Low-power intelligent camera control method and device based on scene adaptive adjustment

Abstract

This invention relates to the field of low-power camera technology, and more particularly to a low-power intelligent camera control method and device based on scene adaptive adjustment. The method includes the following steps: continuously capturing images of a shooting area using an intelligent camera to obtain an image frame sequence; extracting brightness information from the image frame sequence to generate an ambient light change index set for multiple time windows; dynamically adjusting parameters based on the ambient light change index set to output exposure adjustment parameters; detecting scene elements based on the image frame sequence and calculating motion indicators to obtain a scene dynamic feature set; and classifying the scene level based on the scene dynamic feature set to obtain a scene complexity level. This invention effectively reduces device power consumption and computing resource usage while ensuring continuous image quality and target recognition accuracy, thereby extending device battery life.

Description

Technical Field

[0001] This invention relates to the field of low-power cameras, and more particularly to a low-power intelligent camera control method and device based on scene adaptive adjustment. Background Technology

[0002] With the continuous development of video capture technology and embedded intelligent devices, intelligent cameras have been widely used in various fields such as security monitoring, motion recording, live streaming, unmanned equipment sensing, and industrial vision acquisition. Especially against the backdrop of the rapid popularization of mobile shooting devices and wearable devices, intelligent cameras are gradually evolving from fixed scene monitoring to continuous shooting and dynamic scene recording, which places higher demands on their real-time performance, image quality stability, and battery life.

[0003] In actual continuous shooting, significant differences exist between various scenarios, such as static meeting scenes, low-light indoor scenes, and high-dynamic outdoor sports scenes, all exhibiting marked differences in the frequency of light changes, target motion density, and scene complexity. These differences directly lead to varying degrees of computational load fluctuations for cameras across multiple processing stages, including image acquisition, automatic exposure adjustment, video encoding, and image enhancement. In traditional fixed-parameter shooting modes, cameras typically employ uniform frame rates, resolutions, and encoding strategies, making dynamic adjustments based on scene changes difficult. This can easily result in resource waste in low-complexity scenes and image quality degradation or processing delays in high-complexity scenes. Existing cameras often rely on fixed strategies or simple brightness feedback mechanisms for exposure control. When the lighting environment changes rapidly, frequent exposure adjustments or response lags can occur, leading to flickering, overexposure, or underexposure. In video encoding and image processing, more complex scenes significantly increase the computational overhead of motion estimation, inter-frame prediction, and image enhancement algorithms, further exacerbating device power consumption, increasing system energy consumption, and severely impacting the battery life of mobile devices or those used for extended shooting sessions. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a low-power intelligent camera control method and device based on scene adaptive adjustment, thereby solving at least one of the aforementioned technical problems.

[0005] To achieve the above objectives, the present invention provides a low-power intelligent camera control method based on scene adaptive adjustment, comprising the following steps: Step S1: Continuously capture images of the shooting area using a smart camera to obtain an image frame sequence; Step S2: Extract brightness information from the image frame sequence to generate a set of ambient light change indicators for multiple time windows; Step S3: Based on the ambient light change index set, perform dynamic parameter adjustment and output exposure adjustment parameters; Step S4: Detect scene elements based on the image frame sequence and calculate motion indicators to obtain the scene dynamic feature set; Step S5: Classify the scene level based on the scene dynamic feature set to obtain the scene complexity level; Step S6: Perform adaptive configuration based on scene complexity level, and perform low-power automatic shooting processing based on exposure adjustment parameters.

[0006] This specification provides a low-power intelligent camera control device based on scene adaptive adjustment, used to execute the low-power intelligent camera control method based on scene adaptive adjustment as described above, including: The shooting module is used to continuously capture images of the shooting area using a smart camera to obtain a sequence of image frames; The brightness extraction module is used to extract brightness information from the image frame sequence and generate a set of ambient light change indicators for multiple time windows. The exposure adjustment module is used to dynamically adjust parameters based on the ambient light change index set and output exposure adjustment parameters. The element detection module is used to detect scene elements based on image frame sequences and calculate motion indicators to obtain a dynamic feature set of the scene. The scene segmentation module is used to classify scenes into levels based on dynamic feature sets to obtain scene complexity levels. The adaptive configuration module is used to perform adaptive configuration based on the scene complexity level and to perform low-power automatic shooting processing based on exposure adjustment parameters.

[0007] The specific benefits of this invention are as follows: By continuously acquiring image frame sequences of the shooting area through an intelligent camera, real-time scene change information within the monitoring area can be completely recorded, avoiding the problem of missing key images caused by traditional intermittent acquisition; simultaneously, a unified temporal image data foundation is formed, providing reliable data support for ambient light analysis, target motion analysis, and shooting parameter adjustment, improving the continuity and stability of the monitoring process. By extracting brightness information from the image frame sequence and constructing an ambient light change index set for multiple time windows, the change law and fluctuation characteristics of ambient light can be accurately reflected, effectively distinguishing between short-term interference and real light changes, reducing the impact of abnormal brightness fluctuations on camera control strategies, and improving the accuracy of light state recognition. By dynamically generating exposure adjustment parameters based on the ambient light change index set, the camera's exposure time, gain, and white balance parameters can be matched with the current light state, reducing image overexposure or underexposure caused by sudden changes in light, while avoiding system resource consumption caused by frequent adjustments of exposure parameters, improving image quality and operating efficiency.

[0008] By detecting scene elements and calculating motion indicators, the system can accurately identify active targets and their motion states within a scene, quantifying the number of targets, motion intensity, and trends. This enables the camera to perceive the level of activity in the scene, improving the analysis accuracy and dynamic response capabilities for complex monitoring scenarios. By classifying the scene's dynamic feature set into scene levels, complex scene change information can be transformed into quantifiable complexity levels, achieving standardized descriptions of different scene states. This facilitates the establishment of unified control decision-making criteria and improves the targeting and adaptability of low-power control strategies. Through adaptive configuration combining scene complexity levels and exposure adjustment parameters, the system can dynamically adjust the allocation of shooting resources and the intensity of encoding processing based on the level of scene activity. This ensures the monitoring quality of critical scenes while reducing unnecessary computational load and energy consumption, enabling low-power operation of the intelligent camera during continuous shooting. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the steps of a low-power intelligent camera control method based on scene adaptive adjustment according to the present invention. Figure 2 This is a detailed flowchart illustrating the implementation steps of step S1. Figure 3 This is a flowchart illustrating the detailed implementation steps of step S2. Detailed Implementation

[0010] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0011] This application provides a low-power intelligent camera control method and apparatus based on scene adaptive adjustment. The execution entities of the low-power intelligent camera control method and apparatus based on scene adaptive adjustment include, but are not limited to, mechanical equipment, data processing platforms, cloud server nodes, network upload devices, etc., which can be considered as general computing nodes in this application. The data processing platform includes, but is not limited to, at least one of an audio-visual management system, an information management system, and a cloud data management system.

[0012] Please see Figures 1 to 3 This invention provides a low-power intelligent camera control method based on scene adaptive adjustment, comprising the following steps: Step S1: Continuously capture images of the shooting area using a smart camera to obtain an image frame sequence; Step S2: Extract brightness information from the image frame sequence to generate a set of ambient light change indicators for multiple time windows; Step S3: Based on the ambient light change index set, perform dynamic parameter adjustment and output exposure adjustment parameters; Step S4: Detect scene elements based on the image frame sequence and calculate motion indicators to obtain the scene dynamic feature set; Step S5: Classify the scene level based on the scene dynamic feature set to obtain the scene complexity level; Step S6: Perform adaptive configuration based on scene complexity level, and perform low-power automatic shooting processing based on exposure adjustment parameters.

[0013] In the embodiments of the present invention, see Figure 1 This is a flowchart illustrating the steps of a low-power intelligent camera control method based on scene adaptive adjustment according to the present invention. In this example, the steps of the low-power intelligent camera control method based on scene adaptive adjustment include: Step S1: Continuously capture images of the shooting area using a smart camera to obtain an image frame sequence; Step S2: Extract brightness information from the image frame sequence to generate a set of ambient light change indicators for multiple time windows; Step S3: Based on the ambient light change index set, perform dynamic parameter adjustment and output exposure adjustment parameters; Step S4: Detect scene elements based on the image frame sequence and calculate motion indicators to obtain the scene dynamic feature set; Step S5: Classify the scene level based on the scene dynamic feature set to obtain the scene complexity level; Step S6: Perform adaptive configuration based on scene complexity level, and perform low-power automatic shooting processing based on exposure adjustment parameters.

[0014] In this embodiment, a smart camera continuously captures video of the target shooting area to form a basic image data stream, which is then parsed into an image frame sequence in chronological order. During the shooting process, the image sensor continuously outputs raw image data at a fixed frame rate of 30fps to ensure the continuous representation of people, vehicles, and other dynamic targets in the moving scene. During continuous acquisition, each frame contains corresponding timestamp information and exposure status information, thereby constructing a complete time series structure. Due to interference factors such as light fluctuations, slight equipment shaking, and sensor noise in the actual shooting environment, the original image information is simultaneously retained during the frame sequence formation process for brightness analysis and dynamic feature extraction. The image frame sequence is arranged in chronological order, and inter-frame relationships are established at the data structure level, enabling precise tracking of changes between adjacent frames.

[0015] Each frame of the image undergoes grayscale conversion, and the average brightness value is calculated. The image is further divided into central and edge regions to calculate local brightness distribution values, reflecting the spatial non-uniformity of illumination. Simultaneously, the overall brightness statistics and ambient light sensor data are combined to generate ambient light intensity values, improving the robustness of illumination estimation. At an analysis frame rate of 10fps, consecutive image frames are divided into time windows, with each window consisting of 3 to 5 frames. Within each window, brightness information is statistically calculated to obtain indicators such as the rate of change of light intensity, the amplitude of light change, and the frequency of light fluctuations. Through statistical analysis of multiple sliding windows, a set of ambient light change indicators for multiple time windows is formed, ensuring a stable representation of illumination changes over time.

[0016] An exposure adjustment threshold system is established, including thresholds for the rate of change of light intensity, the amplitude of change of light intensity, and the frequency of light fluctuation. These thresholds are then compared with ambient light change indicators within each time window. When the rate of change of light intensity, the amplitude of change of light intensity, and the frequency of light fluctuation all exceed their respective thresholds, a sudden change in lighting is identified, and the exposure parameters are dynamically corrected based on the trend. Exposure time, gain parameters, and white balance parameters are adjusted simultaneously to adapt to rapidly changing lighting conditions. When any one of these indicators does not exceed its corresponding threshold, the lighting is considered stable, and the current exposure parameters are maintained, thus avoiding image fluctuations and additional computational overhead caused by frequent adjustments.

[0017] For each frame of the image, target detection processing is performed to identify moving targets such as people, vehicles, and animals, and their spatial positions are marked. Then, a multi-frame tracking method based on optical flow is used to continuously track the identified targets, acquiring their positional change trajectories between consecutive frames. Under a 10fps analysis frame rate, the displacement of the target between adjacent frames is calculated to obtain the target's motion velocity; the number of motion direction changes is obtained by calculating the angle of trajectory direction change; and the motion trajectory change frequency is obtained by statistically analyzing the trajectory curvature change. Simultaneously, the number of targets in the scene is counted, and density analysis is performed based on the spatial distribution of targets to obtain element quantity distribution information. Finally, the target quantity distribution information is fused with various motion indicators to form a dynamic feature set of the scene.

[0018] The scene complexity is calculated by weighted fusion of dynamic feature sets, and then classified into levels based on this complexity. The distribution of target quantity, target speed, number of changes in direction, and frequency of trajectory changes are normalized to map their values ​​to a uniform range of 0 to 1. Each feature is then assigned a weight coefficient: target quantity distribution (0.30), target speed (0.30), number of changes in direction (0.20), and frequency of trajectory changes (0.20). The scene complexity is obtained by weighted summation. Based on this, a sliding window is used to statistically analyze the complexity of consecutive frames, with each 3 to 5 frames constituting a time window, and the average complexity within the window is calculated. Scenes with an average complexity less than 0.40 are classified as low-complexity scenes, those between 0.40 and 0.70 as medium-complexity scenes, and those greater than or equal to 0.70 as high-complexity scenes, thus achieving scene level classification.

[0019] The shooting system parameters are adaptively configured based on the scene complexity level, and low-power shooting control is implemented in conjunction with exposure adjustment parameters. In medium-complexity scenes, standard shooting configuration parameters are used, including standard dynamic resolution, standard shooting frame rate of 30fps, standard encoding bitrate, and standard image processing intensity, to ensure a balance between image quality and power consumption. In low-complexity scenes, dynamic resolution, encoding bitrate, and image processing intensity are reduced, while maintaining a low-load operation of 10fps in the analysis link to reduce computational power consumption. In high-complexity scenes, dynamic resolution is increased, a high frame rate of 30fps is maintained, the encoding bitrate is increased, and image processing intensity is enhanced to ensure the integrity of motion details. Exposure adjustment parameters are jointly driven by the adaptive configuration parameters, causing exposure time, gain parameters, and white balance parameters to be adjusted synchronously with the encoding and processing strategies, thereby achieving a unified control effect of low-power automatic shooting and stable image quality output under different scene conditions.

[0020] In one specific embodiment, the intelligent camera is deployed in a fixed manner in the target shooting area, continuously shooting the shooting area at a capture rate of 30 frames per second and a resolution of 1920×1080, and outputting the raw video stream. Image enhancement processing is performed on the raw video stream: the first stage performs high-frequency noise filtering, applying a Gaussian low-pass filter to each frame, with a filter kernel size of 5×5 and a standard deviation σ=1.2; the second stage performs temporal smoothing processing, using three consecutive frames as a smoothing window, with the center frame weighted at 0.5 and the preceding and following frames weighted at 0.25 respectively, and the smoothed frames are calculated according to the following weighted formula: ; in Indicates the first The pixel brightness matrix of the frame image. After image enhancement, the enhanced video stream is obtained. The preset decomposition frame rate is 15 frames / second. The enhanced video stream is decomposed by extracting one frame at uniform intervals from every other frame, resulting in a sequence of image frames arranged according to their temporal numbers. The time interval between adjacent frames is fixed at 1. Second.

[0021] Brightness information is extracted frame by frame from the image frame sequence. For the first... Frame Image Convert the RGB color space to luminance components and calculate the average luminance value of the image using the following formula. : ; in , , , , These represent the red, green, and blue channel values ​​of the corresponding pixel, ranging from 0 to 255. The image is divided into 16 4×4 local blocks, and the average brightness of each block is calculated to obtain the local brightness distribution value. , Simultaneously read the exposure feedback parameters of the camera's image sensor to obtain the ambient light intensity value. .

[0022] The brightness change index between image frames is calculated using the difference between adjacent frames. (The text abruptly ends here, likely due to an incomplete sentence or missing information.) Frame and the For example, a frame: ; ; ; in To determine the statistical time window length, it is set to 10 seconds, meaning the window contains 150 frames; the sign flip count refers to the number of times the sign of adjacent difference values ​​changes. The time window length is used as a guideline. The above three indicators are statistically analyzed using a sliding window method, and a set of average values ​​is output for each window, generating a set of ambient light change indicators for multiple time windows. .

[0023] Taking a specific time window as an example, the calculation yields: Brightness unit: per second Unit of brightness times per second.

[0024] Three preset exposure adjustment thresholds: light intensity change rate threshold Brightness unit / second, threshold for light change amplitude Brightness unit, light fluctuation frequency threshold times per second.

[0025] Compare the above thresholds with the ambient light change index set obtained in step S2 item by item: ; ; ; If all three indicators exceed their corresponding thresholds, the current scene is determined to be in a state of sudden illumination change. A light change trend analysis is performed on the ambient light change indicator set, by analyzing the light change within the window... The sequence is fitted with a first-order linear regression to obtain the trend slope. Brightness unit per second, indicating a continuous brightening trend. Based on this brightening trend, the exposure adjustment parameters are dynamically adjusted according to the following rules: the exposure time is adjusted from the current value... shortened to ; Change the gain parameter from Reduced to The white balance parameters were adjusted, with the color temperature changed from 5500K to 6000K, and the output exposure was adjusted accordingly. .

[0026] If any of the three indicators does not exceed the corresponding threshold, the current exposure parameter is locked and the current parameter is directly output as the exposure adjustment parameter.

[0027] Scene element detection was performed based on the image frame sequence, and people, vehicles, and animals in the scene were marked. At this moment, a total of 7 scene elements were detected, including 5 people and 2 vehicles. Dynamic optical flow tracing was performed on the 7 scene elements, and the position change trajectory of each element between adjacent frames was extracted using the Lucas-Kanade optical flow method to obtain the displacement vector sequence of each element.

[0028] Count the number of target elements in the scene The spatial distribution of the seven elements in the image was analyzed. The 1920×1080 image was divided into nine 3×3 blocks, and the number of elements in each block was counted to obtain the element distribution information.

[0029] Calculate the element motion index based on the position change trajectory, with the first... Taking one element as an example: ; in , This represents the number of pixels the element has shifted horizontally and vertically between adjacent frames. Seconds. The average target velocity of the 7 elements is calculated as follows: Pixels per second; Average number of changes in motion direction is times / window; the average frequency of motion trajectory changes is Hours per second. Combining element quantity distribution information with element motion indicators. By fusing the features, a dynamic feature set of the scene is obtained. .

[0030] Set the weights for each feature term in the calculation: target quantity weights Target motion speed weight Weight of the number of changes in direction of motion Weight of frequency of motion trajectory change After normalizing the scene's dynamic feature set, a weighted calculation is performed to obtain the scene complexity score for the current frame. : ; in , , , This represents the score for each feature item after normalization to the (0, 100) interval. Substituting the values ​​from this example, the scores for each item after normalization are as follows: , , , ,but: ; Set the sliding calculation window length to 10 seconds, or 150 frames, and take the average complexity score of all frames within the window: ; According to the following grading standards Make a judgment: For low complexity, Medium complexity This is highly complex. In this embodiment... If the value falls within the range (40, 70), the scene complexity level is determined to be medium complexity.

[0031] Preset standard shooting configuration parameters: standard dynamic resolution is 1920×1080, standard shooting frame rate is 15 frames / second, standard video encoding bitrate is 4Mbps, and standard image processing intensity is base level 100%.

[0032] In this embodiment, the scene complexity level is medium, and the above standard shooting configuration parameters are directly used as adaptive configuration parameters to perform standard shooting processing.

[0033] If the scene complexity level is determined to be low complexity (i.e.) If the dynamic resolution is reduced to 1280×720, the shooting frame rate is reduced to 8 frames / second, the video encoding bitrate is reduced to 2Mbps, the image processing intensity is reduced to 60%, and low-power adaptive configuration parameters are generated.

[0034] If the scene complexity level is determined to be high complexity (i.e.) This increases the dynamic resolution to 2560×1440, the shooting frame rate to 30 frames / second, the video encoding bitrate to 8Mbps, the image processing intensity to 150%, and generates full-effect adaptive configuration parameters.

[0035] The adaptive configuration parameters output in this embodiment Resolution 1920×1080, frame rate 15 frames / second, bitrate 4Mbps, processing intensity 100%. The exposure adjustment parameters output in step S3 The commands are jointly distributed to the various functional control modules of the camera, driving the camera to perform low-power automatic shooting processing with standard power consumption configuration in the current medium-complexity scene, so as to achieve the goal of scene-adaptive power consumption optimization control.

[0036] In this embodiment, see Figure 2 The diagram below illustrates the detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include: The raw video stream is obtained by continuously shooting the shooting area using a smart camera; The original video stream is enhanced to obtain an enhanced video stream; the image enhancement specifically involves performing high-frequency noise filtering and temporal smoothing to eliminate high-frequency components of sensor jitter and environmental interference.

[0037] Set the decomposition frame rate; decompose the enhanced video stream into multiple frames based on the decomposition frame rate to obtain an image frame sequence.

[0038] In this embodiment, the camera operates in continuous recording mode, maintaining fixed video output parameters to ensure complete recording of motion processes in the activity scene. Preferably, the video recording frame rate is maintained at 30fps, allowing continuous capture of human movement, object displacement, and scene changes. The image sensor continuously outputs video data according to set exposure parameters, while assigning corresponding time stamp information to each frame to establish a complete time series relationship. Since the raw video stream originates directly from the image sensor, the image data contains not only real scene information but also interference information caused by ambient light fluctuations, device thermal noise, circuit random noise, and minor device vibrations. For example, in indoor lighting environments, AC-powered light sources may produce periodic brightness fluctuations; in outdoor scenes, cloud cover and changes in reflected light may cause local brightness changes; during handheld shooting, slight shaking may cause high-frequency changes in image details. These interference information are superimposed on the raw video stream, causing discrepancies between image brightness statistics and motion analysis results. By continuously acquiring the raw video stream, a complete video data foundation is established, providing a data source for image quality optimization and scene feature extraction, while ensuring that the camera maintains complete video recording capabilities during low-power control.

[0039] High-frequency noise filtering is applied to consecutive image frames in the original video stream. By analyzing the spatial distribution characteristics between image pixels, high-frequency regions with abnormal grayscale changes are identified, and random noise points and non-realistic texture changes are suppressed, thereby reducing the impact of thermal noise and circuit noise from the image sensor. After high-frequency noise filtering, isolated noise points and local high-frequency interference in the image are effectively weakened, while image edges and subject contours remain clear. A temporal analysis window of 3 to 5 consecutive images is constructed, and the brightness changes of each image frame within the window are smoothed to suppress abnormal brightness fluctuations that occur within a short period. For example, slight shaking during handheld shooting, instantaneous reflections in the environment, or sudden brightness changes caused by partial occlusion usually manifest as short-period changes within the time window. Temporal smoothing can effectively reduce the impact of such interference. During the smoothing process, the continuous change characteristics formed by real moving targets are preserved to avoid loss of target motion information due to over-smoothing. After high-frequency noise filtering and temporal smoothing, an enhanced video stream is obtained. The enhanced video stream exhibits a more stable brightness distribution, significantly reduced random noise, and improved image continuity. This allows for a more accurate reflection of the actual scene conditions, reduces invalid calculations caused by noise, and improves the accuracy of parameter analysis during the camera's low-power control process.

[0040] The decomposition frame rate is used to determine the number of images involved in scene analysis. Its purpose is to reduce the size of the analysis data while preserving the scene's changing characteristics. The camera recording frame rate remains constant at 30fps, and the decomposition frame rate is preferably set to 10fps. The decomposition frame rate only affects the image analysis process and does not change the original video recording quality or video output parameters. The enhanced video stream is periodically sampled in chronological order. When the recording frame rate is 30fps and the decomposition frame rate is 10fps, one frame is extracted as the analysis frame every three frames. This method reduces the number of images involved in the analysis while maintaining the integrity of the scene's dynamic change information. After sampling, the extracted analysis frames are arranged in chronological order to form an image frame sequence, and a correspondence between image frames and time information is established. For each image frame, the exposure state, brightness state, and environmental parameter information corresponding to the acquisition time are recorded, so that the image frame sequence contains not only image content but also the corresponding operational state data. For example, in one second of continuous video data, 30 original images are decomposed into 10 analysis images, reducing the amount of data involved in the analysis by about two-thirds while preserving the overall trend of scene changes. By constructing image frame sequences, the changes in a scene over time can be accurately described, providing a unified data foundation for brightness change analysis, illumination change rate calculation, target motion trajectory extraction, and scene complexity assessment. At the same time, it effectively reduces computational resource consumption and achieves the data preprocessing goal of low-power control.

[0041] In this embodiment, see Figure 3 The diagram below illustrates the detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Brightness information is extracted from the image frame sequence to obtain brightness information for multiple frames; the brightness information includes the average brightness value, local brightness distribution value, and ambient light intensity value of the image. Based on the brightness information, continuous difference calculation is performed between adjacent frames to obtain the brightness change index between image frames; the brightness change index between image frames includes the light intensity change rate, the light intensity change amplitude, and the illumination fluctuation frequency. The brightness change index between image frames is statistically analyzed using time-series windows to generate a set of ambient light change indexes for multiple time windows.

[0042] In this embodiment, pixel-level brightness calculation is performed on a single-frame image. By performing a brightness space transformation on the RGB image, the color image is mapped to a grayscale brightness distribution map to obtain the brightness value of each pixel. Statistical analysis is performed on the brightness values ​​of the entire frame to calculate the average brightness value, which characterizes the overall illumination intensity level of the current frame. The image is divided into multiple local analysis regions, such as the central region, edge regions, and adaptive regions of interest, and the local brightness distribution value within each region is calculated to characterize the non-uniformity of illumination in the spatial dimension. An ambient light intensity value is introduced as an auxiliary feature. This value is obtained by fusing the overall image brightness statistics with data collected by an ambient light sensor to enhance the stability and accuracy of illumination estimation. Each frame image generates corresponding brightness information data including the average image brightness value, local brightness distribution value, and ambient light intensity value, thus forming a multi-frame image brightness information set corresponding to the time series. Based on a time-sequential sequence of image frames, the average brightness values ​​of adjacent frames are differentially analyzed to obtain the rate of change in light intensity, which characterizes how quickly the overall brightness changes per unit time. Differential calculations are also performed on the local brightness distribution values ​​of adjacent frames to obtain the amplitude of light intensity changes, which characterizes the degree of spatial variation in illumination. The periodic fluctuation characteristics of brightness changes are statistically analyzed within a continuous time window to calculate the illumination fluctuation frequency, which characterizes the periodic stability of ambient light source changes, such as periodic flickering caused by indoor AC light or intermittent changes caused by outdoor shading. During the differential calculation process, continuous frame analysis is performed based on a decomposition frame rate of 10fps, meaning 10 frames are analyzed per second. Frame-by-frame differential analysis between adjacent frames can accurately reflect the trend of illumination changes. Through these calculation methods, a set of inter-frame brightness change indicators, including the rate of change in light intensity, the amplitude of light intensity changes, and the illumination fluctuation frequency, is formed, thereby achieving a quantitative description of the dynamic changes in ambient light.

[0043] The brightness variation index between image frames is divided into multiple consecutive time windows according to time sequence. Each time window contains several consecutive image frames, for example, 3 to 5 frames per time analysis window, to smooth the impact of short-term random fluctuations. Within each time window, the rate of change of light intensity, the amplitude of change of light intensity, and the frequency of light fluctuations are statistically calculated, including mean calculation, variance calculation, and trend analysis, to obtain the stability and fluctuation characteristics within that time window. A multi-window sliding analysis method is used to maintain partial overlap between adjacent time windows, thereby enhancing the ability to express temporal continuity and avoiding misjudgments caused by abrupt changes in a single window. The statistical results between different time windows are compared and analyzed to determine the persistence and trend characteristics of ambient light changes, such as continuous enhancement, continuous weakening, or periodic fluctuations. The statistical results of each time window are summarized to form a set of ambient light variation indexes for multiple time windows, representing the light variation patterns at different time scales, providing a stable and reliable basis for adjusting exposure control parameters and scene-adaptive low-power adjustment.

[0044] In this embodiment, step S3 includes the following steps: A preset exposure adjustment threshold is provided; the exposure adjustment threshold includes a light intensity change rate threshold, a light change amplitude threshold, and a light intensity fluctuation frequency threshold. The exposure adjustment threshold is compared with the ambient light change index set. When the ambient light change index set is greater than the exposure adjustment threshold, it is determined that the current scene is in a state of sudden change in lighting. A light change trend analysis was performed on the set of ambient light change indicators to obtain the change trend; The exposure adjustment parameters are dynamically adjusted based on the changing trend and output; the exposure adjustment parameters include exposure time, gain parameters and white balance parameters. If any one of the ambient light change indicators in the set is not greater than the corresponding threshold of the exposure adjustment threshold, the current exposure parameters are locked and remain unchanged.

[0045] In this embodiment, the judgment conditions for exposure control are parameterized to form a threshold system for exposure strategy decision-making. The exposure adjustment threshold consists of three core parameters: the light intensity change rate threshold, the light change amplitude threshold, and the illumination fluctuation frequency threshold. The light intensity change rate threshold limits the maximum acceptable range of brightness change per unit time, distinguishing between slow and sudden light changes; the light change amplitude threshold limits the maximum amplitude of overall brightness change within adjacent frames or time windows, determining whether a significant jump in illumination has occurred; and the illumination fluctuation frequency threshold limits the frequency range of periodic changes in ambient light, identifying periodic interfering light sources such as flickering lights. These thresholds can be preset according to different shooting modes. For example, in indoor conference mode, the light change amplitude threshold is set lower to improve exposure stability, while in outdoor dynamic shooting mode, a higher threshold is allowed to adapt to environmental changes. However, the three types of thresholds maintain a consistent judgment logic structure, thereby ensuring the consistency and transferability of the exposure control strategy.

[0046] The system compares and analyzes the ambient light change index sets generated within multiple time windows against preset exposure adjustment thresholds. Specifically, for each time window, three indicators—light intensity change rate, light change amplitude, and light fluctuation frequency—are extracted and compared with their corresponding thresholds. If, within the same time window, the light intensity change rate exceeds a preset rate threshold, the light change amplitude exceeds a preset amplitude threshold, and the light fluctuation frequency exceeds a preset frequency threshold, the lighting change within that time window is determined to be in a sudden change state. To avoid misjudgments caused by instantaneous anomalies, a multi-window consistency judgment mechanism is adopted. This requires that at least two consecutive time windows meet the condition of "all three indicators exceeding the threshold" before the current scene is finally confirmed to be in a sudden lighting change state. This judgment method can effectively distinguish between real lighting changes and short-term noise fluctuations, such as the rapid entry of vehicle headlights or the switching of indoor lights.

[0047] Time series modeling is performed on the rate and amplitude of light intensity change across multiple consecutive time windows, and the slope and growth trend are calculated using a sliding window approach. For example, if the rate of light intensity change shows a continuous upward trend across multiple time windows, the illumination is determined to be increasing; if it shows a continuous downward trend, the illumination is determined to be decreasing; and if it exhibits periodic fluctuations, it is determined to be a periodic trend based on the frequency characteristics of the illumination fluctuations. During trend analysis, a time series is constructed based on a 10fps decomposed frame rate, ensuring that 10 sampling points per second form stable basic data for trend analysis. Simultaneously, a trend smoothing mechanism is introduced to suppress short-term abnormal fluctuations, making the trend results more consistent with the actual laws of illumination change, thereby avoiding frequent jitter in exposure parameter adjustments.

[0048] When the light intensity is determined to be increasing, the exposure time is gradually shortened and the image sensor gain parameter is reduced to avoid overexposure. When the light intensity is determined to be decreasing, the exposure time is gradually increased and the gain parameter is increased to ensure stable image brightness. When the light intensity shows a periodic change trend, the combination strategy of white balance and gain parameters is prioritized to reduce the impact of color cast and brightness fluctuations. A progressive adjustment method is used during the adjustment process, meaning that the adjustment range of exposure time, gain parameter, and white balance parameter is limited to a preset range of change within adjacent control cycles, thereby avoiding image flicker or visual instability caused by sudden parameter changes. For example, the exposure time adjustment step can be controlled within the range of 5% to 10%, the gain parameter adjustment adopts a graded incremental method, and the white balance parameter is adaptively corrected according to the color temperature change trend. The final output is a stable combination of exposure adjustment parameters, enabling the camera to maintain stable image quality under different lighting conditions.

[0049] The system systematically detects the rate of change of light intensity, the amplitude of light change, and the frequency of light fluctuations within a set of ambient light change indicators. If any one of these indicators does not exceed its corresponding exposure adjustment threshold, it is determined that the current light change has not met the conditions for triggering exposure adjustment, thus entering an exposure lock state. In this state, the current exposure time, gain parameters, and white balance parameters remain unchanged, while the exposure parameter update calculation process is paused, maintaining only a low-frequency monitoring mode to determine if any new light change events have occurred. This mechanism effectively reduces the computational frequency of the exposure control module, lowers the power consumption of the image signal processing link, and avoids frequent adjustments to exposure parameters due to minor fluctuations in stable lighting conditions, thereby improving image stability and system energy efficiency.

[0050] In this embodiment, step S4 includes the following steps: Scene element detection is performed based on image frame sequences, and elements in the scene are labeled; the elements in the scene include people, vehicles and animals; Perform dynamic optical flow tracing on elements in the scene to extract the position change trajectories of multiple elements; Calculate the target number of elements in the scene and perform scene distribution analysis to obtain element quantity distribution information; Motion indices are calculated based on the position change trajectory to obtain elemental motion indices; the elemental motion indices include the target motion speed, the number of changes in motion direction, and the frequency of motion trajectory changes. Dynamic evaluation is performed based on the distribution information of element quantity and element movement indicators to obtain a dynamic feature set of the scene.

[0051] In this embodiment, based on a single-frame image feature extraction and multi-class target detection model, salient targets in the image are classified and identified. The identified categories include at least three typical moving targets: people, vehicles, and animals. Image frames are scale-normalized and feature-enhanced to ensure consistent feature representation under different lighting conditions. Multi-scale region scanning is performed on the image to obtain candidate target regions, which are then classified to determine the target category and its spatial location in the image. In each frame, detected targets are bounding box-marked, and their center coordinates, occupied area, and confidence information are recorded to form target marking results. By performing this detection process on consecutive image frames, a continuous record of target appearance can be established over time, ensuring accurate reflection of the distribution of key elements in the scene even under continuous recording at 30fps.

[0052] The number of people, vehicles, and animals identified in each frame of the image is statistically analyzed to obtain single-frame target quantity information. The target quantity is then accumulated and analyzed over a continuous time window to obtain stable target quantity statistics. Based on the spatial distribution of targets in the image, the image is divided into multiple analysis regions, such as the central region, left region, right region, and edge region. The target quantity distribution within each region is then statistically analyzed to obtain the spatial distribution characteristics of scene elements. Furthermore, independent statistical analysis is performed on different categories of targets (people, vehicles, and animals) to form multi-category target quantity distribution information. Through joint analysis of the temporal and spatial dimensions, the density and distribution of targets in the scene can be accurately reflected.

[0053] The target's velocity is obtained by calculating the change in the Euclidean distance between the target's center position and adjacent time sampling points, which characterizes the speed of the target's movement. Since the decomposition frame rate is 10fps, there are 10 sampling points per second, allowing for a relatively detailed description of the target's motion process. The target's motion direction is vectorized, and the number of changes in the target's motion direction is counted by calculating the change in the directional angle between adjacent trajectory segments, which characterizes the complexity of the target's motion path. The overall shape of the target trajectory is analyzed, and the frequency of trajectory curvature changes and path polygonal changes is counted to obtain the frequency of trajectory changes, which characterizes the stability and randomness of the target's motion. During the calculation process, different types of targets are analyzed independently, and the results are weighted and fused to reflect the intensity of the motion characteristics of the entire scene, thus forming a complete set of elemental motion indicators.

[0054] The target quantity distribution information is normalized to eliminate the impact of differences in the number of targets of different categories. The target movement speed, the number of changes in movement direction, and the frequency of trajectory changes are standardized to ensure comparability of movement indicators of different dimensions. A comprehensive calculation of target quantity density, spatial distribution concentration, and movement intensity yields a scene dynamic evaluation value. This evaluation value simultaneously reflects the density and activity level of targets in the scene; for example, it is significantly higher in densely populated and frequently moving scenes, and significantly lower in static meetings or low-activity scenes.

[0055] In this embodiment, step S5 includes the following steps: Set computational weights and perform complexity-weighted calculations on the dynamic feature set of the scene to obtain the scene complexity. Set a sliding calculation window, extract the continuous frame complexity of the scene complexity, and calculate the mean to obtain the average scene complexity of the current time window; The scene complexity levels are determined by classifying the scene complexity based on the average scene complexity; the scene complexity levels include low complexity, medium complexity, and high complexity.

[0056] In this embodiment, after obtaining the scene dynamic feature set, the scene complexity is obtained by performing a complexity-weighted calculation on each dynamic feature. The scene dynamic feature set includes at least target quantity distribution information, target movement speed, number of movement direction changes, and movement trajectory change frequency. For the above four types of features, calculation weights are set respectively, with the weight coefficient for target quantity distribution information set to 0.30, the weight coefficient for target movement speed set to 0.30, the weight coefficient for the number of movement direction changes set to 0.20, and the weight coefficient for movement trajectory change frequency set to 0.2. During the calculation process, each feature is normalized to uniformly map its numerical range to between 0 and 1, thereby eliminating the influence of differences in the dimensions of different features.

[0057] After normalization, each feature value is multiplied by its corresponding weight coefficient, and then weighted and summed to obtain the scene complexity value C at the current moment. Among these, the target quantity distribution information characterizes the density of targets in the scene and contributes the most to the complexity; the target movement speed characterizes the overall movement intensity and has a secondary impact on complexity; the number of changes in movement direction characterizes the complexity of the movement path; and the frequency of changes in movement trajectory characterizes the randomness and jitter of the movement. Through this weighting method, the scene complexity can simultaneously reflect both spatial distribution characteristics and temporal motion characteristics, thus achieving a unified quantitative expression for different activity scenarios.

[0058] Based on the image frame sequence corresponding to a 10fps decomposition frame rate, several consecutive frames are divided into a time analysis window. In a preferred embodiment, every 3 to 5 frames constitute a sliding calculation window, and 1 or 2 frames are used as the sliding step size, thereby forming multiple overlapping time windows.

[0059] Within each sliding window, the scene complexity values ​​for the corresponding consecutive frames are extracted frame by frame. All complexity values ​​are summed and then divided by the number of frames within the window to obtain the average scene complexity for the current time window. This average calculation method effectively eliminates abnormal fluctuations in complexity caused by false detections in single frames, partial occlusion, or instantaneous changes in illumination. The overlapping design of the sliding windows ensures continuity between adjacent time windows, resulting in a smooth trend in scene complexity over time. This approach guarantees both real-time performance and stability in the complexity assessment results, providing reliable input for subsequent scene level classification.

[0060] A threshold for classifying complexity levels is pre-defined, and the average scene complexity is compared with the corresponding threshold. When the average scene complexity is less than 0.40, the current scene is determined to be a low-complexity scene. This type of scene is usually characterized by a small number of targets, low movement speed, and relatively gentle trajectory changes, such as meeting recording or static indoor shooting scenes. When the average scene complexity is greater than or equal to 0.40 and less than 0.70, it is determined to be a medium-complexity scene. This type of scene usually has a certain number of moving targets and moderate levels of movement changes, such as daily activity recording or ordinary indoor and outdoor shooting scenes. When the average scene complexity is greater than or equal to 0.70, it is determined to be a high-complexity scene. This type of scene is usually characterized by dense targets, fast movement speed, and frequent trajectory changes, such as sports shooting or scenes of dense crowds.

[0061] By using the above-mentioned hierarchical method, continuously changing complexity values ​​are mapped to discrete level labels, enabling complexity information to be directly used for subsequent camera parameter control decisions, thereby realizing a low-power adaptive shooting control strategy based on scene complexity.

[0062] In this embodiment, the specific steps of step S6 are as follows: Based on the scene complexity level, perform adaptive optimization configuration of low-power shooting parameters and output adaptive configuration parameters; The camera is driven to perform low-power automatic shooting processing based on adaptive configuration parameters and exposure adjustment parameters.

[0063] In this embodiment, standard shooting configuration parameters are set as the baseline operating state. These parameters include standard dynamic resolution, standard shooting frame rate, standard video encoding bitrate, and standard image processing intensity. The standard shooting frame rate maintains the same 30fps recording capability as the aforementioned video acquisition to ensure video continuity. The standard dynamic resolution is used to maintain basic image quality output under different load conditions. The standard video encoding bitrate ensures a balance between video compression quality and transmission stability. The standard image processing intensity controls the computational overhead of image enhancement and ISP processing. When the scene complexity level is determined to be medium complexity, the standard shooting configuration parameters are directly output as adaptive configuration parameters without additional enhancement or reduction adjustments, ensuring the shooting system maintains a stable standard operating state while balancing image quality and power consumption. When the scene complexity level is determined to be low, the standard shooting configuration parameters are downgraded based on the characteristics of low-load scenes. Specifically, the dynamic resolution is reduced by one level to decrease pixel processing overhead; the shooting frame rate is adaptively reduced while maintaining a low-power control logic corresponding to 10fps in the analysis path; the video encoding bitrate is simultaneously reduced to decrease encoding computational complexity; and the image processing intensity is reduced to decrease ISP computational load. This generates low-power adaptive configuration parameters, enabling the camera to maintain basic image quality output and significantly reduce power consumption in static or low-activity scenes. When the scene complexity level is determined to be high, the standard shooting configuration parameters are enhanced. The dynamic resolution is increased to a high-quality output level to retain more detail; the shooting frame rate is maintained at 30fps with enhanced real-time processing capabilities; the video encoding bitrate is increased to adapt to high-motion-information scenes; and the image processing intensity is simultaneously increased to enhance motion compensation and image stabilization capabilities. This generates high-performance adaptive configuration parameters, enabling the camera to maintain complete image quality and motion capture capabilities in high-dynamic scenes.

[0064] Adaptive configuration parameters are used to control the operating status of the video encoding link and the image processing link, including dynamic resolution adjustment, shooting frame rate control, video encoding bitrate control, and image processing intensity adjustment; exposure adjustment parameters are used to control the imaging conditions of the image sensor front end, including exposure time, gain parameters, and white balance parameters, thereby affecting the quality of the original image acquisition.

[0065] In actual operation, when the camera enters a low-complexity scene, the adaptive configuration parameters reduce encoding complexity and image processing load, while the exposure adjustment parameters remain locked under stable ambient light conditions, thereby reducing the frequency of control calculations and achieving the lowest power consumption operation. When the camera is in a medium-complexity scene, both the adaptive configuration parameters and the exposure adjustment parameters remain in standard configuration, allowing the camera to maintain normal shooting quality under stable power consumption. When the camera is in a high-complexity scene, the adaptive configuration parameters enhance encoding and image processing capabilities to handle high dynamic range input, while the exposure adjustment parameters are dynamically adjusted according to the aforementioned lighting change trends. For example, exposure time and gain parameters are simultaneously optimized when lighting changes intensify to ensure image brightness stability and detail reproduction. Through the combined driving method of the above adaptive configuration parameters and exposure adjustment parameters, the camera can dynamically adjust the allocation of internal computing resources and imaging parameter configuration under different complexity scenes, thereby achieving overall power consumption optimization and improved battery life while ensuring video continuity and image quality stability.

[0066] In this embodiment, a low-power intelligent camera control device based on scene adaptive adjustment is provided, for executing the low-power intelligent camera control method based on scene adaptive adjustment as described above, including: The shooting module is used to continuously capture images of the shooting area using a smart camera to obtain a sequence of image frames; The brightness extraction module is used to extract brightness information from the image frame sequence and generate a set of ambient light change indicators for multiple time windows. The exposure adjustment module is used to dynamically adjust parameters based on the ambient light change index set and output exposure adjustment parameters. The element detection module is used to detect scene elements based on image frame sequences and calculate motion indicators to obtain a dynamic feature set of the scene. The scene segmentation module is used to classify scenes into levels based on dynamic feature sets to obtain scene complexity levels. The adaptive configuration module is used to perform adaptive configuration based on the scene complexity level and to perform low-power automatic shooting processing based on exposure adjustment parameters.

[0067] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0068] In the embodiments provided in this application, it should be understood that the disclosed apparatus / devices and methods can be implemented in other ways. For example, the apparatus / device embodiments described above are merely illustrative. For instance, the division of modules or 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 coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0069] 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.

[0070] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0071] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein are implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A low-power intelligent camera control method based on scene adaptive adjustment, characterized in that, Includes the following steps: Step S1: Continuously capture images of the shooting area using a smart camera to obtain an image frame sequence; Step S2: Extract brightness information from the image frame sequence to generate a set of ambient light change indicators for multiple time windows; Step S3: Based on the ambient light change index set, perform dynamic parameter adjustment and output exposure adjustment parameters; Step S4: Detect scene elements based on the image frame sequence and calculate motion indicators to obtain the scene dynamic feature set; Step S5: Classify the scene level based on the scene dynamic feature set to obtain the scene complexity level; Step S6: Perform adaptive configuration based on scene complexity level, and execute low-power automatic shooting processing based on exposure adjustment parameters.

2. The low-power intelligent camera control method based on scene adaptive adjustment according to claim 1, characterized in that, The specific steps of step S1 are as follows: The raw video stream is obtained by continuously shooting the shooting area using a smart camera; The original video stream is enhanced to obtain an enhanced video stream; Set the frame rate; The enhanced video stream is decomposed into multiple frames based on the decomposed frame rate to obtain an image frame sequence.

3. The low-power intelligent camera control method based on scene adaptive adjustment according to claim 2, characterized in that, The image enhancement specifically involves performing high-frequency noise filtering and temporal smoothing to eliminate high-frequency components caused by sensor jitter and environmental interference.

4. The low-power intelligent camera control method based on scene adaptive adjustment according to claim 2, characterized in that, The specific steps of step S2 are as follows: Brightness information is extracted from the image frame sequence to obtain brightness information for multiple frames; the brightness information includes the average brightness value, local brightness distribution value, and ambient light intensity value of the image. Based on the brightness information, continuous difference calculation is performed between adjacent frames to obtain the brightness change index between image frames; the brightness change index between image frames includes the light intensity change rate, the light intensity change amplitude, and the illumination fluctuation frequency. The brightness change index between image frames is statistically analyzed using time-series windows to generate a set of ambient light change indexes for multiple time windows.

5. The low-power intelligent camera control method based on scene adaptive adjustment according to claim 4, characterized in that, Step S3 is as follows: A preset exposure adjustment threshold is provided; the exposure adjustment threshold includes a light intensity change rate threshold, a light change amplitude threshold, and a light intensity fluctuation frequency threshold. The exposure adjustment threshold is compared with the ambient light change index set. When the ambient light change index set is greater than the exposure adjustment threshold, it is determined that the current scene is in a state of sudden change in lighting. A light change trend analysis was performed on the set of ambient light change indicators to obtain the change trend; The exposure adjustment parameters are dynamically adjusted based on the changing trend and output; the exposure adjustment parameters include exposure time, gain parameters and white balance parameters. If any one of the ambient light change indexes is not greater than the corresponding threshold of the exposure adjustment threshold, the current exposure parameters are locked and remain unchanged.

6. The low-power intelligent camera control method based on scene adaptive adjustment according to claim 5, characterized in that, The specific steps of step S4 are as follows: Scene element detection is performed based on image frame sequences, and elements in the scene are labeled; the elements in the scene include people, vehicles and animals; Perform dynamic optical flow tracing on elements in the scene to extract the position change trajectories of multiple elements; Calculate the target number of elements in the scene and perform scene distribution analysis to obtain element quantity distribution information; Motion indices are calculated based on the position change trajectory to obtain elemental motion indices; the elemental motion indices include the target motion speed, the number of changes in motion direction, and the frequency of motion trajectory changes. Dynamic evaluation is performed based on the distribution information of element quantity and element movement indicators to obtain a dynamic feature set of the scene.

7. The low-power intelligent camera control method based on scene adaptive adjustment according to claim 6, characterized in that, The specific steps of step S5 are as follows: Set computational weights and perform complexity-weighted calculations on the dynamic feature set of the scene to obtain the scene complexity. Set a sliding calculation window, extract the continuous frame complexity of the scene complexity, and calculate the mean to obtain the average scene complexity of the current time window; The scene complexity levels are obtained by classifying the scenes according to their average complexity. The scene complexity levels are categorized into low complexity, medium complexity, and high complexity.

8. The low-power intelligent camera control method based on scene adaptive adjustment according to claim 7, characterized in that, The specific steps of step S6 are as follows: Based on the scene complexity level, perform adaptive optimization configuration of low-power shooting parameters and output adaptive configuration parameters; The camera is driven to perform low-power automatic shooting processing based on adaptive configuration parameters and exposure adjustment parameters.

9. The low-power intelligent camera control method based on scene adaptive adjustment according to claim 8, characterized in that, The specific steps for performing adaptive optimization configuration of low-power shooting parameters based on the scene complexity level and outputting adaptive configuration parameters are as follows: Set standard shooting configuration parameters, including standard dynamic resolution, standard shooting frame rate, standard video encoding bitrate, and standard image processing intensity; When the scene complexity level is medium complexity, the standard shooting configuration parameters are used as adaptive configuration parameters for standard shooting processing. When the scene complexity level is low, reduce the dynamic resolution, reduce the shooting frame rate, reduce the video encoding bitrate, and reduce the image processing intensity to generate adaptive configuration parameters to maintain basic image quality output. When the scene complexity level is high, the dynamic resolution, shooting frame rate, video encoding bitrate, and image processing intensity are increased to generate adaptive configuration parameters for full-effect shooting and processing.

10. A low-power intelligent camera control device based on scene adaptive adjustment, characterized in that, The method for implementing the low-power intelligent camera control method based on scene adaptive adjustment as described in claim 1 includes: The shooting module is used to continuously capture images of the shooting area using a smart camera to obtain a sequence of image frames; The brightness extraction module is used to extract brightness information from the image frame sequence and generate a set of ambient light change indicators for multiple time windows. The exposure adjustment module is used to dynamically adjust parameters based on the ambient light change index set and output exposure adjustment parameters. The element detection module is used to detect scene elements based on image frame sequences and calculate motion indicators to obtain a dynamic feature set of the scene. The scene segmentation module is used to classify scenes into levels based on dynamic feature sets to obtain scene complexity levels. The adaptive configuration module is used to perform adaptive configuration based on the scene complexity level and to perform low-power automatic shooting processing based on exposure adjustment parameters.

Images