemented through software, hardware, or a combination of both, becoming all or part of the overall system. For example... Figure 3 As shown, the system includes: a data acquisition module 301, a target detection module 302, a scene transformation module 303, a performance decision module 304, a power consumption calculation module 305, and a power consumption adjustment module 306, wherein:

[0116] The data acquisition module 301 is configured to acquire infrared image data from the infrared thermal imager in real time, as well as power consumption data of the target imaging component in the infrared thermal imager.

[0117] Target detection module 302 is configured to perform feature analysis on infrared image data using a preset target detection model to obtain preset targets in the infrared image data;

[0118] The scene conversion module 303 is configured to classify the importance of the current detection scene corresponding to the infrared image data based on a preset target, obtain the attention level, and convert the attention level into the corresponding quantitative attention parameter through a preset conversion rule.

[0119] The performance decision module 304 is configured to determine the target performance profile based on the quantified attention parameters. The target performance profile includes running parameters, including the target frame rate and the target resolution.

[0120] The power consumption calculation module 305 is configured to obtain the minimum target power consumption value required to meet the operating parameters based on the operating parameters and through a preset performance-power consumption mapping relationship.

[0121] The power consumption adjustment module 306 is configured to generate a power consumption adjustment signal based on the lowest target power consumption value and power consumption data, and to adjust the power consumption of the target imaging component according to the power consumption adjustment signal.

[0122] Based on the above embodiments, as an optional embodiment, the target detection module 302 is further configured to:

[0123] Infrared image data is processed by multiple convolutional layers of the backbone network in the target detection model. Each stage of convolutional processing outputs a feature map of the corresponding resolution. The feature map output by the current stage is downsampled and then input into the next adjacent stage to obtain a set of feature maps consisting of all feature maps output by at least two stages. This set of feature maps is used as the multi-scale feature of the preset target. The feature map includes thermal contour and shape information.

[0124] Through the connection layer in the object detection model, the first feature map generated by the later stage of any two adjacent stages in the multi-scale feature is upsampled, and the upsampled first feature map is fused with the second feature map to obtain a fused feature map. The second feature map is generated by the earlier stage of any two adjacent stages. The fused feature map is processed by multi-branch convolution through a preset receptive field enhancement module to generate enhanced fused features.

[0125] The enhanced fusion features are classified using the detection head in the target detection model to determine the target category. The enhanced fusion features are then regressed using the regression head in the detection head to determine the bounding box. By combining the target category and the bounding box, the preset target is obtained.

[0126] Based on the above embodiments, as an optional embodiment, the scene conversion module 303 is further used for:

[0127] Acquire environmental parameters, including temperature, from sensors pre-configured in the infrared thermal imager.

[0128] Based on the target type and environmental parameters of the preset objectives, the attention level of the current detection scenario is determined through a preset scenario classification and grading system;

[0129] Through a pre-defined mapping relationship, the level of attention is converted into a quantitative parameter of attention.

[0130] Based on the above embodiments, as an optional embodiment, the scene conversion module 303 is further used for:

[0131] If there is a first target among the preset targets, the first target belongs to the preset high importance category, or the first target's movement speed exceeds the first speed threshold, or the first target's thermal anomaly change rate exceeds the first change rate threshold, or the temperature exceeds the first temperature threshold, then the attention level is determined to be the first level.

[0132] If the first target is not among the preset targets, then determine whether there is a second target among the preset targets. The second target belongs to the preset medium importance category, or the movement speed of the second target exceeds the second speed threshold, or the thermal anomaly change rate of the second target exceeds the second change rate threshold. The second speed threshold is less than the first speed threshold, and the second change rate threshold is less than the first change rate threshold. If they exist, then the attention level is determined to be the second level.

[0133] When there are no first or second objectives among the preset goals, the attention level will be set to the third level, where the first level is higher than the second level, and the second level is higher than the third level.

[0134] Based on the above embodiments, as an optional embodiment, the performance decision module 304 is further configured to:

[0135] Acquire the device status data of the infrared thermal imager, which includes the remaining battery power and the temperature of the target imaging components;

[0136] Based on the preset conservative power consumption mapping relationship and the quantized attention parameters, the first candidate performance profile is calculated. The first candidate performance profile includes the first operating parameters and the first imaging quality threshold.

[0137] Based on the preset priority performance mapping relationship and the quantified attention parameters, the second candidate performance profile is calculated. The second candidate performance profile includes the second operating parameters and the second imaging quality threshold.

[0138] When the remaining power is lower than the preset power threshold and / or the temperature is higher than the preset temperature threshold, the first candidate performance profile is selected as the target performance profile.

[0139] When the remaining battery power is higher than the battery power threshold and the temperature is lower than the temperature threshold, the second candidate performance profile is selected as the target performance profile. The total number of pixels processed per unit time corresponding to the running parameters in the first candidate performance profile is lower than the total number of pixels processed per unit time corresponding to the running parameters in the second candidate performance profile. The total number of pixels processed per unit time is the product of the target frame rate and the target resolution.

[0140] Based on the above embodiments, as an optional embodiment, the power consumption adjustment module 306 is further used for:

[0141] The imaging quality index of the target imaging component is determined based on the signal-to-noise ratio of the infrared image data;

[0142] Obtain the target imaging quality threshold from the target performance configuration file;

[0143] When the imaging quality index is lower than the target imaging quality threshold, the quality difference between the imaging quality index and the target imaging quality threshold is calculated, and the quality difference is converted into a power consumption enhancement signal through a preset mapping relationship.

[0144] When the imaging quality index is not lower than the target imaging quality threshold, the power consumption error is calculated based on the lowest target power consumption value and power consumption data, and a power consumption reduction signal is generated based on the power consumption error.

[0145] The power consumption of the target imaging component is adjusted based on the power consumption enhancement signal or the power consumption reduction signal.

[0146] Based on the above embodiments, as an optional embodiment, the power consumption adjustment module 306 is further used for:

[0147] In infrared image data, signal regions and noise regions are identified and divided. Signal regions are image regions that contain preset targets, while noise regions are image regions that do not contain preset targets.

[0148] Calculate the arithmetic mean of the pixel values ​​of all pixels within the signal region to obtain the signal average.

[0149] Calculate the standard deviation of pixel values ​​for all pixels within the noise region to obtain the noise standard deviation;

[0150] The signal-to-noise ratio (SNR) is obtained by dividing the average signal value by the noise standard deviation, and is used as an indicator of image quality.

[0151] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0152] This embodiment also discloses an electronic device, as shown in the reference. Figure 4 The electronic device may include: at least one processor 401, at least one communication bus 402, user interface 403, network interface 404, and at least one memory 405.

[0153] The communication bus 402 is used to enable communication between these components.

[0154] The user interface 403 may include a display screen and a camera. Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0155] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0156] The processor 401 may include one or more processing cores. The processor 401 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 405, and by calling data stored in memory 405. Optionally, the processor 401 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 401 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 401.

[0157] The memory 405 may include random access memory (RAM) or read-only memory. Optionally, the memory 405 may include a non-transitory computer-readable storage medium. The memory 405 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 405 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 405 may also be at least one storage device located remotely from the aforementioned processor 401. Figure 4 As shown, the memory 405, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a power consumption control method for an infrared thermal imager.

[0158] exist Figure 4 In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 401 can be used to call an application program stored in the memory 405 for a power consumption control method of an infrared thermal imager. When executed by one or more processors 401, the electronic device performs one or more methods as described in the above embodiments.

[0159] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0160] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0161] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0162] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0163] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0164] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 405 and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory 405 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0165] The above description is merely an exemplary embodiment of this application and should not be construed as limiting the scope of this application. Any equivalent changes and modifications made in accordance with the teachings of this application shall still fall within the scope of this application. Those skilled in the art will readily conceive of other embodiments of this application upon considering the disclosure of this specification. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not described in this application. The specification and embodiments are considered exemplary only, and the scope of disclosure of this application is defined by the claims.

Claims

1. A power consumption control method for an infrared thermal imager, characterized in that, The method, applied to a power consumption control platform, includes: In real time, acquire infrared image data from the infrared thermal imager, as well as power consumption data of the target imaging component in the infrared thermal imager; The infrared image data is analyzed for features using a preset target detection model to obtain the preset target in the infrared image data; Based on the preset target, the importance of the current detection scene corresponding to the infrared image data is classified to obtain the attention level, and the attention level is converted into the corresponding quantitative attention parameter through the preset conversion rule. Based on the quantified attention parameters, a target performance profile is determined. The target performance profile includes runtime parameters, including a target frame rate and a target resolution. Based on the operating parameters, the minimum target power consumption value required to meet the operating parameters is obtained through a preset performance-power consumption mapping relationship; Based on the minimum target power consumption value and the power consumption data, a power consumption adjustment signal is generated, and the power consumption of the target imaging component is adjusted according to the power consumption adjustment signal.

2. The method according to claim 1, characterized in that, The step of performing feature analysis on the infrared image data using a preset target detection model to obtain a preset target in the infrared image data specifically includes: The infrared image data is processed by multiple convolutional layers of the backbone network in the target detection model. Each stage of convolutional processing outputs a feature map of the corresponding resolution. The feature map output by the current stage is downsampled and then input into the next adjacent stage to obtain a set of feature maps composed of all feature maps output by at least two stages. The set of feature maps is used as the multi-scale feature of the preset target. The feature map includes thermal contour and shape information. Through the connection layer in the target detection model, the first feature map generated by the later stage of any two adjacent stages in the multi-scale feature is upsampled, and the upsampled first feature map is fused with the second feature map to obtain a fused feature map. The second feature map is generated by the earlier stage of any two adjacent stages. The fused feature map is processed by multi-branch convolution through a preset receptive field enhancement module to generate enhanced fused features. The enhanced fusion features are classified using the detection head in the target detection model to determine the target category. The enhanced fusion features are then regressed using the regression head in the detection head to determine the bounding box. The target category and the bounding box are then combined to obtain the preset target.

3. The method according to claim 1, characterized in that, Based on the preset target, the importance of the current detection scene corresponding to the infrared image data is classified to obtain an attention level, and the attention level is converted into a corresponding quantitative attention parameter through a preset conversion rule, specifically including: The environmental parameters, including temperature, are acquired by sensors pre-configured in the infrared thermal imager. Based on the target type of the preset target and the environmental parameters, the attention level of the current detection scene is determined through a preset scene classification and grading system; The attention level is converted into the quantitative attention parameter through a preset mapping relationship.

4. The method according to claim 3, characterized in that, The determination of the attention level of the current detection scene based on the target type of the preset target and the environmental parameters, through a preset scene classification and grading system, specifically includes: If a first target exists among the preset targets, and the first target belongs to a preset high importance category, or the movement speed of the first target exceeds a first speed threshold, or the thermal anomaly change rate of the first target exceeds a first change rate threshold, or the temperature exceeds a first temperature threshold, then the attention level is determined to be the first level. If the first target is not among the preset targets, then determine whether there is a second target among the preset targets. The second target belongs to a preset medium importance category, or the movement speed of the second target exceeds a second speed threshold, or the thermal anomaly change rate of the second target exceeds a second change rate threshold. The second speed threshold is less than the first speed threshold, and the second change rate threshold is less than the first change rate threshold. If they exist, then the attention level is determined to be the second level. When neither the first target nor the second target exists among the preset targets, the attention level is determined to be the third level, wherein the first level is higher than the second level, and the second level is higher than the third level.

5. The method according to claim 1, characterized in that, The step of determining the target performance configuration file based on the quantified attention parameters specifically includes: Acquire the device status data of the infrared thermal imager, the device status data including the remaining battery power and the temperature of the target imaging component; Based on the preset conservative power consumption mapping relationship and the quantized attention parameters, a first candidate performance profile is calculated. The first candidate performance profile includes a first operating parameter and a first imaging quality threshold. Based on the preset priority performance mapping relationship and the quantized attention parameters, a second candidate performance profile is calculated. The second candidate performance profile includes a second operating parameter and a second imaging quality threshold. When the remaining power is lower than a preset power threshold and / or the temperature is higher than a preset temperature threshold, the first candidate performance profile is selected as the target performance profile. When the remaining battery power is higher than the battery power threshold and the temperature is lower than the temperature threshold, the second candidate performance profile is selected as the target performance profile. The total number of pixels processed per unit time corresponding to the running parameters in the first candidate performance profile is lower than the total number of pixels processed per unit time corresponding to the running parameters in the second candidate performance profile. The total number of pixels processed per unit time is the product of the target frame rate and the target resolution.

6. The method according to claim 1, characterized in that, The step of generating a power consumption adjustment signal based on the minimum target power consumption value and the power consumption data, and adjusting the power consumption of the target imaging component according to the power consumption adjustment signal, specifically includes: The imaging quality index of the target imaging component is determined based on the signal-to-noise ratio of the infrared image data; Obtain the target imaging quality threshold from the target performance configuration file; When the imaging quality index is lower than the target imaging quality threshold, the quality difference between the imaging quality index and the target imaging quality threshold is calculated, and the quality difference is converted into a power consumption enhancement signal through a preset mapping relationship. When the imaging quality index is not lower than the target imaging quality threshold, the power consumption error is calculated based on the minimum target power consumption value and the power consumption data, and a power consumption reduction signal is generated based on the power consumption error. The power consumption of the target imaging component is adjusted according to the power consumption enhancement signal or the power consumption reduction signal.

7. The method according to claim 6, characterized in that, The step of determining the imaging quality index of the target imaging component based on the signal-to-noise ratio of the infrared image data specifically includes: In the infrared image data, signal regions and noise regions are identified and divided. The signal region is the image region containing the preset target, and the noise region is the image region that does not contain the preset target. Calculate the arithmetic mean of the pixel values ​​of all pixels within the signal region to obtain the signal average value; Calculate the standard deviation of pixel values ​​for all pixels within the noise region to obtain the noise standard deviation; The signal-to-noise ratio (SNR) is obtained by dividing the average signal value by the noise standard deviation, and the SNR is used as the imaging quality index.

8. A power consumption control system for an infrared thermal imager, characterized in that, The system includes a data acquisition module, a target detection module, a scene transformation module, a performance decision module, a power consumption calculation module, and a power consumption adjustment module, among which: The data acquisition module is configured to acquire infrared image data from the infrared thermal imager in real time, as well as power consumption data of the target imaging component in the infrared thermal imager. The target detection module is configured to perform feature analysis on the infrared image data using a preset target detection model to obtain a preset target in the infrared image data. The scene conversion module is configured to classify the importance of the current detection scene corresponding to the infrared image data based on the preset target, obtain the attention level, and convert the attention level into the corresponding quantitative attention parameter through a preset conversion rule. The performance decision module is configured to determine a target performance profile based on the quantified attention parameters. The target performance profile includes running parameters, including a target frame rate and a target resolution. The power consumption calculation module is configured to obtain the minimum target power consumption value required to meet the operating parameters based on the operating parameters and through a preset performance-power consumption mapping relationship. The power consumption adjustment module is configured to generate a power consumption adjustment signal based on the minimum target power consumption value and the power consumption data, and to adjust the power consumption of the target imaging component according to the power consumption adjustment signal.

9. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-7.

Images