Exposure control method, image capture device, and computer-readable storage medium

Abstract

This application provides an exposure control method, an image acquisition device, and a computer-readable storage medium. The method includes: if a preset wake-up condition is met, triggering the image acquisition device to switch to a wake-up state and acquiring a first image at a first sampling density; determining a first brightness change value based on the first image; if the first brightness change value is greater than or equal to a first change threshold and less than a second change threshold, acquiring a second image at a second sampling density; determining a second brightness change value based on the second image; if the second brightness change value is greater than or equal to the second change threshold, predicting the ambient light brightness at the target acquisition time, and determining a second exposure parameter based on the predicted first ambient brightness, a preset expected ambient brightness, and a first exposure parameter, and acquiring an image based on the second exposure parameter and the target sampling density. This application can reduce device power consumption while improving the convergence capability of automatic exposure.

Description

Technical Field

[0001] This application relates to the field of image acquisition technology, and in particular to an exposure control method, an image acquisition device, and a computer-readable storage medium. Background Technology

[0002] With the rapid development of image acquisition technology and IoT technology, image acquisition devices (such as security cameras and smart vision terminals) are widely used in various scenarios. To meet the long battery life requirements in some scenarios, image acquisition devices typically employ low-power operating mechanisms. This means the device remains in sleep mode most of the time, only being awakened at specific times to acquire images, and then returning to sleep mode after completing the acquisition task. During the image acquisition process, ambient lighting conditions are often dynamically changing. To ensure good visibility and clarity of the acquired images, the device usually needs to perform an automatic exposure (AE) convergence process. This involves passively and iteratively adjusting exposure parameters by analyzing the brightness of the currently acquired image until the image brightness reaches a suitable level.

[0003] However, the image acquisition methods in related technologies are relatively simple. Regardless of the magnitude of changes in ambient light, the device often executes a fixed exposure adjustment process. This method has a long exposure adjustment time, consumes a lot of computing resources, and results in high power consumption during the wake-up period. Summary of the Invention

[0004] This application provides an exposure control method, an image acquisition device, a computer-readable storage medium, and a computer program product, which can improve the convergence capability of automatic exposure while reducing device power consumption through progressive sampling and predictive exposure estimation.

[0005] The technical solution of this application embodiment is implemented as follows: This application provides an exposure control method applied to an image acquisition device, the method comprising: While the image acquisition device is in a sleep state, if a preset wake-up condition is met, the image acquisition device is triggered to switch from the sleep state to the wake-up state, and in the wake-up state, a first image is acquired at a first sampling density; a first brightness feature of the first image is extracted, and a first brightness change value is determined based on the first brightness feature; if the first brightness change value is greater than or equal to a first change threshold and less than a second change threshold, a second image is acquired at a second sampling density; wherein, the second change threshold is greater than the first change threshold, and the second sampling density is higher than the first sampling density; a second brightness feature of the second image is extracted, and a second brightness change value is determined based on the second brightness feature; if the second brightness change value is greater than or equal to the second change threshold, the brightness of the ambient light at the target acquisition time is predicted to obtain a first predicted ambient brightness; a second exposure parameter is determined based on the first predicted ambient brightness, a preset expected ambient brightness, and a first exposure parameter when acquiring the first image; image acquisition is performed based on the second exposure parameter and the target sampling density; wherein, the target sampling density is higher than the second sampling density.

[0006] In the above scheme, the method further includes: acquiring historical environmental data; classifying the historical environmental data using a clustering algorithm to establish an environmental pattern database; recording exposure convergence success records; establishing a convergence decision tree model based on the exposure convergence success records; and determining whether to trigger a switch from the dormant state to the awake state based on the convergence decision tree model.

[0007] This application provides an image acquisition device, including: The wake-up control module is used to trigger the image acquisition device to switch from the sleep state to the wake-up state if a preset wake-up condition is met while the image acquisition device is in a sleep state. An image processing module is configured to: acquire a first image at a first sampling density in the wake-up state; extract a first brightness feature from the first image and determine a first brightness change value based on the first brightness feature; if the first brightness change value is greater than or equal to a first change threshold and less than a second change threshold, acquire a second image at a second sampling density; wherein the second change threshold is greater than the first change threshold and the second sampling density is higher than the first sampling density; extract a second brightness feature from the second image and determine a second brightness change value based on the second brightness feature; The prediction module is used to predict the ambient light brightness at the target acquisition time if the second brightness change value is greater than or equal to the second change threshold, and obtain a first predicted ambient brightness; and determine a second exposure parameter based on the first predicted ambient brightness, the preset expected ambient brightness and the first exposure parameter when acquiring the first image. The image processing module is further configured to acquire an image based on the second exposure parameter and the target sampling density; wherein the target sampling density is higher than the second sampling density.

[0008] In the above scheme, the wake-up control module is further configured to determine whether the preset wake-up condition is met by at least one of the following: for any evaluation time during the period when the image acquisition device is in a sleep state, if the time interval between the evaluation time and the previous wake-up time reaches a preset time period, the preset wake-up condition is determined to be met; if the brightness change rate of the ambient light at the evaluation time is greater than the change rate threshold, the preset wake-up condition is determined to be met; if the evaluation time reaches the pre-predicted target wake-up time, the preset wake-up condition is determined to be met.

[0009] In the above scheme, the wake-up control module is further configured to acquire the first ambient brightness of the ambient light at the time to be evaluated, the second ambient brightness of the ambient light at the first historical acquisition time, and the first ambient brightness sequence acquired within a preset historical period; determine the brightness change rate based on the first ambient brightness, the second ambient brightness, and the time interval between the time to be evaluated and the first historical acquisition time; determine the median absolute deviation of the first ambient brightness sequence, and use the product of the median absolute deviation and a preset sensitivity coefficient as the change rate threshold; wherein, the preset sensitivity coefficient is used to constrain the wake-up sensitivity of the image acquisition device.

[0010] In the above scheme, the wake-up control module is further configured to acquire a second ambient brightness sequence acquired before the second historical acquisition time; the second historical acquisition time is the wake-up time preceding the time to be evaluated; a short-term prediction model and a long-term trend model are constructed based on the second ambient brightness sequence; wherein, the short-term prediction model represents the weighted sum of ambient brightness over different time spans from the second historical acquisition time; the long-term trend model is determined based on the brightness change value within a first preset time window and the brightness change slope within a second preset time window; the time span of the second preset time window is greater than the time span of the first preset time window; based on the short-term prediction model and the long-term trend model, a second predicted ambient brightness at the time to be predicted after the second historical acquisition time is predicted; if the difference between the second predicted ambient brightness and the third ambient brightness corresponding to the second historical acquisition time is greater than a preset change amount, then the time to be predicted is used as a reference, and a preset time offset is shifted forward to obtain the pre-predicted target wake-up time.

[0011] In the above scheme, the image processing module is further configured to determine the first global average brightness of the first image and use the first global average brightness as the first brightness feature; obtain the second global average brightness of the third image; wherein the third image is acquired at the previous wake-up time when the image acquisition device switches from the sleep state to the wake-up state; determine the absolute value of the first difference between the first global average brightness and the second global average brightness; and use the ratio of the absolute value of the first difference to the preset maximum brightness value of the image acquisition device as the first brightness change value.

[0012] In the above scheme, the image processing module is further configured to maintain the first exposure parameter if the first brightness change value is less than the first change threshold, and perform image acquisition based on the first exposure parameter and the target sampling density; the wake-up control module is further configured to trigger the image acquisition device to switch from the wake-up state to the sleep state in response to the image acquisition end command.

[0013] In the above scheme, the image processing module is further configured to: determine multiple regions of interest (ROIs) pre-divided for the second image; determine the initial average brightness and region weight of each ROI; determine the average brightness of each ROI based on the initial average brightness and region weight; take the average of the average brightness of the multiple ROIs as the third global average brightness of the second image; determine the total number of pixels in the second image and the number of overexposed pixels with brightness values ​​greater than a preset brightness limit, and take the ratio of the number of overexposed pixels to the total number of pixels as the overexposed pixel ratio of the second image; take the third global average brightness and the overexposed pixel ratio as the second brightness feature; acquire the second global average brightness of the third image; wherein the third image was acquired at the previous wake-up time when the image acquisition device switches from the sleep state to the wake-up state; determine the absolute value of the second difference between the third global average brightness and the second global average brightness; take the ratio of the absolute value of the second difference to the preset maximum brightness value of the image acquisition device as the initial brightness change value; and correct the initial brightness change value based on the overexposed pixel ratio to obtain the second brightness change value.

[0014] In the above scheme, the second brightness feature includes the third global average brightness and the overexposed pixel ratio of the second image; the image processing module is further configured to determine an exposure offset based on the third global average brightness and the overexposed pixel ratio if the second brightness change value is less than the second change threshold; adjust the first exposure parameter based on the exposure offset to obtain a third exposure parameter, and perform image acquisition based on the third exposure parameter and the target sampling density; the wake-up control module is further configured to trigger the image acquisition device to switch from the wake-up state to the sleep state in response to the image acquisition end command.

[0015] In the above scheme, the prediction module is further configured to determine the first acquisition time when acquiring the second image, and obtain the third ambient brightness sequence acquired before the first acquisition time; perform a weighted summation of the fourth ambient brightness corresponding to the first acquisition time and each ambient brightness in the third ambient brightness sequence to obtain the first predicted ambient brightness corresponding to the target acquisition time, wherein the target acquisition time is the next image acquisition time after the first acquisition time.

[0016] In the above scheme, the prediction module is further configured to determine the ratio of the first predicted ambient brightness to the desired ambient brightness, and use the product of the ratio and the first exposure parameter as the second exposure parameter.

[0017] In the above scheme, the wake-up control module is further configured to determine the number of valid wake-ups based on wake-up records within a preset historical period prior to the time to be evaluated; determine the effective wake-up rate based on the number of valid wake-ups and the wake-up records; adjust the preset time period based on the difference between the preset target wake-up rate and the effective wake-up rate; and determine that the preset wake-up condition is met if the time interval between the time to be evaluated and the previous wake-up time reaches the adjusted preset time period.

[0018] This application provides an image acquisition device, the image acquisition device comprising: Memory is used to store executable instructions or computer programs. When a processor executes computer-executable instructions or computer programs stored in the memory, it implements the exposure control method provided in the embodiments of this application.

[0019] This application provides a computer-readable storage medium storing a computer program or computer-executable instructions for implementing the exposure control method provided in this application when executed by a processor.

[0020] This application provides a computer program product, including a computer program or computer executable instructions. When the computer program or computer executable instructions are executed by a processor, they implement the exposure control method provided in this application.

[0021] The embodiments of this application have the following beneficial effects: In this embodiment, the image acquisition device triggers wake-up when a preset wake-up condition is met. It then acquires images at a first sampling density and a second sampling density, extracting the corresponding brightness change values. When the brightness change value reaches a second threshold, the brightness of the ambient light at the target acquisition moment is predicted. Based on the predicted ambient brightness, the desired ambient brightness, and the initial exposure parameters, the second exposure parameters are directly calculated. Finally, image acquisition is performed using the highest target sampling density and the second exposure parameters. This scheme balances environmental perception agility with hardware resource consumption through hierarchical state evaluation based on increasing sampling density and exposure parameter prediction. Specifically, this method first reduces background noise interference by using gradient sampling density, thus reducing computational resource consumption when ambient light does not change significantly. Second, it sets a forward-looking data boundary for exposure control by predicting the brightness at the target acquisition moment, which is introduced by significant light changes. This approach solves the problem in related technologies where passive probing using a single full-resolution image across multiple frames leads to an inability to adapt to dynamic lighting environment changes and an inability to balance the battery life of low-power devices with exposure response speed. This application's hierarchical sampling method reduces unnecessary computational load caused by unnecessarily changing light conditions, allowing the system's computing power to be highly focused on core computational changes requiring intervention. Furthermore, predictive exposure estimation can find an exposure parameter update direction that balances current hardware timing and future lighting trends. This makes image acquisition equipment more efficient and intelligent, reducing power consumption while improving the convergence capability of automatic exposure. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the architecture of the image acquisition device provided in the embodiments of this application; Figure 2 This is a schematic diagram of the structure of the image acquisition device provided in the embodiments of this application; Figure 3 This is a first flowchart illustrating the exposure control method provided in an embodiment of this application; Figure 4 This is a flowchart illustrating the method for determining the first brightness change value provided in an embodiment of this application; Figure 5 This is a flowchart illustrating the method for determining the second brightness change value provided in an embodiment of this application; Figure 6 This is a flowchart illustrating the first method for predicting ambient brightness provided in an embodiment of this application. Figure 7This is a flowchart illustrating the method for determining the second exposure parameter provided in an embodiment of this application; Figure 8 This is a schematic diagram of the second process of the exposure control method provided in the embodiments of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0025] In the following description, the terms "first, second, third, fourth" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, third, fourth" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0026] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0027] Unless otherwise defined, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in the embodiments of this application is for the purpose of describing the embodiments of this application only and is not intended to limit this application.

[0028] Before providing a further detailed description of the embodiments of this application, the nouns and terms involved in the embodiments of this application will be explained, and the nouns and terms involved in the embodiments of this application shall be interpreted as follows.

[0029] Response: Used to indicate the conditions or states on which the operation performed depends. When the conditions or states on which it depends are met, one or more operations performed can be performed in real time or with a set delay. Unless otherwise specified, there is no restriction on the order in which multiple operations are performed.

[0030] In low-power image acquisition, traditional automatic exposure mechanisms typically rely on continuously acquiring multiple frames of images for passive and iterative parameter adjustments. When faced with dynamic lighting changes, this slow convergence process generates significant unnecessary computational overhead and power consumption, making it difficult for image acquisition devices to simultaneously meet the requirements of light change response and power consumption control. This is a technical pain point that urgently needs to be addressed in the industry.

[0031] Taking a typical Always-On-Video (AOV) monitor (i.e., an AOV image acquisition device, also known as an AOV device) as an example, in low-power mode, it typically wakes up the main controller briefly at fixed intervals (e.g., every 2 seconds) to output one frame of image, and then immediately enters sleep mode. However, when ambient lighting conditions change dynamically, the AOV device must perform automatic exposure convergence to ensure the clarity of the output image. In related technologies, automatic exposure convergence often requires the AOV device to switch to constant-power mode and passively and iteratively adjust the exposure parameters by continuously acquiring multiple frames of images. This method has a long exposure adjustment time and consumes a lot of computing resources, resulting in a surge in power consumption of the AOV device during the wake-up period, which violates the original intention of power saving of the AOV device; and if a fixed exposure adjustment process is still used when the ambient light changes abruptly, it is easy to cause overexposure or underexposure of the acquired image.

[0032] Therefore, embodiments of this application provide an exposure control method, an image acquisition device, a computer-readable storage medium, and a computer program product that can improve the convergence capability of automatic exposure while reducing device power consumption through progressive sampling and predictive exposure estimation.

[0033] The following describes the application scenarios of the exposure control method provided in the embodiments of this application. The exposure control method of this application can be applied to various application scenarios involving image acquisition, such as security monitoring, ecological monitoring, smart home and smart access control, vehicle parking monitoring, portable wearable and mobile acquisition, and industrial inspection. Examples are given below.

[0034] In some embodiments, the exposure control method of this application can be applied to outdoor passive security surveillance and ecological monitoring scenarios. In this scenario, the image acquisition device can be: a solar-powered or battery-powered wireless network camera, an infrared camera for monitoring wild flora and fauna, an unattended monitoring node for farmland / forestry / water conservancy, etc.

[0035] In some embodiments, the exposure control method of this application can be applied to smart home and smart access control scenarios. In such scenarios, the image acquisition device can be a smart video doorbell, a smart electronic peephole, a home battery-powered camera, etc.

[0036] In some embodiments, the exposure control method of this application can also be applied to vehicle parking monitoring scenarios. In such scenarios, the image acquisition device can be a dashcam with sentry mode or parking security function, a vehicle surround view camera, etc.

[0037] In some embodiments, the exposure control method of this application can be applied to portable wearable and mobile law enforcement scenarios. In such scenarios, the image acquisition device can be: a portable law enforcement recorder, a smart helmet (such as a food delivery rider or security patrol helmet) camera, or an action camera.

[0038] In some embodiments, the exposure control method of this application can also be applied to industrial inspection and warehouse logistics automation scenarios. In this scenario, the image acquisition device can be: a drone power or pipeline periodic inspection camera, an intermittent vision sensor in front of an automated guided vehicle, a passive timed monitoring node for a large warehouse, etc.

[0039] See Figure 1 , Figure 1 This is a schematic diagram of the architecture of the image acquisition device 100 provided in this application embodiment. To support an image acquisition application, the image acquisition device 100 is configured with a main control chip 110 and peripheral hardware components. The peripheral hardware components may specifically include a timer, an image sensor, and an ambient light sensor. The main control chip integrates multiple functional modules for executing the image acquisition and processing logic of this application, specifically including: a wake-up control module 1101, an image processing engine 1102 (also called an image processing module), and a prediction learning optimization module 1103 (also called a prediction module).

[0040] Specifically, the interaction relationships and functional implementation of the various modules within the image acquisition device 100 are as follows: The wake-up control module 1101 is communicatively connected to the timer. The wake-up control module 1101 is primarily responsible for monitoring and switching the state of the image acquisition device 100 in low-power sleep mode. During the sleep state of the image acquisition device 100, the wake-up control module 1101 continuously or periodically evaluates the state. If it determines that the preset wake-up conditions are met (e.g., receiving a timeout interrupt from the timer, or receiving other forms of wake-up trigger signals), it triggers the image acquisition device 100 to switch from sleep mode to wake-up mode, thereby providing normal operating voltage and operating environment for other image processing units within the main control chip and peripheral sensors.

[0041] The image processing engine 1102 (i.e., the image processing module) is communicatively connected to the image sensor and is used to drive and control the image sensor to perform "progressive" image acquisition and preliminary evaluation operations. Specifically, when the image acquisition device 100 enters the wake-up state: First, the image processing engine 1102 controls the image sensor to acquire a first image at a first sampling density and extracts a first brightness feature from the first image, and then determines a first brightness change value based on the first brightness feature. Second, the image processing engine 1102 analyzes the magnitude of the brightness change. If it is determined that the first brightness change value is greater than or equal to a first change threshold and less than a second change threshold (i.e., indicating a moderate change in ambient light, which needs further confirmation), then the image sensor is controlled to acquire a second image at a second sampling density. At this stage, the image processing engine 1102 further extracts a second brightness feature from the second image and determines a second brightness change value based on this. The second change threshold is greater than the first change threshold, and the second sampling density is higher than the first sampling density.

[0042] The predictive learning optimization module 1103 (i.e., the prediction module) is communicatively connected to the ambient light sensor and is specifically designed for rapid exposure convergence (AE convergence) decisions when there are drastic changes in illumination. When the image processing engine 1102 calculates that the second brightness change value is greater than or equal to the second change threshold (i.e., indicating a sudden or drastic change in ambient light), the predictive learning optimization module 1103 is triggered. The predictive learning optimization module 1103 (combining ambient brightness data acquired by the ambient light sensor) actively predicts the brightness of the ambient light at the time of target acquisition, obtaining a first predicted ambient brightness. Subsequently, based on the calculated first predicted ambient brightness, the preset desired ambient brightness (i.e., the optimal exposure target value), and the first exposure parameters used when acquiring the aforementioned first image, the predictive learning optimization module 1103, through mathematical modeling or proportional conversion, calculates the second exposure parameters required for the target in one step. This method breaks away from the traditional passive trial-and-error exposure adjustment approach and achieves pre-convergence of exposure parameters under complex lighting conditions.

[0043] Finally, after the predictive learning optimization module 1103 completes parameter estimation, the image processing engine 1102 receives and calls the second exposure parameters to control the image sensor to perform final image acquisition at the target sampling density (e.g., a full-resolution high-definition image). The target sampling density is higher than the aforementioned second sampling density. Through the coordinated operation of these modules, the image acquisition device 100 outputs a full-resolution image only when absolutely necessary, significantly reducing system power consumption while ensuring rapid response to image brightness.

[0044] It should be noted that in actual physical implementation, the aforementioned main control chip can be a microprocessor (MCU), digital signal processor (DSP), application-specific integrated circuit (ASIC), system-on-a-chip (SoC), or a standalone image signal processor (ISP), etc. The wake-up control module, image processing engine, and predictive learning optimization module mentioned above can be software program code / instructions stored in a storage medium and read and executed by the main control chip; or they can be hardware coprocessors or logic computing units directly embedded inside the main control chip through hardware circuits such as logic gate arrays. This application embodiment does not impose specific limitations on this.

[0045] See Figure 2 , Figure 2 This is a schematic diagram of the structure of the image acquisition device 400 provided in the embodiments of this application. Figure 2 The image acquisition device 400 shown includes at least one processor 410, a memory 450, at least one network interface 420, and a user interface 430. The various components in the image acquisition device 400 are coupled together via a bus system 440. It is understood that the bus system 440 is used to implement communication between these components. In addition to a data bus, the bus system 440 also includes a power bus, a control bus, and a status signal bus. However, for clarity, ... Figure 2 The general labeled all buses as Bus System 440.

[0046] Processor 410 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0047] User interface 430 includes one or more output devices 431 that enable the presentation of media content, including one or more speakers and / or one or more visual displays. User interface 430 also includes one or more input devices 432, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

[0048] The memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state storage, hard disk drives, optical disk drives, etc. The memory 450 may optionally include one or more storage devices physically located away from the processor 410.

[0049] The memory 450 may include volatile memory or non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), and the volatile memory may be random access memory (RAM). The memory 450 described in this application embodiment is intended to include any suitable type of memory.

[0050] In some embodiments, memory 450 is capable of storing data to support various operations, examples of which include programs, modules, and data structures or subsets or supersets thereof, as illustrated below.

[0051] Operating system 451 includes system programs for handling various basic system services and performing hardware-related tasks, such as the framework layer, core library layer, driver layer, etc., for implementing various basic business functions and handling hardware-based tasks; The network communication module 452 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 420, exemplary network interfaces 420 including: Bluetooth, WiFi, and Universal Serial Bus (USB), etc. Presentation module 453 is configured to enable the presentation of information (e.g., a user interface for operating peripheral devices and displaying content and information) via one or more output devices 431 associated with user interface 430 (e.g., a display screen, a speaker, etc.). The input processing module 454 is used to detect and translate one or more user inputs or interactions from one or more input devices 432.

[0052] In some embodiments, the apparatus provided in this application can be implemented in software. Figure 2 An exposure control device 455 stored in memory 450 is shown. This device can be software in the form of programs and plug-ins, and includes the following software modules: wake-up control module 4551, image processing module 4552, and prediction module 4553. These modules are logically linked and can therefore be arbitrarily combined or further separated according to their implemented functions. The functions of each module will be described below.

[0053] The exposure control method provided in the embodiments of this application will be described below. As mentioned above, the electronic device implementing the exposure control method in the embodiments of this application can be an image acquisition device. Therefore, the executing entity of each step will not be described again below.

[0054] See Figure 3 , Figure 3This is a first flowchart illustrating the exposure control method provided in this application embodiment, which will be combined with... Figure 3 The steps shown are explained below. Figure 3 The main component of the process is the image acquisition device.

[0055] In step 101, if the preset wake-up condition is met while the image acquisition device is in a sleep state, the image acquisition device is triggered to switch from a sleep state to a wake-up state, and in the wake-up state, the first image is acquired at a first sampling density.

[0056] Here, "sleep mode" refers to a low-power standby mode that image acquisition devices enter to maximize energy conservation and extend battery life. During sleep mode, the main control chip, image processing engine, and main data channel of the image sensor are typically powered off or deeply suspended, and the image acquisition device stops outputting regular video streams. At this time, only the extremely low-power peripherals within the image acquisition device (such as ambient light sensors and timers) are continuously running or polling, waiting for physical changes in the environment. When the low-power peripherals detect that environmental or temporal characteristics reach a threshold (i.e., the preset wake-up conditions are met), they send an interrupt signal to the main control chip, triggering the image acquisition device to switch from sleep mode to wake-up mode. During this process, the main data path of the image acquisition device is powered on or activated again, enabling the image processing module to re-acquire and process image data. Taking a typical AOV (Always-On Video) operating mode as an example, AOV devices achieve long battery life by switching between wake-up and sleep modes. For example, in a 2-second AOV workflow, the AOV device may spend approximately 100ms in the wake-up state to complete image acquisition and algorithm processing, and then immediately enter a sleep state lasting approximately 1900ms. The switch from sleep state to wake-up state described in this application embodiment corresponds to the key node in this cycle where the AOV device is triggered to activate from a purely low-power standby period.

[0057] When the image acquisition device first enters the wake-up state, since it has not yet confirmed whether the ambient light has changed substantially, to avoid unnecessary battery consumption caused by directly starting full-power image metering operations, the image acquisition device does not immediately output high-definition video frames for end-user viewing. Instead, as a trial, it acquires a first image at a first sampling density. Here, sampling density refers to the density of pixels in the spatial dimension of an image, representing the total amount of pixel data contained in a single frame. Reducing sampling density can be achieved by directly reducing the sampling resolution of image acquisition, or by methods that reduce the total number of pixels, such as pixel merging, skip sampling, or interlaced or inter-column sampling. The first sampling density refers to an extremely low pixel density (e.g., the total number of pixels in a single frame is less than a preset pixel threshold, or a highly sparse sampling strategy is used). For example, the image sensor can output a thumbnail with a scale of 1 / 16 of the normal resolution (e.g., reducing or downsampling the 1920×1080 target acquisition size to 480×270) as the first image through pixel merging or skip sampling.

[0058] The first image, serving as the first frame after wake-up, does not require the preservation of clear texture details or use in final video storage and display; its sole purpose is to provide the most basic global illumination data. This design, acquiring the first image at an extremely low sampling density (first sampling density), reduces pixel processing directly at the data source while still achieving the goal of detecting the true illumination state of the current environment. This minimizes the data throughput of the image sensor and the computational overhead of the main control chip, laying the foundation for a subsequent low-power hierarchical photometric interception mechanism.

[0059] In some embodiments, determining that a preset wake-up condition is met includes at least one of the following: First, for any evaluation time during the image acquisition device's sleep state, if the time interval between the evaluation time and the previous wake-up time reaches a preset time period, the preset wake-up condition is determined to be met. Second, if the rate of change of ambient light brightness at the evaluation time is greater than the rate of change threshold, the preset wake-up condition is determined to be met. Third, if the evaluation time reaches the pre-predicted target wake-up time, the preset wake-up condition is determined to be met.

[0060] It's important to note that the wake-up moment signifies a state switch in the core processing resources of an image processing device. Typically, no business data is immediately generated upon wake-up; it merely marks the starting point for preparing for work. Conversely, the acquisition moment signifies the occurrence of data acquisition actions, such as image acquisition. This requires the image processing device to be actively powered. Generally, after being woken up, the image acquisition device immediately acquires an image; therefore, the wake-up moment is equivalent to one acquisition moment. However, multiple acquisition moments can occur after the image acquisition device is woken up (i.e., one image is acquired upon wake-up, and multiple more images can be acquired subsequently).

[0061] First, let's explain the determination method for the first one.

[0062] Here, the "time to be evaluated" refers to any specific time point on the timeline where the image acquisition device is in sleep mode, either in observation or timing evaluation mode. It serves as the anchor point for determining whether a wake-up action needs to be triggered. The "previous wake-up time" refers to the exact physical time point on the historical timeline where the image acquisition device most recently successfully switched from sleep mode to wake-up mode. The "preset time period" is a fixed duration parameter pre-stored in the image acquisition device's underlying layer. It is used to regulate the maximum time span that the sleep mode can maintain, ensuring that the image acquisition device can still perform periodic self-wake-up and environmental checks even in the absence of external triggering events.

[0063] In practical applications, the moment the image acquisition device enters sleep mode, the underlying hardware low-power timer records the current moment as the previous wake-up moment. Subsequently, the hardware low-power timer begins to run independently in ultra-low power mode and accumulates the time. When the timeline advances to any moment to be evaluated, the hardware low-power timer calculates the time difference between the moment to be evaluated and the recorded previous wake-up moment, thus obtaining the time interval. The calculated time interval is then compared with a configured preset time period. When the time interval is equal to or greater than the preset time period, the hardware low-power timer triggers a hardware interrupt signal. Upon receiving the hardware interrupt signal, the wake-up control module generates a wake-up command and then determines that the preset wake-up conditions are met.

[0064] As an example, in AOV (Always On Video) mode, when each preset time period T is reached and wake-up is triggered, the image acquisition device will immediately execute the subsequent hierarchical image acquisition process (for example, first outputting a 1 / 16 thumbnail frame for automatic exposure prediction). If it is determined that the current ambient light is stable and there is no need for complex exposure parameter convergence, the image acquisition device will directly output a full-resolution complete image and immediately trigger hardware power-off after outputting, returning to sleep mode to wait for the next cycle.

[0065] Secondly, the determination method for the second one will be explained.

[0066] Here, the rate of change of brightness is used to characterize the rate at which ambient light intensity changes within a unit of time, quantifying the severity of scene light fluctuations, as opposed to pure absolute light differences. The rate of change threshold is used to define whether ambient light fluctuations constitute a sensitive boundary parameter worthy of triggering a wake-up event, serving as a criterion for filtering out slowly varying background ambient light noise and occasional lighting fluctuations.

[0067] In practical applications, while the image acquisition device is in sleep mode, the ambient light sensor configured on the device's periphery maintains low power consumption. The ambient light sensor acquires the current ambient brightness value at the time of evaluation. Since calculating the brightness change rate requires a time dimension, it is also necessary to extract the ambient brightness values ​​cached from historical moments and combine them with the time difference between the historical moments and the time of evaluation to perform differentiation or slope calculations, thereby accurately deriving the brightness change rate at the current time of evaluation. Subsequently, the calculated brightness change rate is compared in absolute terms with a pre-configured change rate threshold. Once it is determined that the brightness change rate is significantly greater than the change rate threshold, a sudden and significant change in ambient light is confirmed, and a wake-up trigger flag is output, indicating that the preset wake-up conditions have been met.

[0068] Finally, the determination methods for the three mentioned above will be explained.

[0069] Here, the target wake-up time is a forward-looking time parameter, which refers to a specific future time node calculated based on historical data and pre-set compensation offset. It is used to actively control the image acquisition device to wake up from hibernation before the predicted drastic change in light occurs.

[0070] In practical applications, before the image acquisition device enters its current sleep state, the predictive learning optimization module has already completed complex model calculations and written the calculated exact time point (i.e., the pre-predicted target wake-up time) into the real-time clock alarm register. During sleep, the real-time clock continuously compares the current system time stream with the target wake-up time in the real-time clock alarm register. The real-time clock performs a query and match at every tiny time step (i.e., the continuously advancing time to be evaluated). When the time to be evaluated on the timeline numerically matches the pre-predicted target wake-up time stored in the real-time clock alarm register, the match is successful. At this point, the real-time clock hardware triggers a dedicated alarm interrupt pin to go low or high, receiving the predictive alarm interrupt, thus determining that the preset wake-up conditions are met.

[0071] By combining three diverse triggering conditions—time-period guarantee, brightness change rate drive, and target time prediction—a robust and environmentally adaptable hybrid wake-up architecture has been constructed, breaking the limitations of traditional single hardware timer or single sensor wake-up. In stable scenarios, the time-period condition serves as the basic wake-up mechanism, ensuring that the image acquisition device does not remain dormant due to prolonged lack of light stimulation. In scenarios with frequent sudden physical events, the brightness change rate condition enables the image acquisition device to respond through physical interruptions, achieving rapid acquisition of data in response to sudden interference. In scenarios with regular environmental changes, the introduction of target time prediction allows the image acquisition device to wake up in advance, reducing computational lag caused by passive responses. The parallel operation of these three conditions, coupled with hardware logic mutual exclusion and preemption priority, ensures that the image acquisition device can accurately switch states in response to changes in the external environment while maintaining low power consumption.

[0072] In some embodiments, for the first one described above, before determining that the preset wake-up condition is met if the time interval between the time to be evaluated and the previous wake-up time reaches a preset time period, the image acquisition device may further determine the number of valid wake-ups based on wake-up records in a preset historical period before the time to be evaluated; determine the effective wake-up rate based on the number of valid wake-ups and the wake-up records; and adjust the preset time period based on the difference between the preset target wake-up rate and the effective wake-up rate. Accordingly, the above step "determining that the preset wake-up condition is met if the time interval between the time to be evaluated and the previous wake-up time reaches a preset time period" can be implemented in the following way: if the time interval between the time to be evaluated and the previous wake-up time reaches the adjusted preset time period, then the preset wake-up condition is met.

[0073] Here, "wake-up record" refers to the historical event log and corresponding processing results of the image acquisition device switching from sleep to wake-up state. It is used to statistically analyze the activity status and triggering situations of the image acquisition device over a past period. "Effective wake-up count" refers to the sum of the number of times, within the total historical wake-up events, the number of times a real and substantial change in ambient light caused the image acquisition device to perform subsequent image processing or exposure parameter updates. It measures the frequency at which the image acquisition device did not perform invalid checks after waking up. "Effective wake-up rate" refers to the ratio of effective wake-up counts to the total number of extracted wake-up record entries. It is used to quantitatively evaluate the rationality of the current sleep-wake frequency setting and the true dynamic attributes of the environment. "Preset target wake-up rate" refers to the desired ideal effective wake-up ratio, used as a benchmark reference target for closed-loop feedback control adjustment to balance power consumption and image capture timeliness.

[0074] In practical applications, during the operation of image acquisition equipment, an internal log database is maintained to record historical wake-up events and post-wake-up processing status. Before determining a timed wake-up, data entries within a preset historical time period are extracted from the log database as wake-up records. The extracted wake-up records are traversed, and the number of entries marked as having experienced changes in ambient light is counted and accumulated to determine the effective wake-up count. Next, the total number of extracted wake-up records is counted, and a division operation is performed, dividing the effective wake-up count by the total number of wake-up records to determine the effective wake-up rate. Subsequently, the internally stored preset target wake-up rate is read, and the calculated effective wake-up rate is subtracted from the preset target wake-up rate to obtain the wake-up rate difference. The wake-up rate difference is multiplied by the internal proportional adjustment parameter to obtain the amount of time to be compensated, and this compensation amount is added to the original time period parameter to complete the adjustment of the preset time period. The timer continuously calculates the time interval between the time to be evaluated and the previous wake-up time, and compares the calculated time interval with the adjusted preset time period in real time. When the value of the time interval increases to be equal to or greater than the adjusted preset time period, the timer triggers an interrupt signal, determining that the preset wake-up condition is met.

[0075] By introducing an adaptive adjustment mechanism based on historical operating performance, the underlying hardware timer can be adjusted according to the complexity of the external physical environment. In scenarios with stable ambient lighting but local fluctuations and interference, the frequency of invalid wake-ups of the image acquisition device increases. This adaptive adjustment mechanism automatically extends the timer's duration through difference calculation, thereby reducing the number of invalid wake-ups and lowering the overall standby power consumption of the device. Conversely, in scenarios with frequent changes in ambient lighting, this mechanism automatically shortens the timer's duration, increasing the sampling density of the image acquisition device for changes in environmental conditions and reducing the omission of critical image acquisition. This adaptive mechanism overcomes the limitation of fixed timers in balancing low power consumption and responsiveness, making the hardware's operating cycle adaptable to the complexity of the physical environment.

[0076] In some embodiments, for the second one described above, the change rate threshold and the brightness change rate can be determined by the following method: acquiring the first ambient brightness of the ambient light at the time to be evaluated, the second ambient brightness of the ambient light at the first historical acquisition time, and the first ambient brightness sequence acquired within a preset historical period; determining the brightness change rate based on the first ambient brightness, the second ambient brightness, and the time interval between the time to be evaluated and the first historical acquisition time; determining the median absolute deviation of the first ambient brightness sequence, and using the product of the median absolute deviation and a preset sensitivity coefficient as the change rate threshold; wherein, the preset sensitivity coefficient is used to constrain the wake-up sensitivity of the image acquisition device.

[0077] Here, ambient brightness (such as first ambient brightness, second ambient brightness) refers to the physical value of the objective light intensity of the external environment measured at a specific time point, used as the basic data source for calculating the rate of light intensity fluctuation over a short period. The first ambient brightness sequence refers to a data set of multiple ambient brightness values ​​recorded and saved sequentially within a previously set time window, used to reflect the background noise characteristics of ambient light. The median absolute deviation is a statistical measure of data dispersion, representing the median of the absolute differences between each value in the dataset and the data median. In this application, it is used to filter out extreme abnormal noise in the first ambient brightness sequence and extract the true background fluctuation amplitude of ambient light. The preset sensitivity coefficient refers to a pre-configured numerical multiplier parameter in the device, used to proportionally amplify or reduce the background noise assessment value in the mathematical model, thereby adjusting the sensitivity boundary to changes in illumination. Its purpose is to constrain the wake-up sensitivity of the image acquisition device.

[0078] In practical applications, when the timeline of the image acquisition device is at any evaluation moment, the ambient light sensor of the image acquisition device samples and obtains the first ambient light brightness at the evaluation moment. The image acquisition device extracts the second ambient light brightness of the ambient light at the first historical acquisition moment from the storage unit, and simultaneously extracts the first ambient light brightness sequence continuously acquired within a preset historical time period. The image acquisition device reads the timestamps of the evaluation moment and the first historical acquisition moment, and calculates the time interval generated by the difference between the two. The absolute value of the difference between the first ambient light brightness and the second ambient light brightness is divided by the time interval to calculate the current ambient light brightness change rate. Next, the image acquisition device sorts all ambient light brightness values ​​in the first ambient light brightness sequence and extracts the median. The absolute value of the difference between each ambient light brightness value in the sequence and the extracted median is calculated to form an absolute deviation sequence. The absolute deviation sequence is sorted again and the median is extracted. The extracted median is determined as the median absolute deviation of the first ambient light brightness sequence. Finally, the image acquisition device retrieves a preset sensitivity coefficient, executes a multiplication instruction, multiplies the median absolute deviation by the preset sensitivity coefficient, and sets the calculated product value as the change rate threshold.

[0079] As an example, the rate of change in brightness The calculation formula can be found in formula (1): Formula (1) in, This represents the ambient brightness value collected at the current moment, which is the first ambient brightness at the moment to be evaluated. express The ambient brightness value collected before the time was used; here it is the second ambient brightness value at the first historical collection time.

[0080] Rate of change threshold The determination method can be found in formula (2): Formula (2) in, This indicates the preset sensitivity coefficient; Indicates the first ambient brightness sequence; This represents the median absolute deviation of the first ambient brightness sequence.

[0081] By using the brightness change rate instead of a fixed brightness difference, this method reduces invalid triggers caused by slowly changing lighting conditions such as sunrise and sunset, improving the image acquisition device's responsiveness to sudden, valid events. The change rate threshold is calculated using the absolute deviation of the median, mitigating the mathematical flaws of traditional mean or standard deviation algorithms, which are easily swayed by transient extreme noise (such as sudden flashes of car headlights at night or lightning). This extreme noise is isolated during median extraction, ensuring the generated change rate threshold accurately reflects the actual background fluctuations of the environment. This approach overcomes the limitations of static thresholds in adapting to changing environments, giving the image acquisition device noise resistance and trigger agility in complex outdoor environments.

[0082] In some embodiments, for the third one described above, before determining that the preset wake-up condition is met, the target wake-up time can be determined by the following method: acquiring a second ambient brightness sequence collected before the second historical acquisition time; the second historical acquisition time is the wake-up time preceding the time to be evaluated; constructing a short-term prediction model and a long-term trend model based on the second ambient brightness sequence; wherein, the short-term prediction model represents the weighted sum of ambient brightness over different time spans from the second historical acquisition time; the long-term trend model is determined based on the brightness change value within the first preset time window and the brightness change slope within the second preset time window; based on the short-term prediction model and the long-term trend model, predicting the second predicted ambient brightness at the time to be predicted after the second historical acquisition time; if the difference between the second predicted ambient brightness and the third ambient brightness corresponding to the second historical acquisition time is greater than a preset change amount, then using the time to be predicted as a reference, shifting forward by a preset time offset amount to obtain the pre-predicted target wake-up time.

[0083] Here, the second historical acquisition moment refers to the exact time point before the current judgment moment (e.g., the moment to be evaluated) when the image acquisition device was awakened from sleep mode to perform sensing operations. It serves as a time anchor point for future extrapolation and prediction. The second ambient brightness sequence refers to a set of data continuously recorded over a past period reflecting objective light intensity. The short-term prediction model is a mathematical evaluation structure reflecting recent changes in lighting inertia, used to capture the transient evolution of ambient light over a short time span. The long-term trend model is a mathematical evaluation structure reflecting the macroscopic direction of lighting evolution, used to extract the overall rise and fall trend of ambient light over a longer time scale. The moment to be predicted refers to a specific virtual evaluation node located in the future direction of the time axis, used as a verification point to check whether a drastic change in future lighting will occur. The preset change amount refers to the boundary parameter defining whether the difference between the predicted light intensity (i.e., the second predicted ambient brightness) and the reference light intensity (i.e., the third ambient brightness) constitutes a drastic environmental change. It is used to filter out normal gradual changes and identify sudden lighting events that require early intervention. The preset time offset is a pre-configured time difference value that needs to be deducted in advance. It is used to offset the internal time required for the image acquisition device to wake up and calculate parameters, ensuring that the image acquisition device is ready before the actual light change occurs.

[0084] In practical applications, when the image acquisition device predicts the target wake-up time at a second historical acquisition time prior to the time to be evaluated, it retrieves light intensity values ​​recorded chronologically before the second historical acquisition time and integrates them to form a second ambient brightness sequence. The image acquisition device performs dual-track parallel calculations on the second ambient brightness sequence. In one calculation track, the image acquisition device extracts discrete ambient brightness values ​​at different time spans from the second historical acquisition time, multiplies them by their corresponding weighting factors, and sums them to output a short-term prediction model. In the other calculation track, the image acquisition device extracts the brightness values ​​at the beginning and end of a first preset time window from the second ambient brightness sequence and calculates the difference. Simultaneously, it extracts continuous data points within the second preset time window to fit the brightness rise and fall slope, and combines these two values ​​to output a long-term trend model. The image acquisition device algebraically fuses and superimposes the outputs of the short-term prediction model and the long-term trend model to directly deduce the second predicted ambient brightness at the future time to be predicted. The image acquisition device also extracts the third ambient brightness acquired at the second historical acquisition time. The image acquisition device calculates the absolute difference by subtracting the third ambient brightness from the second predicted ambient brightness, and then compares this absolute difference with a preset change value stored internally. When the absolute difference is greater than the preset change value, the image acquisition device confirms that a sudden change in illumination will occur at the time to be predicted. The image acquisition device aligns the timestamp with the time to be predicted, subtracts a preset time offset along the past direction on the time axis, and the calculated new time point is established as the target wake-up time, which is written into the hardware alarm register of the image acquisition device to await triggering.

[0085] It is understandable that the prediction of the target wake-up time is carried out synchronously with the wake-up of the image acquisition device. That is, if the preset wake-up conditions are met at the time to be evaluated, thereby triggering the wake-up of the image acquisition device, the time to be evaluated can be used as a benchmark to predict at least one target wake-up time after the time to be evaluated.

[0086] As an example, a short-term forecasting model can be found in formula (3): Formula (3) in, This indicates the current time of prediction; in this case, it is the second historical data collection time. Indicates the time to be predicted; Indicates in The ambient brightness after time, that is, the short-term predicted ambient brightness component at the time to be predicted; This represents the brightness weighting coefficients for different time spans, and must satisfy the following conditions: ; This represents the ambient brightness value collected at the current moment, which here is the third ambient brightness value at the second historical collection moment; , This indicates the ambient brightness at different time spans from the second historical data collection time.

[0087] For a long-term trend model, please refer to formula (4): Formula (4) in, This indicates the current time of prediction; in this case, it is the second historical data collection time. It represents the long-term trend component, that is, the long-term trend value; Indicates the trend weighting coefficient; Indicates the first preset time window; This represents the brightness change value within the first preset time window; This represents the slope of brightness change within the second preset time window.

[0088] This approach introduces a feedforward pre-detection and temporal backoff reconstruction mechanism, upgrading the traditional passive response to ambient light changes to active predictive control. It integrates high-frequency transient light changes with low-frequency day-night cycles, improving the mathematical accuracy of environmental projection. By calculating the target wake-up time through backoff offset on the time axis, the latency required for hardware reset and parameter calculation is reduced. This temporal planning mechanism reduces overexposure or underexposure phenomena caused by sudden changes in lighting conditions, enabling low-power image acquisition devices to improve exposure convergence capabilities when facing sudden environmental changes.

[0089] See also Figure 3 The following explanation follows step 101 above.

[0090] In step 102, the first brightness feature of the first image is extracted, and the first brightness change value is determined based on the first brightness feature.

[0091] Here, the first brightness feature is used to characterize the overall illumination distribution of the environment in which the first image is located. Since the first image was acquired at a first sampling density, the pixel data volume of the first image is lower than that of the image at the target sampling density. Therefore, the first brightness feature focuses on macroscopic statistical information. In specific implementations, in addition to using the global average brightness of all pixels in the first image, the first brightness feature can also be configured as the median of the brightness values ​​of all pixels in the first image, the percentage of pixels in a specific brightness range, or statistical parameters such as the peak brightness extracted based on the brightness histogram of the first image. By extracting the above-mentioned first brightness feature, basic illumination information of the current environment can be obtained while maintaining a low data throughput. After extracting the first brightness feature, it is necessary to establish a numerical comparison relationship between the current acquisition time and the historical reference state to determine the first brightness change value. The first brightness change value is used to objectively measure the magnitude of the fluctuation of the current ambient light compared to the historical environment.

[0092] In some embodiments, Figure 4 This is a flowchart illustrating the method for determining the first brightness change value provided in an embodiment of this application. See also... Figure 4 Step 102 can be achieved through steps 1021 to 1024: In step 1021, the first global average brightness of the first image is determined and used as the first brightness feature.

[0093] Here, the first global average brightness can be the mathematical average of the brightness channel values ​​of all pixels in the first image, used as a benchmark feature for evaluating the current transient macroscopic ambient light intensity.

[0094] In practical applications, the image processing engine of the image acquisition device traverses all pixels at spatial coordinates within the first image, extracts the brightness value corresponding to each pixel, and sums all the extracted pixel brightness values ​​to obtain the total brightness value. Next, it obtains the total number of pixels in the first image, executes a division instruction to divide the total brightness value by the total number of pixels, and calculates the mathematical average, which is then determined as the first global average brightness. The image acquisition device can directly establish the first global average brightness as the first brightness feature for temporary storage.

[0095] In step 1022, the second global average brightness of the third image is obtained; wherein the third image is acquired at the previous wake-up moment when the image acquisition device switches from sleep state to wake-up state.

[0096] Here, the second global average brightness can be the mathematical average of the brightness channel values ​​of all pixels in the historically acquired third image, used as a historical reference benchmark for comparing current illumination changes. It should be noted that the calculation method for the second global average brightness can also refer to the calculation method for the third global average brightness described later; no limitation is made here.

[0097] In practical applications, the image acquisition device traces back along the timeline based on the current wake-up trigger time series label (i.e. the time to be evaluated), accurately locates the previous wake-up time, and directly retrieves and reads the second global average brightness data corresponding to the third image acquired and stored at the previous wake-up time.

[0098] In step 1023, the absolute value of the first difference between the first global average brightness and the second global average brightness is determined.

[0099] Here, the absolute value of the first difference is the result of taking the positive value after subtracting the two global average brightness values, which is used to quantify the absolute shift of the ambient light intensity within the two acquisition intervals.

[0100] In practical applications, the image acquisition device performs a subtraction operation to calculate the algebraic difference between the first and second global average brightness. Subsequently, the absolute value of this algebraic difference is taken, removing the positive and negative signs and directional attributes, and retaining only the pure offset magnitude value, which is determined as the absolute value of the first difference.

[0101] In step 1024, the ratio of the absolute value of the first difference to the preset maximum brightness value of the image acquisition device is used as the first brightness change value.

[0102] Here, the preset maximum brightness value refers to the upper limit of brightness quantization that the ambient light sensor or data bit width of the image acquisition device can output. It is used as the denominator factor in the normalization calculation to convert the absolute brightness difference into a relative fluctuation ratio. The first brightness change value refers to the relative environmental fluctuation ratio parameter after normalization processing. It is used to eliminate the evaluation error caused by the absolute brightness difference of the scene and serves as the comparison credential to trigger subsequent high-precision image acquisition logic.

[0103] In practical applications, the image acquisition device reads a preset maximum brightness value, performs a division operation using the absolute value of the first difference as the dividend and the preset maximum brightness value as the divisor. The resulting quotient is then established as the final output first brightness change value.

[0104] As an example, the formula for calculating the first brightness change value can be found in formula (5): Formula (5) in, This indicates the first change in brightness. This represents the first global average brightness; Indicates the second global average brightness; This indicates the preset maximum brightness value.

[0105] By employing the above method, in the initial stage when the image acquisition device is just awakened from sleep mode, a global average value extraction method with low computational complexity is used, reducing the clock cycles required to extract basic physical quantities. A preset maximum brightness value is introduced for division normalization, converting brightness jumps into a relative percentage scale. Regardless of whether the image acquisition device is in a bright or dark environment, the normalization process reduces the impact of ambient light base brightness differences on fluctuation assessment. This low-computation processing method allows the image acquisition device to assess ambient light change trends with low power consumption in the initial wake-up phase, providing a basis for determining whether more hardware resources need to be allocated for subsequent image acquisition.

[0106] See also Figure 3 The following explanation follows step 102 above.

[0107] In step 103, if the first brightness change value is greater than or equal to the first change threshold and less than the second change threshold, then the second image is acquired at the second sampling density.

[0108] Among them, the second change threshold is greater than the first change threshold, and the second sampling density is higher than the first sampling density.

[0109] Here, the first and second change thresholds represent the grading criteria set for the magnitude of changes in ambient light. The first change threshold distinguishes whether the ambient light is in a stable state or has changed, while the second change threshold distinguishes whether the ambient light has changed by a moderate amount or has undergone a significant switch in illumination. Therefore, in parameter configuration, the value of the second change threshold is greater than the value of the first change threshold.

[0110] If the judgment result shows that the first brightness change value is greater than or equal to the first change threshold and less than the second change threshold, it indicates that the overall brightness change of the current ambient light falls within a preset intermediate fluctuation range. Within this range, relying solely on the macroscopic statistical characteristics of the aforementioned first image may not accurately determine whether the brightness change is caused by a change in the overall ambient light or by the movement of local light sources in the field of view (e.g., local reflections from vehicle headlights or movement of local tree shadows). To avoid misjudgment and to avoid unnecessary power consumption caused by directly calling high-computing-power exposure prediction algorithms, the device is configured to initiate a verification mechanism, i.e., control the image acquisition device to acquire a second image at a second sampling density.

[0111] The second sampling density represents an intermediate image resolution between the initial pathfinding acquisition and the final imaging acquisition. To meet the requirements of verifying local features while maintaining low power consumption, the second sampling density is higher than the first sampling density and lower than the target sampling density used for the final video stream image output. By acquiring the second image using the second sampling density, the image acquisition device obtains more image pixels than the first image. The increased pixel data allows the second image to retain specific spatial distribution information and local texture boundaries within the scene.

[0112] See also Figure 3 The following explanation follows step 103 above.

[0113] In step 104, the second brightness feature of the second image is extracted, and the second brightness change value is determined based on the second brightness feature.

[0114] Here, the second brightness feature characterization is based on a set of image parameters extracted from the second image that reflect the details of illumination distribution and local light saturation in different areas of the scene. Because the sampling density of the second image is higher than that of the first image, the image acquisition device can extract more accurate brightness information for different areas from the second image.

[0115] The methods for extracting the second brightness feature may include, but are not limited to: using the region-weighted brightness and the proportion of overexposed pixels as the second brightness feature; dividing the second image into a foreground subject region and a background region based on a preset image segmentation algorithm (such as threshold segmentation or edge detection), and extracting the brightness histogram information of the foreground subject region and the background region as the second brightness feature respectively; or, statistically analyzing the dispersion of pixel clusters in the second image that are in the highlight threshold range and the shadow threshold range (such as brightness variance), and using it as the second brightness feature characterizing the scene light contrast.

[0116] In some embodiments, Figure 5 This is a flowchart illustrating the method for determining the second brightness change value provided in an embodiment of this application. See also... Figure 5 Step 104 can be achieved through steps 1041 to 1049: In step 1041, multiple regions of interest (ROIs) are pre-divided for the second image, and the initial average brightness and region weight of each ROI are determined.

[0117] Here, Region of Interest (ROI) is a standard term in computer vision, referring to a pre-defined spatial pixel block (such as a road or gate in the center of an image) that has significance for business intent. Its purpose is to isolate background interference at the edges and focus the lighting status of the core subject during brightness evaluation. The initial average brightness of the region refers to the simple arithmetic mean of the brightness channel values ​​of all pixels within a single ROI, used to characterize the absolute basic brightness of that single ROI. Region weights are pre-assigned product coefficients to different ROIs, used to characterize the relative importance of different ROIs within the overall image.

[0118] In practical applications, the image acquisition device reads the coordinate configuration information of multiple regions of interest (ROIs) that have been pre-defined for the second image, extracts all pixels within each ROI based on the coordinate configuration, calculates the average brightness of these pixels to determine the initial average brightness of each ROI, and simultaneously reads the region weights corresponding to each ROI in the configuration table.

[0119] In step 1042, for each region of interest, the average brightness of the region of interest is determined based on the initial average brightness of the region and the region weight.

[0120] Here, the regional average brightness is the equivalent brightness value obtained by mathematically weighting the initial regional average brightness with regional weights. Its purpose is to reflect the true proportion of the contribution of the region of interest in the global brightness assessment.

[0121] In practical applications, within each region of interest, the image acquisition device performs a multiplication operation between the extracted initial average brightness of the region and the corresponding region weight. The result of the multiplication is determined as the average brightness of the region of interest.

[0122] In step 1043, the average of the regional average brightness of multiple regions of interest is used as the third global average brightness of the second image.

[0123] Here, the third global average brightness refers to the comprehensive weighted brightness representing the overall viewpoint of the image, calculated by fusing and superimposing all the divided regions of interest according to their regional weights. Its purpose is to serve as the final brightness evaluation benchmark for the second image of the current frame after eliminating background interference.

[0124] In practical applications, all regions of interest are traversed, and the calculated average brightness of each region is summed to obtain the fused result, which is then used as the third global average brightness of the second image.

[0125] In step 1044, the total number of pixels in the second image and the number of overexposed pixels with brightness values ​​greater than a preset brightness limit are determined, and the ratio of the number of overexposed pixels to the total number of pixels is taken as the overexposed pixel ratio of the second image.

[0126] Here, the number of overexposed pixels refers to the total number of discrete pixels in the image array whose brightness quantization value exceeds the preset brightness limit. Its purpose is to quantify the area in the image where texture details are lost due to direct, strong local light. The overexposed pixel ratio refers to the percentage of overexposed pixels in the total number of pixels in the entire second image.

[0127] In practical applications, the entire pixel array of the second image is traversed, and the total number of pixels is counted. During the traversal, the brightness value of each pixel is fed into a comparator and compared with a preset brightness limit. If the brightness value is greater than the preset limit, an internal counter is incremented. After the traversal is complete, the value of the counter is determined as the number of overexposed pixels. Then, a division operation is performed by dividing the number of overexposed pixels by the total number of pixels, and the quotient is used as the overexposed pixel ratio of the second image.

[0128] In step 1045, the third global average brightness and the overexposed pixel ratio are used as the second brightness feature.

[0129] In step 1046, the second global average brightness of the third image is obtained; wherein the third image is acquired at the previous wake-up moment when the image acquisition device switches from sleep state to wake-up state.

[0130] For a description of the third image and the second global average brightness, please refer to the aforementioned related embodiments, which will not be repeated here.

[0131] In step 1047, the absolute value of the second difference between the third global average brightness and the second global average brightness is determined.

[0132] Here, the absolute value of the second difference refers to a purely physical quantity that takes the positive value after subtracting the third global average brightness from the second global average brightness. Its purpose is to measure the absolute change in brightness of the core area of ​​interest.

[0133] In practical applications, the image acquisition device calculates the difference between the currently extracted third global average brightness and the historical second global average brightness, discards the sign bit of the difference result, and takes the absolute value to determine the second difference absolute value.

[0134] In step 1048, the ratio of the absolute value of the second difference to the preset maximum brightness value of the image acquisition device is used as the initial brightness change value.

[0135] Here, the initial brightness change value refers to the preliminary relative fluctuation ratio after the absolute value of the second difference is normalized. It is used to rate the basic brightness jump without considering extreme overexposure conditions.

[0136] In practical applications, the image acquisition device divides the absolute value of the second difference by the preset maximum brightness value fixed by the image acquisition device, and the resulting basic relative fluctuation value is used as the initial brightness change value.

[0137] In step 1049, the initial brightness change value is corrected based on the overexposed pixel ratio to obtain the second brightness change value.

[0138] Here, the second brightness change value is a comprehensive evaluation parameter obtained by correcting the initial brightness change value based on the overexposed pixel ratio. It characterizes the magnitude of brightness change in the region of interest of the second image relative to the images acquired at historical times. In practical applications, the image acquisition device retrieves the overexposed pixel ratio previously stored in the register and uses it as a penalty gain factor to perform a fusion operation with the initial brightness change value (e.g., direct multiplication or conditional step amplification) to complete the correction operation on the initial brightness change value. The final result of the correction output is determined as the second brightness change value.

[0139] As an example, the formula for calculating the third global average brightness can be found in formula (6): Formula (6) in, This represents the third global average brightness; This represents the region weight corresponding to each region of interest; This represents the initial average brightness of each region of interest.

[0140] Overexposed pixel ratio The calculation formula can be found in formula (7): Formula (7) in, This indicates the preset maximum brightness value; This indicates the number of pixels whose brightness value exceeds 95% of the preset maximum brightness value; This represents the total number of pixels in the second image.

[0141] Through the above approach, in the medium-power evaluation after the image acquisition device is woken up, the technical implementation abandons the global undifferentiated mean algorithm in the first-level evaluation and instead adopts a weighted spatial decoupling method. At medium resolution, it uses region-weighted brightness calculation, which reduces the interference of irrelevant backgrounds (such as sky and reflections) on brightness change detection. Simultaneously, it introduces the overexposed pixel ratio as a correction factor; when local overexposure is severe, even if the weighted average brightness change is not significant, it can still improve the second brightness change value. This feature extraction mode, combining subject weighting and overexposure fallback, enables the image acquisition device to achieve both superior anti-interference capabilities compared to global coarse evaluation and the ability to prevent missed detections in extremely high-exposure and saturated scenes at medium resolution with extremely low computational overhead.

[0142] See also Figure 3 The following explanation follows step 104 above.

[0143] In step 105, if the second brightness change value is greater than or equal to the second change threshold, the brightness of the ambient light at the target acquisition time is predicted to obtain the first predicted ambient brightness.

[0144] After determining the second brightness change value, which has been verified for accuracy, the image acquisition device compares the second brightness change value with the second change threshold. If the result shows that the second brightness change value is greater than or equal to the second change threshold, it indicates that the ambient light of the current scene has undergone a significant change beyond the normal fluctuation range (e.g., transitioning from day to night, or the main indoor light source being suddenly turned on or off). When such a significant change occurs, if a conventional feedback automatic exposure algorithm is used, the image acquisition device needs to continuously capture multiple frames of images to gradually bring the exposure parameters closer to the correct value. To avoid the delay and power consumption caused by such multi-frame image acquisition in low-power operating mode, the system abandons repeated attempts based on the brightness at the current moment and instead triggers a prediction mechanism, that is, predicting the brightness of the ambient light at the target acquisition moment.

[0145] It is understood that the target acquisition time is not the current moment when the evaluation calculation is being performed, but rather refers to a future point in time on the timeline after the current acquisition time, based on which the image acquisition device will output image data. Specifically, since the issuance of exposure control commands by the image sensor, sensor integration exposure, and data readout all require system time differences (e.g., tens of milliseconds or longer), the target acquisition time characterizes the time point after the image acquisition device applies the latest exposure parameters.

[0146] In some embodiments, Figure 6 This is a flowchart illustrating the first method for predicting ambient brightness provided in this application embodiment. See also... Figure 6The step 105, "predicting the brightness of the ambient light at the target acquisition time to obtain the first predicted ambient brightness", can be achieved through steps 1051 to 1052.

[0147] In step 1051, the first acquisition time when acquiring the second image is determined, and the third ambient brightness sequence acquired before the first acquisition time is obtained.

[0148] Here, the third ambient brightness sequence is a set of data reflecting the intensity of objective ambient light that is continuously recorded on the timeline earlier than the first acquisition time. Its purpose is to provide the historical trajectory of light evolution over a period of time.

[0149] In practical applications, the image acquisition device locks the absolute value of the current hardware tick counter the instant the second image is read, marking this locked absolute value as the first acquisition moment. Then, using this first acquisition moment as a time cutoff anchor point, the image acquisition device backtracks along the timeline, extracting multiple discrete ambient brightness values ​​continuously measured at a fixed sampling period before this time cutoff anchor point, and combining them to form a third ambient brightness sequence.

[0150] In step 1052, the environmental brightness of each element in the fourth and third environmental brightness sequences corresponding to the first acquisition time is weighted and summed to obtain the first predicted environmental brightness corresponding to the target acquisition time, where the target acquisition time is the next image acquisition time after the first acquisition time.

[0151] Here, the fourth ambient brightness is the absolute value of the ambient light intensity measured and extracted by the image acquisition device at the first acquisition moment. The target acquisition moment is a future time point after the first acquisition moment, where the image acquisition device plans to output the next frame of a clear image with the target sampling density. The first predicted ambient brightness is the expected intensity of the ambient light at the future target acquisition moment, derived by the image acquisition device using a mathematical model.

[0152] In practical applications, the image acquisition device extracts the currently observed fourth ambient brightness and sequentially retrieves each discrete ambient brightness contained in the third ambient brightness sequence. It reads a pre-configured weight allocation matrix and assigns a unique weight coefficient to each ambient brightness in both the fourth and third ambient brightness sequences. The fourth ambient brightness is multiplied by its assigned weight coefficient to obtain the first intermediate product. This process is repeated, multiplying each ambient brightness in the third ambient brightness sequence by its corresponding weight coefficient to generate multiple historical intermediate products. Finally, a continuous summation instruction is executed, mathematically adding the first intermediate product to all historical intermediate products. The final sum is directly determined as the first predicted ambient brightness. Since the first predicted ambient brightness is a physical projection forward along the timeline based on historical trends, pointing to future physical states, the image acquisition device, in terms of temporal logic, maps the time label corresponding to the first predicted ambient brightness to the next image acquisition time after the first acquisition time, thus establishing the target acquisition time.

[0153] As an example, the formula for calculating the first predicted ambient brightness can be found in formula (8): Formula (8) in, This represents the first predicted ambient brightness at the time of target acquisition. This indicates the current first data collection moment; This indicates the next image acquisition time after the first acquisition time, i.e., the target acquisition time. This represents the fourth ambient brightness at the first acquisition time. : Represents the historical ambient brightness in the extracted third ambient brightness sequence. This represents the weighting coefficient assigned to each ambient brightness level, and it must follow the principle of assigning higher weights to recent values ​​than to distant values. .

[0154] By introducing a time-series prediction mechanism, the traditional passive exposure response is transformed into active time-series feedforward control. When faced with sudden changes in illumination exceeding a threshold, traditional automatic exposure algorithms rely on the exposure result of the current frame to adjust parameters for the next frame, easily leading to an exposure response that lags behind the rate of change in physical illumination. By introducing a weighted summation time-series prediction mechanism, the image acquisition device incorporates the historical trends of illumination fluctuations. By assigning differentiated weights to data from different time dimensions, the illumination intensity that will be encountered when actually performing an exposure action in the future is accurately estimated, improving the exposure convergence capability of the image acquisition device.

[0155] See also Figure 3 The following explanation follows step 105 above.

[0156] In step 106, based on the first predicted ambient brightness, the preset desired ambient brightness, and the first exposure parameters when acquiring the first image, the second exposure parameters are determined, and the image is acquired based on the second exposure parameters and the target sampling density.

[0157] The target sampling density is higher than the second sampling density.

[0158] Here, the first exposure parameter represents the initial photosensitive configuration used by the image acquisition device when it is first activated and acquires the first image. Specifically, it can include a combination of the exposure time of the photosensitive element, the analog gain coefficient, and the digital gain coefficient. The desired ambient brightness represents the standard brightness baseline that the output image should achieve. This value is used to ensure that the image does not exhibit large-area saturation overflow or loss of detail in dark areas. For example, the desired ambient brightness can be set to 45% of the maximum value of the image sensor's quantization range.

[0159] In some embodiments, Figure 7 This is a flowchart illustrating the method for determining the second exposure parameter provided in an embodiment of this application. See also... Figure 7 The step 106, "determine the second exposure parameter based on the first predicted ambient brightness, the preset expected ambient brightness, and the first exposure parameter when acquiring the first image", can be achieved through step 1061.

[0160] In step 1061, the ratio of the first predicted ambient brightness to the desired ambient brightness is determined, and the product of the ratio and the first exposure parameter is used as the second exposure parameter.

[0161] Here, the desired ambient brightness is a pre-configured target reference brightness baseline value to ensure that the final image is not severely overexposed or underexposed. Its purpose is to provide a benchmark anchor point when calculating exposure parameter adjustment ratios. The second exposure parameter is a new photosensitive configuration data derived through mathematical algebraic calculations, applicable to the target acquisition time in future new lighting environments.

[0162] In practical applications, the image acquisition device can read the desired ambient brightness value, retrieve the first predicted ambient brightness calculated in the previous steps, and perform a division operation using the desired ambient brightness value as the dividend and the first predicted ambient brightness value as the divisor. The image acquisition device then temporarily stores the quotient as the ratio for adjustment. Next, the image acquisition device reads the first exposure parameter that has not yet been modified in the current state, performs a multiplication operation between the first exposure parameter and the ratio, and directly establishes the calculated product value as the second exposure parameter, completing the update calculation of the old and new photosensitivity parameters. Finally, it uses the second exposure parameter and the target sampling density to perform image acquisition.

[0163] In this embodiment, the image acquisition device triggers wake-up when a preset wake-up condition is met. It then acquires images at a first sampling density and a second sampling density, extracting the corresponding brightness change values. When the brightness change value reaches a second threshold, the brightness of the ambient light at the target acquisition moment is predicted. Based on the predicted ambient brightness, the desired ambient brightness, and the initial exposure parameters, the second exposure parameters are directly calculated. Finally, image acquisition is performed using the highest target sampling density and the second exposure parameters. This scheme balances environmental perception agility with hardware resource consumption through hierarchical state evaluation based on increasing sampling density and exposure parameter prediction. Specifically, this method first reduces background noise interference by using gradient sampling density, thus reducing computational resource consumption when ambient light does not change significantly. Second, it sets a forward-looking data boundary for exposure control by predicting the brightness at the target acquisition moment, which is introduced by significant light changes. This approach solves the problem in related technologies where passive probing using a single full-resolution image across multiple frames leads to an inability to adapt to dynamic lighting environment changes and an inability to balance the battery life of low-power devices with exposure response speed. This application's hierarchical sampling method reduces unnecessary computations caused by unnecessarily changing light conditions, allowing the system's computing power to be highly focused on core computational changes requiring intervention. Furthermore, predictive exposure estimation can find an exposure parameter update direction that balances current hardware timing and future lighting trends. This makes image acquisition equipment more efficient and intelligent, reducing power consumption while improving the convergence capability of automatic exposure.

[0164] In some embodiments, after determining the first brightness change value based on the first brightness feature, if the first brightness change value is greater than the second change threshold, the brightness of the ambient light at the first acquisition time can be directly predicted to obtain the third predicted ambient brightness. Based on the third predicted ambient brightness, the expected ambient brightness, and the first exposure parameters, the fourth exposure parameters are determined, and image acquisition is performed at or after the first acquisition time based on the fourth exposure parameters and the target sampling density.

[0165] In some embodiments, after determining the first brightness change value based on the first brightness feature, if the first brightness change value is less than the first change threshold, the first exposure parameter is maintained, and image acquisition is performed based on the first exposure parameter and the target sampling density; in response to the image acquisition end command, the image acquisition device is triggered to switch from the wake-up state to the sleep state.

[0166] Here, the image acquisition end command is an interrupt feedback signal sent by the underlying hardware module (such as an image sensor or image signal processing bus) to the main control chip after it has completely output one or more frames of image data that meet the preset sampling density.

[0167] In practical applications, when the value of the first brightness change is less than the value of the first change threshold, it can be determined that the light intensity of the current external physical environment has not undergone a substantial change worthy of adjusting the exposure strategy. Based on this determination, no new parameter update instructions can be issued to the photosensitive components, thereby forcibly retaining and maintaining the first exposure parameters used in the initial wake-up phase. Immediately afterwards, the image sensor can directly switch the pixel array readout mode to the pre-set highest resolution configuration, i.e., the target sampling density, and continue to perform image acquisition using the unmodified first exposure parameters at this target sampling density, completing the physical capture of one or more frames of a regular clear image. After the underlying image sensor completes image acquisition at the target sampling density, it can send an image acquisition end command to the main control chip. Upon receiving the image acquisition end command, the main control chip confirms that all core business processes that must be performed within the current wake-up cycle have been completed, immediately initiates the power-down sequence, and only retains the extremely low-power wake-up monitoring peripheral to continue operating, thereby triggering the image acquisition device to exit the high-power wake-up state and switch back to the low-power sleep state.

[0168] By determining that the value of the first brightness change is less than a first change threshold, subsequent feature extraction (such as defining the region of interest) and exposure estimation calculations are terminated. This method of directly maintaining the first exposure parameters reduces the internal clock cycle overhead of performing floating-point operations and parameter iterations. Without modifying the first exposure parameters, the image is output at the target sampling density, and the main circuit power is cut off by a termination command, shortening the wake-up time of the image acquisition device in stable lighting conditions. This processing strategy reduces the overall power consumption of the image acquisition device under stable monitoring conditions, thereby improving the standby life of the image acquisition device.

[0169] In some embodiments, the second brightness feature includes the third global average brightness and the overexposed pixel ratio of the second image; after determining the second brightness change value based on the second brightness feature, if the second brightness change value is less than the second change threshold, an exposure offset is determined based on the third global average brightness and the overexposed pixel ratio; the first exposure parameter is adjusted based on the exposure offset to obtain the third exposure parameter, and image acquisition is performed based on the third exposure parameter and the target sampling density; in response to the image acquisition end command, the image acquisition device is triggered to switch from a wake-up state to a sleep state.

[0170] Here, the third exposure parameter is a photosensitive configuration combination synthesized by superimposing the exposure offset on the first exposure parameter inherited when the image acquisition device is first woken up, in order to adapt to the current ambient light intensity after slight changes.

[0171] In practical applications, the image acquisition device obtains the calculated second brightness change value and a pre-configured second change threshold, and performs a numerical comparison operation. When the comparison result indicates that the second brightness change value is strictly less than the second change threshold, the image acquisition device extracts the third global average brightness and overexposed pixel ratio contained in the second brightness feature. It then divides the third global average brightness by the preset desired brightness to obtain the basic adjustment ratio, and simultaneously evaluates the magnitude of the overexposed pixel ratio. The overexposed pixel ratio is then used to perform a safety limit or negative attenuation operation on the basic adjustment ratio, and the final determined exposure offset is output. Next, the image acquisition device reads the first exposure parameters used in the initial wake-up phase, performs a mathematical merging operation with the exposure offset, and generates a finely adjusted third exposure parameter. Under the target sampling density configuration, the image acquisition device completes the final image acquisition according to the third exposure parameter. After the image acquisition is completed and an image acquisition end command is triggered, the image acquisition device immediately initiates a power-off timing mechanism, directly switching to and maintaining a very low-power sleep state.

[0172] In scenarios with moderate fluctuations in ambient light, this method avoids the main control chip performing full-space exposure prediction modeling. Instead, it uses the third global average brightness and overexposed pixel ratio generated in the early assessment as features to calculate the exposure offset. Based on this offset, the basic parameters are corrected using a one-time multiplicative superposition, reducing the process of multi-frame image feedback adjustment and allowing the photosensitive element to adapt to the new lighting environment more quickly. Combined with a control strategy that outputs images at the target sampling density after parameter application and then shuts down for sleep after acquisition, this closed-loop mechanism of single-step fine-tuning and image output enables the image acquisition device to achieve exposure compensation for ambient light fluctuations with low power consumption, extending the battery life of low-power devices.

[0173] In some embodiments, the image acquisition device can also acquire historical environmental data; classify the historical environmental data using a clustering algorithm to establish an environmental pattern database; record successful exposure convergence records; establish a convergence decision tree model based on the successful exposure convergence records; and determine whether to trigger a switch from a dormant state to a wake-up state based on the convergence decision tree model.

[0174] The following will describe exemplary applications of the embodiments of this application in real-world application scenarios.

[0175] Always-On Video (AOV) monitors (i.e., image acquisition devices) have two operating modes: constant power mode and AOV mode. In constant power mode, the main controller never loses power and continuously outputs a video stream. AOV mode operates on a cycle of T seconds. Every T seconds, the main controller is woken up to output one frame of video, and then goes into sleep mode for the remaining time. For example, if the AOV wake-up cycle T is 2 seconds, it means that one frame is output every two seconds, after which the main controller goes into sleep mode. If the ambient light intensity changes significantly, after-effect (AE) convergence is required to ensure that the output video stream is not overexposed or underexposed. Generally, AE convergence requires switching to constant power mode, which consumes more power. The AE convergence methods in related technologies typically suffer from the following problems: Issue 1: Lack of dedicated optimizations for AOV mode.

[0176] The related technologies treat constant power mode and AOV mode equally, failing to fully utilize the predictive and intermittent characteristics of AOV mode, and thus failing to maximize energy-saving effects while ensuring functionality.

[0177] Question 2: The wake-up strategy is too simple.

[0178] The existing technologies rely solely on fixed time intervals for wake-up, failing to dynamically adjust to changes in the actual environment. This results in excessively frequent wake-ups when the environment is stable and delayed responses when the environment changes drastically. A single time-driven mechanism cannot adapt to dynamic environments, exhibiting delayed responses during drastic changes and excessively frequent wake-ups even when the environment is stable.

[0179] Question 3: Convergence path is fixed.

[0180] Regardless of the magnitude of environmental changes, the same convergence process is executed. This wastes computational resources for small changes and is not fast enough for large changes.

[0181] Question 4: Lack of adaptive learning mechanism.

[0182] It cannot self-optimize based on usage scenarios and user habits, resulting in stagnant long-term operational efficiency. Each wake-up is handled as an independent event, failing to optimize future behavior based on historical experience and thus unable to adapt to user habits.

[0183] Question 5: Lack of predictive ability.

[0184] Image acquisition devices based on related technologies can only passively respond to environmental changes and cannot make predictive adjustments based on historical data and environmental trends.

[0185] Therefore, the purpose of this application is to overcome the shortcomings of related technologies and provide a prediction-driven AOV fast AE convergence method, namely an exposure control method. By constructing four core modules—a multi-dimensional intelligent wake-up mechanism, a three-level progressive sampling convergence engine, a predictive convergence algorithm, and adaptive learning optimization—the following technical effects are achieved: significantly reducing power consumption and extending battery life; improving AE convergence speed and enhancing user experience; and enhancing intelligence to adapt to complex and ever-changing usage environments.

[0186] Specifically, the image acquisition device proposed in this application embodiment has the following core mechanisms: a multi-dimensional intelligent wake-up mechanism, a three-level progressive sampling convergence mechanism, a predictive convergence control mechanism, and an adaptive learning optimization mechanism.

[0187] First, a detailed explanation of the multi-dimensional intelligent wake-up mechanism will be provided.

[0188] The multi-dimensional intelligent wake-up mechanism includes three wake-up methods: passive wake-up, active wake-up, and predictive wake-up.

[0189] For the passive wake-up method, a basic wake-up cycle can be set, which is controlled by a timer. Each time the timer expires, a wake-up command is generated, and the image acquisition device enters the wake-up state.

[0190] For the active wake-up method, an ambient light sensor can be deployed to continuously monitor the ambient brightness and calculate the brightness change rate in real time, as shown in formula (1). When the brightness change rate exceeds the dynamic threshold (change rate threshold, see formula 2), active wake-up is triggered immediately.

[0191] For the predictive wake-up method, a predictive model can be established based on historical ambient light data, such as a short-term predictive model (see Formula 3) and a long-term trend model (see Formula 4). When the prediction error (absolute difference) exceeds the preset change amount, the image acquisition device is woken up in advance by τ_pre time (preset time offset).

[0192] Next, we will explain the three-level progressive sampling convergence mechanism.

[0193] Level 1: Quick and rough assessment.

[0194] Upon waking, a 1 / 16 thumbnail (first image) is first output, and the average brightness change (first brightness change value) is calculated. If the average brightness change is less than a first threshold (first change threshold), the current exposure parameters (first exposure parameters) are maintained, and the system can directly enter sleep mode.

[0195] Level 2: Medium precision verification.

[0196] When the average brightness change is greater than or equal to the first threshold and less than the second threshold (second change threshold), a 1 / 4 medium-resolution image (second image) is output for weighted brightness analysis and saturation detection to obtain the brightness change (second brightness change value). If the brightness change is less than the second threshold, fine-tuning convergence is performed in AOV mode.

[0197] Level 3: High-precision complete convergence.

[0198] When the brightness change exceeds the second threshold, switch to constant power mode, output a full-resolution image (image acquired based on the target sampling density), and execute the standard AE convergence algorithm until complete convergence, that is, the sampling predictive convergence control mechanism performs exposure convergence.

[0199] Here, we will explain the predictive convergence control mechanism.

[0200] Referring to formula (8), an ambient light change prediction model is established, and the weighted moving average method is used to predict the brightness value at the next moment, i.e., the first predicted ambient brightness at the target acquisition time. Then, based on the AE parameters required for brightness prediction: Exposure_pred = Exposure_current (L_target / L_pred), where Exposure_pred represents the predicted exposure parameters (e.g., the second exposure parameter), Exposure_current represents the current exposure parameters (the first exposure parameter), L_target is the target brightness value (i.e., the expected brightness value), and L_pred is the predicted brightness value (the first predicted ambient brightness). This method uses the predicted brightness to infer the required exposure parameters, achieving pre-convergence.

[0201] Finally, the adaptive learning optimization mechanism will be explained.

[0202] The adaptive learning optimization mechanism includes: learning environmental change characteristics by classifying historical environmental data using clustering algorithms to establish an environmental pattern database, including daytime stable patterns, nighttime abrupt change patterns, and transitional period patterns; and periodic adaptive adjustment by monitoring the effectiveness of the most recent N wake-ups, calculating the effective wake-up rate, and dynamically adjusting the preset time period based on the effective wake-up rate.

[0203] For the method of dynamically adjusting the base period (preset time period), please refer to formula (9): Formula (9) in, Indicates the adjusted base period; Indicates the current base cycle; Indicates the adjustment factor; This indicates the preset target effective wake-up rate; This indicates the effective wake-up rate.

[0204] The adaptive cycle adjustment is used to: extend the cycle when the efficiency is low and shorten the cycle when the efficiency is high.

[0205] The adaptive learning optimization mechanism also includes: accumulating convergence experience, recording successful cases of each convergence, building a convergence decision tree model, and guiding the handling of similar scenarios in the future.

[0206] Figure 8 This is a schematic diagram of the second process of the exposure control method provided in the embodiments of this application. See also... Figure 8 After the device (image acquisition device) starts up and initializes its parameters, it enters AOV mode. When wake-up is triggered by any of the passive wake-up, active wake-up, or predictive wake-up methods, a 1 / 16 thumbnail is output, and the first brightness change value is calculated. Then, it is determined whether the first brightness change value is less than the first threshold (i.e., the first variable threshold): if so, the original exposure parameters (first exposure parameters) are maintained, and historical data updates and adaptive cycle adjustments are performed sequentially. Finally, the image acquisition device enters sleep mode. If the first brightness change value is not less than the first threshold (i.e., it is determined not to be), a 1 / 4 medium image (second image) is output, and the second brightness change value is calculated. Then, it is determined whether the second brightness change value is less than the second threshold (i.e., the second variable threshold): if so, the exposure parameters are fine-tuned, and historical data updates and adaptive cycle adjustments are performed sequentially. Finally, the image acquisition device enters sleep mode. If the second brightness change value is not less than the second threshold (i.e., it is determined not to be), the device switches to constant power mode, performs full-resolution convergence and records convergence experience sequentially, then returns to AOV mode, and finally, the image acquisition device enters sleep mode.

[0207] As can be seen, the exposure control method of this application embodiment has the following advantages: power consumption is significantly reduced, and the power consumption during AE convergence is reduced through intelligent wake-up and hierarchical processing; response speed is improved, the average number of convergence frames can be reduced from 4.2 frames to 1.8 frames, the response time is shortened by 72%, and the user experience is greatly improved; the level of intelligence is enhanced, with environmental prediction and self-learning capabilities, which can adapt to different usage scenarios and continuously optimize performance; stability is improved, and through multi-dimensional wake-up mechanism and anomaly handling strategy, it exhibits stronger robustness in complex environments.

[0208] The following description continues to illustrate the exemplary structure of the exposure control device 455 provided in the embodiments of this application as a software module. In some embodiments, such as... Figure 2 As shown, the software module stored in the exposure control device 455 in the memory 450 may include: The wake-up control module 4551 is used to trigger the image acquisition device to switch from sleep state to wake-up state if the preset wake-up conditions are met while the image acquisition device is in sleep state.

[0209] Image processing module 4552 is used to acquire a first image at a first sampling density in a wake-up state; extract a first brightness feature of the first image and determine a first brightness change value based on the first brightness feature; if the first brightness change value is greater than or equal to a first change threshold and less than a second change threshold, then acquire a second image at a second sampling density; wherein the second change threshold is greater than the first change threshold and the second sampling density is higher than the first sampling density; extract a second brightness feature of the second image and determine a second brightness change value based on the second brightness feature.

[0210] The prediction module 4553 is used to predict the ambient light brightness at the target acquisition time if the second brightness change value is greater than or equal to the second change threshold, and obtain the first predicted ambient brightness; and determine the second exposure parameters based on the first predicted ambient brightness, the preset expected ambient brightness and the first exposure parameters when acquiring the first image.

[0211] The image processing module 4552 is also used to acquire images based on the second exposure parameters and the target sampling density; wherein the target sampling density is higher than the second sampling density.

[0212] In some embodiments, the wake-up control module 4551 is further configured to determine whether the preset wake-up condition is met by at least one of the following: for any time to be evaluated during the period when the image acquisition device is in a sleep state, if the time interval between the time to be evaluated and the previous wake-up time reaches a preset time period, the preset wake-up condition is determined to be met; if the rate of change of ambient light brightness at the time to be evaluated is greater than the rate of change threshold, the preset wake-up condition is determined to be met; if the time to be evaluated reaches a pre-predicted target wake-up time, the preset wake-up condition is determined to be met.

[0213] In some embodiments, the wake-up control module 4551 is further configured to acquire a first ambient brightness of the ambient light at the time to be evaluated, a second ambient brightness of the ambient light at the first historical acquisition time, and a first ambient brightness sequence acquired within a preset historical period; determine the brightness change rate based on the first ambient brightness, the second ambient brightness, and the time interval between the time to be evaluated and the first historical acquisition time; determine the median absolute deviation of the first ambient brightness sequence, and use the product of the median absolute deviation and a preset sensitivity coefficient as a change rate threshold; wherein the preset sensitivity coefficient is used to constrain the wake-up sensitivity of the image acquisition device.

[0214] In some embodiments, the wake-up control module 4551 is further configured to acquire a second ambient brightness sequence acquired before the second historical acquisition time; the second historical acquisition time is the wake-up time preceding the time to be evaluated; construct a short-term prediction model and a long-term trend model based on the second ambient brightness sequence; wherein, the short-term prediction model represents the weighted sum of ambient brightness over different time spans from the second historical acquisition time; the long-term trend model is determined based on the brightness change value within a first preset time window and the brightness change slope within a second preset time window; based on the short-term prediction model and the long-term trend model, predict the second predicted ambient brightness at the time to be predicted after the second historical acquisition time; if the difference between the second predicted ambient brightness and the third ambient brightness corresponding to the second historical acquisition time is greater than a preset change amount, then the time to be predicted is used as a reference, and a preset time offset is shifted forward to obtain the pre-predicted target wake-up time.

[0215] In some embodiments, the image processing module 4552 is further configured to determine a first global average brightness of the first image and use the first global average brightness as a first brightness feature; acquire a second global average brightness of the third image; wherein the third image is acquired at the previous wake-up time when the image acquisition device switches from a sleep state to a wake-up state; determine a first absolute value of the difference between the first global average brightness and the second global average brightness; and use the ratio of the first absolute value of the difference to the preset maximum brightness value of the image acquisition device as a first brightness change value.

[0216] In some embodiments, the image processing module 4552 is further configured to maintain the first exposure parameter if the first brightness change value is less than the first change threshold, and perform image acquisition based on the first exposure parameter and the target sampling density; the wake-up control module 4551 is further configured to trigger the image acquisition device to switch from the wake-up state to the sleep state in response to the image acquisition end command.

[0217] In some embodiments, the image processing module 4552 is further configured to: determine multiple regions of interest (ROIs) pre-divided for the second image; determine the initial average brightness and region weight of each ROI; determine the average brightness of each ROI based on the initial average brightness and region weight; take the average of the average brightness of the multiple ROIs as the third global average brightness of the second image; determine the total number of pixels in the second image and the number of overexposed pixels with brightness values ​​greater than a preset brightness limit, and take the ratio of the number of overexposed pixels to the total number of pixels as the overexposed pixel ratio of the second image; take the third global average brightness and the overexposed pixel ratio as the second brightness feature; acquire the second global average brightness of the third image; wherein the third image is acquired at the previous wake-up time when the image acquisition device switches from the sleep state to the wake-up state; determine the absolute value of the second difference between the third global average brightness and the second global average brightness; take the ratio of the absolute value of the second difference to the preset maximum brightness value of the image acquisition device as the initial brightness change value; and correct the initial brightness change value based on the overexposed pixel ratio to obtain the second brightness change value.

[0218] In some embodiments, the second brightness feature includes the third global average brightness and the overexposed pixel ratio of the second image; the image processing module 4552 is further configured to determine an exposure offset based on the third global average brightness and the overexposed pixel ratio if the second brightness change value is less than the second change threshold; adjust the first exposure parameter based on the exposure offset to obtain the third exposure parameter, and perform image acquisition based on the third exposure parameter and the target sampling density; the wake-up control module 4551 is further configured to trigger the image acquisition device to switch from a wake-up state to a sleep state in response to an image acquisition end command.

[0219] In some embodiments, the prediction module 4553 is further configured to determine the first acquisition time when acquiring the second image, and obtain the third ambient brightness sequence acquired before the first acquisition time; perform a weighted summation of the fourth ambient brightness corresponding to the first acquisition time and each ambient brightness in the third ambient brightness sequence to obtain the first predicted ambient brightness corresponding to the target acquisition time, wherein the target acquisition time is the next image acquisition time after the first acquisition time.

[0220] In some embodiments, the prediction module 4553 is further configured to determine the ratio of the first predicted ambient brightness to the desired ambient brightness, and use the product of the ratio and the first exposure parameter as the second exposure parameter.

[0221] In some embodiments, the wake-up control module 4551 is further configured to determine the number of valid wake-ups based on wake-up records in a preset historical period before the time to be evaluated; determine the effective wake-up rate based on the number of valid wake-ups and wake-up records; adjust the preset time period based on the difference between the preset target wake-up rate and the effective wake-up rate; and determine that the preset wake-up condition is met if the time interval between the time to be evaluated and the previous wake-up time reaches the adjusted preset time period.

[0222] This application provides a computer program product including a computer program or computer-executable instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer-executable instructions from the computer-readable storage medium and executes the computer-executable instructions, causing the electronic device to perform the exposure control method described above in this application.

[0223] This application provides a computer-readable storage medium storing computer-executable instructions or a computer program. When the computer-executable instructions or the computer program are executed by a processor, the processor will execute the exposure control method provided in this application. For example, ... Figure 3 The exposure control method shown.

[0224] In some embodiments, the computer-readable storage medium may be a memory such as RAM, ROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a variety of devices including one or any combination of the above-mentioned memories.

[0225] In some embodiments, computer-executable instructions may take the form of programs, software, software modules, scripts, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as stand-alone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

[0226] As an example, computer-executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple co-located files (e.g., files that store one or more modules, subroutines, or code sections).

[0227] As an example, computer-executable instructions can be deployed to execute on a single electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed across multiple locations and interconnected via a communication network.

[0228] In summary, through the embodiments of this application, the image acquisition device triggers wake-up when a preset wake-up condition is met, and acquires images stepwise at a first sampling density and a second sampling density, extracting the corresponding brightness change values. When the brightness change value reaches a second threshold, the brightness of the ambient light at the target acquisition time is predicted, and the second exposure parameter is directly calculated based on the predicted ambient brightness, the expected ambient brightness, and the initial exposure parameters. Finally, image acquisition is performed using the highest target sampling density and the second exposure parameters. This scheme balances environmental perception agility and hardware resource consumption through hierarchical state evaluation based on increasing sampling density and exposure parameter prediction. Specifically, this method first reduces background noise interference by using gradient sampling density, thus reducing computational resource consumption when the ambient light does not change significantly. Second, it provides a forward-looking data reference for exposure control by predicting the brightness at the target acquisition time for significant light changes. This approach solves the problem in related technologies where a single full-resolution image is used for passive multi-frame probing, resulting in an inability to adapt to dynamic lighting environment changes and difficulty in balancing the battery life of low-power devices with exposure response speed. This application's hierarchical sampling method reduces unnecessary computations caused by unnecessarily changing lighting conditions, allowing the system's computing power to be highly focused on core computational changes requiring intervention. Furthermore, predictive exposure estimation can find an exposure parameter update direction that balances current hardware timing and future lighting trends. This makes image acquisition equipment more efficient and intelligent, reducing power consumption while improving the convergence capability of automatic exposure.

[0229] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, and improvements made within the spirit and scope of this application are included within the scope of protection of this application.

Claims

1. An exposure control method, characterized in that, Applied to an image acquisition device, the method includes: While the image acquisition device is in a sleep state, if a preset wake-up condition is met, the image acquisition device is triggered to switch from the sleep state to the wake-up state, and in the wake-up state, a first image is acquired at a first sampling density. Extract the first brightness feature of the first image, and determine the first brightness change value based on the first brightness feature; If the first brightness change value is greater than or equal to the first change threshold and less than the second change threshold, then the second image is acquired at the second sampling density; wherein the second change threshold is greater than the first change threshold, and the second sampling density is higher than the first sampling density; Extract the second brightness feature from the second image, and determine the second brightness change value based on the second brightness feature; If the second brightness change value is greater than or equal to the second change threshold, then the brightness of the ambient light at the target acquisition time is predicted to obtain the first predicted ambient brightness. Based on the first predicted ambient brightness, the preset desired ambient brightness, and the first exposure parameters when acquiring the first image, a second exposure parameter is determined, and image acquisition is performed based on the second exposure parameter and the target sampling density; wherein, the target sampling density is higher than the second sampling density.

2. The method according to claim 1, characterized in that, The determination that the preset wake-up condition is met includes at least one of the following: For any evaluation time during the sleep state of the image acquisition device, if the time interval between the evaluation time and the previous wake-up time reaches a preset time period, it is determined that the preset wake-up condition is met. If the rate of change of ambient light brightness at the time to be evaluated is greater than the rate of change threshold, then the preset wake-up condition is determined to be met. If the time to be evaluated reaches the pre-predicted target wake-up time, then the preset wake-up condition is determined to be met.

3. The method according to claim 2, characterized in that, Before determining that the preset wake-up condition is met if the rate of change of ambient light brightness at the time to be evaluated is greater than the rate of change threshold, the method further includes: Acquire the first ambient brightness of the ambient light at the time to be evaluated, the second ambient brightness of the ambient light at the first historical acquisition time, and the first ambient brightness sequence acquired within a preset historical time period; The brightness change rate is determined based on the first ambient brightness, the second ambient brightness, and the time interval between the time to be evaluated and the first historical acquisition time. The median absolute deviation of the first ambient brightness sequence is determined, and the product of the median absolute deviation and a preset sensitivity coefficient is used as the rate of change threshold; wherein, the preset sensitivity coefficient is used to constrain the wake-up sensitivity of the image acquisition device.

4. The method according to claim 2, characterized in that, Before determining that the preset wake-up condition is met if the time to be evaluated reaches the pre-predicted target wake-up time, the method further includes: Acquire a second ambient brightness sequence acquired before the second historical acquisition time; the second historical acquisition time is the wake-up time preceding the time to be evaluated. A short-term prediction model and a long-term trend model are constructed based on the second ambient brightness sequence. The short-term prediction model represents the weighted sum of ambient brightness across different time spans from the second historical acquisition time. The long-term trend model is determined based on the brightness change value within a first preset time window and the slope of the brightness change within a second preset time window. The time span of the second preset time window is greater than that of the first preset time window. Based on the short-term prediction model and the long-term trend model, the second predicted ambient brightness at the time to be predicted, which is located after the second historical acquisition time, is predicted. If the difference between the second predicted ambient brightness and the third ambient brightness corresponding to the second historical acquisition time is greater than a preset change amount, then the target wake-up time is obtained by shifting forward by a preset time offset based on the time to be predicted.

5. The method according to claim 1, characterized in that, The step of extracting a first brightness feature from the first image and determining a first brightness change value based on the first brightness feature includes: Determine the first global average brightness of the first image, and use the first global average brightness as the first brightness feature; Obtain the second global average brightness of the third image; wherein the third image was acquired at the previous wake-up time when the image acquisition device switches from the sleep state to the wake-up state; Determine the absolute value of a first difference between the first global average brightness and the second global average brightness; The ratio of the absolute value of the first difference to the preset maximum brightness value of the image acquisition device is taken as the first brightness change value.

6. The method according to claim 1, characterized in that, After determining the first brightness change value based on the first brightness feature, the method further includes: If the first brightness change value is less than the first change threshold, maintain the first exposure parameter, and perform image acquisition based on the first exposure parameter and the target sampling density; In response to an image acquisition end command, the image acquisition device is triggered to switch from the wake-up state to the sleep state.

7. The method according to claim 1, characterized in that, The step of extracting the second brightness feature of the second image and determining the second brightness change value based on the second brightness feature includes: Multiple regions of interest (ROIs) are pre-divided for the second image, and the initial average brightness and region weight of each ROI are determined. For each region of interest, the average brightness of the region of interest is determined based on the initial average brightness of the region and the region weight; The average of the average brightness of the regions of interest is taken as the third global average brightness of the second image. Determine the total number of pixels in the second image and the number of overexposed pixels whose brightness values ​​are greater than a preset brightness limit, and use the ratio of the number of overexposed pixels to the total number of pixels as the overexposed pixel ratio of the second image; The third global average brightness and the overexposed pixel ratio are used as the second brightness feature; Obtain the second global average brightness of the third image; wherein the third image was acquired at the previous wake-up time when the image acquisition device switches from the sleep state to the wake-up state; Determine the absolute value of the second difference between the third global average brightness and the second global average brightness; The ratio of the absolute value of the second difference to the preset maximum brightness value of the image acquisition device is used as the initial brightness change value; The initial brightness change value is corrected based on the overexposed pixel ratio to obtain the second brightness change value.

8. The method according to claim 1, characterized in that, The second brightness feature includes the third global average brightness and the overexposed pixel ratio of the second image; after determining the second brightness change value based on the second brightness feature, the method further includes: If the second brightness change value is less than the second change threshold, then the exposure offset is determined based on the third global average brightness and the overexposed pixel ratio. The first exposure parameter is adjusted based on the exposure offset to obtain the third exposure parameter, and image acquisition is performed based on the third exposure parameter and the target sampling density; In response to an image acquisition end command, the image acquisition device is triggered to switch from the wake-up state to the sleep state.

9. The method according to claim 1, characterized in that, The step of predicting the ambient light brightness at the target acquisition time to obtain a first predicted ambient brightness includes: Determine the first acquisition time when acquiring the second image, and obtain the third ambient brightness sequence acquired before the first acquisition time; The first predicted ambient brightness corresponding to the target acquisition time is obtained by weighted summing of the fourth ambient brightness corresponding to the first acquisition time and each ambient brightness in the third ambient brightness sequence, wherein the target acquisition time is the next image acquisition time after the first acquisition time.

10. The method according to claim 1, characterized in that, The step of determining the second exposure parameter based on the first predicted ambient brightness, the preset desired ambient brightness, and the first exposure parameter when acquiring the first image includes: The ratio of the first predicted ambient brightness to the desired ambient brightness is determined, and the product of the ratio and the first exposure parameter is used as the second exposure parameter.

11. The method according to claim 2, characterized in that, Before determining that the preset wake-up condition is met, if the time interval between the time to be evaluated and the previous wake-up time reaches a preset time period, the method further includes: The number of valid wake-ups is determined based on wake-up records within a preset historical period prior to the time to be evaluated; Based on the number of effective wake-ups and the wake-up records, the effective wake-up rate is determined; The preset time period is adjusted based on the difference between the preset target wake-up rate and the effective wake-up rate. If the time interval between the time to be evaluated and the previous wake-up time reaches a preset time period, then the preset wake-up condition is determined to be met, including: If the time interval between the time to be evaluated and the previous wake-up time reaches the adjusted preset time period, then the preset wake-up condition is determined to be met.

12. An image acquisition device, characterized in that, The image acquisition device includes: The wake-up control module is used to trigger the image acquisition device to switch from the sleep state to the wake-up state if a preset wake-up condition is met while the image acquisition device is in a sleep state. An image processing module is configured to: acquire a first image at a first sampling density in the wake-up state; extract a first brightness feature from the first image and determine a first brightness change value based on the first brightness feature; if the first brightness change value is greater than or equal to a first change threshold and less than a second change threshold, acquire a second image at a second sampling density; wherein the second change threshold is greater than the first change threshold and the second sampling density is higher than the first sampling density; extract a second brightness feature from the second image and determine a second brightness change value based on the second brightness feature; The prediction module is used to predict the ambient light brightness at the target acquisition time if the second brightness change value is greater than or equal to the second change threshold, and obtain a first predicted ambient brightness; and determine a second exposure parameter based on the first predicted ambient brightness, the preset expected ambient brightness and the first exposure parameter when acquiring the first image. The image processing module is further configured to acquire an image based on the second exposure parameter and the target sampling density; wherein the target sampling density is higher than the second sampling density.

13. A computer-readable storage medium storing computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, the exposure control method according to any one of claims 1 to 11 is implemented.

Images