Scene recognition method, automatic exposure control method, chip and storage medium

Abstract

The invention discloses a scene recognition method, an automatic exposure control method, a chip and a storage medium. The scene recognition method comprises the following steps: acquiring exposure statistical information corresponding to a current scene according to corresponding exposure images of the current scene in a multi-exposure mode, and acquiring motion statistical information corresponding to the current scene according to corresponding exposure images of the current scene in a reference exposure mode and previous exposure images; the category of the current scene is identified according to the exposure statistical information and the motion statistical information corresponding to the current scene, and the category comprises at least one of a non-high-dynamic scene, a high-dynamic motion scene and a high-dynamic static scene, so that dynamic identification of the current shooting scene in a video shooting process is realized; therefore, dynamic control of multiple exposures in different shooting scenes in the video shooting process can be realized.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a scene recognition method, an automatic exposure control method, a chip, and a storage medium. Background Technology

[0002] In digital imaging devices, multiple exposures are often used to improve the dynamic range of images or videos. Existing automatic exposure control schemes are mostly designed for processing single-frame images. Furthermore, video shooting often involves changes in the shooting scene. Therefore, existing automatic exposure control schemes for single-frame images lack both control over multiple exposures during video shooting and recognition of the shooting scene, making it even more impossible to achieve dynamic control of multiple exposures in different scenes. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art in that it cannot dynamically identify the shooting scene during video shooting and thus cannot dynamically control multiple exposures under different scenes, and to provide a scene recognition method, an automatic exposure control method, a chip and a storage medium.

[0004] The present invention solves the above-mentioned technical problems through the following technical solutions:

[0005] A scene recognition method, the method comprising:

[0006] Based on the exposure image corresponding to the current scene in multiple exposure mode, obtain the exposure statistics information corresponding to the current scene, wherein different exposure modes correspond to different exposure amounts;

[0007] Based on the exposure image and the prior exposure image corresponding to the current scene in the reference exposure mode, motion statistics information corresponding to the current scene is obtained, wherein the reference exposure mode is one of multiple exposure modes, and the exposure image corresponding to the reference exposure mode is at least one exposure cycle apart from the prior exposure image;

[0008] Based on the exposure statistics and the motion statistics, the category of the current scene is identified, wherein the category includes at least one of a non-high dynamic scene, a high dynamic motion scene, and a high dynamic static scene.

[0009] Preferably, the multiple exposure modes include a long exposure mode and a short exposure mode. The step of obtaining exposure statistics information corresponding to the current scene based on the exposure image corresponding to the current scene in the multiple exposure modes includes:

[0010] Based on the long exposure image corresponding to the current scene in the long exposure mode, a long exposure statistical factor is obtained, which is used to characterize the bright and dark features of the long exposure image.

[0011] Based on the short exposure image corresponding to the current scene in the short exposure mode, a short exposure statistical factor is obtained, which is used to characterize the highlight features of the short exposure image;

[0012] The exposure statistics information is obtained based on the long exposure statistics factor and the short exposure statistics factor.

[0013] Preferably, the long exposure statistical factors include:

[0014] The first bright feature is used to characterize the average pixel value of all pixels in the long exposure image whose pixel values ​​fall within the first range;

[0015] The second bright part feature is used to characterize the proportion of the number of pixels in the long exposure image whose pixel values ​​fall within the first range to the total number of pixels in the long exposure image;

[0016] A first dark area feature is used to characterize the average pixel value of all pixels in the long exposure image whose pixel values ​​fall within a second range, wherein the maximum value of the second range is less than the minimum value of the first range.

[0017] The short exposure statistical factors include:

[0018] The third highlight feature is used to characterize the proportion of the number of pixels in the short exposure image whose pixel values ​​fall within the third range to the total number of pixels in the short exposure image.

[0019] Preferably, obtaining the exposure statistics information based on the long exposure statistics factor and the short exposure statistics factor includes:

[0020] The exposure statistics are obtained using the following formula:

[0021]

[0022] in, , , Characterize weights, and , Characterizing the bit width of the exposed image, Characterizing the first bright part features, Characterize the first dark area feature. Characterizing the second bright part feature, Characterizes the third bright part feature.

[0023] Preferably, identifying the category of the current scene based on the exposure statistics and the motion statistics includes:

[0024] If the exposure statistics are not greater than the high dynamic range threshold, then the current scene is identified as a non-high dynamic range scene.

[0025] If the exposure statistics are greater than the high dynamic range threshold and the motion statistics are greater than the motion threshold, then the current scene is identified as a high dynamic range motion scene.

[0026] If the exposure statistics are greater than the high dynamic range threshold and the motion statistics are not greater than the motion threshold, then the current scene is identified as the high dynamic range static scene.

[0027] And / or,

[0028] After obtaining the exposure statistics information corresponding to the current scene, the method further includes:

[0029] The exposure statistics are then subjected to temporal smoothing.

[0030] An automatic exposure control method is provided, the method being used to automatically update exposure parameters of at least one of multiple exposure modes, the multiple exposure modes including a long exposure mode and a short exposure mode, the method comprising:

[0031] Obtain the brightness information and category of the current scene, wherein the category includes at least one of non-high dynamic scene, high dynamic motion scene and high dynamic static scene, and the category is obtained according to any of the above scene recognition methods;

[0032] Determine the target exposure mode among the multiple exposure modes according to the category;

[0033] The target brightness of the target exposure mode is determined based on the brightness information;

[0034] The exposure parameters of the target exposure mode are updated based on the target brightness of the target exposure mode.

[0035] Preferably, updating the exposure parameters of the target exposure mode based on the target brightness of the target exposure mode includes:

[0036] Based on the target exposure image corresponding to the current scene in the target exposure mode, obtain the reference brightness of the target exposure image;

[0037] The target exposure amount of the target exposure mode is obtained based on the current exposure amount of the target exposure mode, the target brightness, and the reference brightness;

[0038] The exposure parameters of the target exposure mode are updated based on the target exposure amount of the target exposure mode.

[0039] Preferably, determining the target exposure mode among the multiple exposure modes according to the category includes at least one of the following:

[0040] When the category is the non-high dynamic range scene, the target exposure mode includes the long exposure mode;

[0041] When the category is the high dynamic range motion scene, the target exposure mode includes the long exposure mode and the short exposure mode;

[0042] When the category is the high dynamic range static scene, the target exposure mode includes the long exposure mode and the short exposure mode;

[0043] And / or,

[0044] The exposure parameters include exposure gain and exposure time, and updating the exposure parameters of the target exposure mode according to the target exposure amount of the target exposure mode includes at least one of the following:

[0045] When the category is the high dynamic range motion scene, the exposure gain of the long exposure mode and the short exposure mode is updated preferentially;

[0046] When the category is the high dynamic range static scene, the exposure time of the long exposure mode and the short exposure mode are updated preferentially;

[0047] And / or,

[0048] After updating the exposure parameters of the target exposure mode according to the target exposure amount of the target exposure mode, the method further includes:

[0049] The exposure parameters of the target exposure mode are updated according to the constraints.

[0050] A chip includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the scene recognition methods or any of the automatic exposure control methods described above.

[0051] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the above-described scene recognition methods or the steps of any of the above-described automatic exposure control methods.

[0052] The positive and progressive effects of this invention are as follows: This invention obtains the exposure statistics of the current scene based on the exposure image corresponding to the current scene in the multiple exposure mode, and obtains the motion statistics of the current scene based on the exposure image corresponding to the current scene in the reference exposure mode and the prior exposure image. Then, it identifies the category of the current scene based on the exposure statistics and motion statistics of the current scene. The category includes at least one of non-high dynamic range scene, high dynamic range motion scene and high dynamic range still scene. In this way, dynamic identification of the current shooting scene is realized during video shooting, thereby enabling dynamic control of multiple exposures under different shooting scenes during video shooting. Attached Figure Description

[0053] Figure 1 This is a flowchart of a scene recognition method according to Embodiment 1 of the present invention.

[0054] Figure 2 This is the exposure image and its Y-channel histogram corresponding to the triple exposure mode in the scene recognition method according to Embodiment 1 of the present invention.

[0055] Figure 3 This is the Y-channel histogram corresponding to the long exposure image in the scene recognition method according to Embodiment 1 of the present invention.

[0056] Figure 4 This is the Y-channel histogram corresponding to the short-exposure image in the scene recognition method according to Embodiment 1 of the present invention.

[0057] Figure 5 This is a flowchart of an automatic exposure control method according to Embodiment 2 of the present invention. Detailed Implementation

[0058] The present invention will be further illustrated by way of embodiments below, but the invention is not limited to the scope of the embodiments described herein.

[0059] Example 1

[0060] This embodiment provides a scene recognition method, specifically for dynamic recognition of the current shooting scene during video recording. (Refer to...) Figure 1 The scene recognition method in this embodiment includes:

[0061] S11. Based on the exposure image corresponding to the current scene in the multiple exposure mode, obtain the exposure statistics information corresponding to the current scene.

[0062] In this embodiment, the exposed image can come from the image data stream acquired by a digital imaging device using an image sensor, and different exposure modes correspond to different exposure amounts. Multiple exposure modes can include long exposure modes, short exposure modes, etc. Multiple exposure schemes include, but are not limited to, frame-based multiple exposure, line-based multiple exposure, and pixel-group-based multiple exposure. Currently, in the field of video imaging, line-based double exposure modes (long exposure / short exposure) or triple exposure modes (long exposure / medium exposure / short exposure) are commonly used, that is, the same row of pixels is exposed two or three times in different modes. Then, the exposed images corresponding to the multiple exposures are fused and processed with tone and brightness mapping to obtain a high dynamic range video stream that preserves the details of both bright and dark areas in the shooting scene.

[0063] In this embodiment, the multiple exposure mode includes at least a long exposure mode and a short exposure mode, and step S11 may specifically include the following steps:

[0064] Based on the long exposure image corresponding to the current scene in long exposure mode, obtain the long exposure statistical factor, which is used to characterize the bright and dark features of the long exposure image.

[0065] Based on the short exposure image corresponding to the current scene in short exposure mode, obtain the short exposure statistical factor, which is used to characterize the highlight features of the short exposure image;

[0066] Exposure statistics are obtained based on long exposure statistical factors and short exposure statistical factors.

[0067] Specifically, in this embodiment, exposure statistics factors (including the aforementioned long exposure statistics factors and short exposure statistics factors) are used to characterize the brightness features of the exposed image. Exposure statistics factors include, but are not limited to, those based on the Y-channel histogram. The Y-channel histogram was obtained. It can be obtained through the R / G / B channel histogram, for example, through any of the following methods:

[0068] 1)

[0069] 2)

[0070] 3)

[0071] in, Characterization channels Histogram, , Characterizing the exposure mode, for example, in triple exposure mode there is , Time represents long exposure mode, Exposure mode in time characterization Time represents short exposure mode, ,in, Characterize the left and right intervals of the calculated histogram.

[0072] Furthermore, It can be obtained using the following formula:

[0073]

[0074] Among them, the exposed images include 1 pixel Characterizing exposure mode Corresponding exposure image Channels at pixels pixel values, The Dirac function is characterized by taking the value 1 when the parameter is 0, and 0 otherwise.

[0075] In this embodiment, statistical information can be calculated from the exposure images corresponding to the triple exposure mode to obtain better scene metering results. Figure 2 The image shows the current scene in triple exposure mode and its corresponding Y-channel histogram.

[0076] Furthermore, for long exposure mode, which primarily reveals details in the darker areas of an image and controls the overall brightness level, long exposure statistical factors can include the first bright area feature. Second bright part features First dark area features Among them, the first bright part features The second highlight feature is used to characterize the average pixel value of all pixels in a long-exposure image whose pixel values ​​fall within the first range. The first dark area feature is used to characterize the proportion of pixels whose pixel values ​​fall within a first range in a long-exposure image out of the total number of pixels in the long-exposure image. The average pixel value used to characterize all pixels in a long-exposure image whose pixel values ​​fall within a second range, where the maximum value of the second range is less than the minimum value of the first range. (Refer to...) Figure 3 The first range can be the third threshold. and the fourth threshold The second range can be the range of pixel values ​​between the first threshold and the second range. Second threshold The range of pixel values ​​between these values ​​is as follows: , , ,in, , Characterizes the bit width of the exposed image.

[0077] For short exposure mode, the main focus is on presenting details in the bright areas of the image. Short exposure statistics can include third bright area features. This is used to characterize the proportion of pixels in a short-exposure image whose pixel values ​​fall within the third range out of the total number of pixels in the short-exposure image. (Refer to...) Figure 4 The third range can be greater than or equal to the fifth threshold. The range of pixel values, at this time, is: .

[0078] For the medium exposure mode, it is mainly used in the transition area between long and short exposures, and has little impact on the overall dynamic range expansion. It can be further determined based on the long exposure mode and the short exposure mode.

[0079] Based on this, exposure statistics can be obtained using the following formula:

[0080]

[0081] in, , , Characteristic weights, The three weights can be adjusted to improve adaptability based on scene changes. For example, when the scene is dark and noisy, the weights can be appropriately reduced. ,improve and .

[0082] Reference Figure 1 The scene recognition method in this embodiment further includes:

[0083] S12. Based on the exposure image and the prior exposure image corresponding to the current scene in the reference exposure mode, obtain the motion statistics information corresponding to the current scene.

[0084] In this embodiment, the reference exposure mode is one of the multiple exposure modes. For example, the reference exposure mode can be the medium exposure mode of a triple exposure mode or the long exposure mode of a double exposure mode. The exposure image corresponding to the reference exposure mode is at least one exposure cycle apart from the previously exposed image. For example, it can be the exposure images corresponding to two consecutive exposure modes in the image data stream. Motion statistics can be obtained from the frame difference between two frames, optical flow, or IMU measurement units, etc. Specifically, in this embodiment, motion statistics can be obtained according to the following formula:

[0085]

[0086] in, Characterizing exposure mode The corresponding exposure image, Characterizing exposure mode The corresponding previously exposed image.

[0087] Reference Figure 1 The scene recognition method in this embodiment further includes:

[0088] S13. Identify the category of the current scene based on exposure statistics and motion statistics.

[0089] Specifically, in this embodiment, if exposure statistics information Not greater than the high dynamic range threshold If the current scene is identified as a non-high dynamic scene, then the exposure statistics information will be considered. Greater than the high dynamic range threshold And sports statistics information Greater than the motion threshold If the current scene is identified as a high-dynamic motion scene, then the exposure statistics information will be used. Greater than the high dynamic range threshold And sports statistics information Not greater than the motion threshold If so, the current scene is identified as a high-dynamic static scene.

[0090] In this embodiment, the current scene can be judged in real time based on the exposure statistics and motion statistics obtained in real time, thereby enabling more efficient and accurate control of multiple exposures in different scenes. The current scene category can include at least one of non-high dynamic scene, high dynamic motion scene and high dynamic static scene.

[0091] Furthermore, since video shooting involves scene transitions, to mitigate the impact of statistical information oscillations during scene changes, exposure and motion statistics can be temporally smoothed separately. The current scene can then be identified and judged based on the temporally smoothed exposure and motion statistics, thereby enhancing the stability of the exposure and motion statistics and improving the robustness of scene judgment. The specific smoothing method can be selected according to the actual application; for example, temporal smoothing can be performed using the following formula:

[0092]

[0093] in, Characterizes statistical information after time-domain smoothing. Characteristic weights, Representing the statistical information obtained last time, This represents the statistical information currently obtained. Further, in this embodiment, the statistical information may be exposure statistics. It can also be sports statistics. And expose statistical information and sports statistics information The corresponding weights can be the same or different.

[0094] This embodiment obtains the exposure statistics of the current scene based on the exposure images corresponding to the current scene in the multiple exposure mode, and obtains the motion statistics of the current scene based on the exposure image of the current scene in the reference exposure mode and the previously exposed image. Then, the category of the current scene is identified based on the exposure statistics and motion statistics of the current scene. The category includes at least one of non-high dynamic scene, high dynamic motion scene and high dynamic still scene. In this way, the dynamic identification of the current shooting scene during video shooting is realized, and thus the dynamic control of multiple exposure under different shooting scenes during video shooting can be realized.

[0095] Example 2

[0096] This embodiment provides an automatic exposure control method for automatically updating the exposure parameters of at least one of multiple exposure modes, including a long exposure mode and a short exposure mode. (Refer to...) Figure 5 The method in this embodiment includes:

[0097] S21. Obtain the brightness information and category of the current scene.

[0098] In this embodiment, the brightness information of the current scene can be obtained based on existing technology. The scene category can include at least one of non-high dynamic scene, high dynamic motion scene and high dynamic static scene. The current scene category can be obtained according to the scene recognition method of Embodiment 1.

[0099] Reference Figure 5 The method in this embodiment further includes:

[0100] S22. Determine the target exposure mode in the multiple exposure mode according to the category.

[0101] In this embodiment, the target exposure mode refers to the exposure mode in the multiple exposure modes that requires updating the exposure parameters.

[0102] For example, in non-high dynamic range scenes, the target exposure mode can include a long exposure mode. Furthermore, short and medium exposure modes can maintain the original exposure parameters and do not participate in subsequent exposure fusion, essentially using only the long exposure mode to avoid motion ghosting issues caused by multi-frame fusion. In high dynamic range moving scenes, the target exposure mode includes both long and short exposure modes. In high dynamic range stationary scenes, the target exposure mode includes both long and short exposure modes.

[0103] Reference Figure 5 The method in this embodiment further includes:

[0104] S23. Determine the target brightness of the target exposure mode based on the brightness information.

[0105] In this embodiment, the target brightness can be customized according to the actual application. For example, the target brightness can be obtained based on the brightness index (BV) corresponding to the brightness information of the current scene and the exposure table interpolation. The exposure table presets different combinations of exposure time, exposure gain, and aperture used under different BV values ​​to make the average value (or metering value) of the exposed image reach the target value. Furthermore, there is a target brightness for long exposure images. For short-exposure images with target brightness .

[0106] Reference Figure 5 The method in this embodiment further includes:

[0107] S24. Update the exposure parameters of the target exposure mode according to the target brightness of the target exposure mode.

[0108] Specifically, in this embodiment, step S24 may include:

[0109] Based on the target exposure image corresponding to the current scene in the target exposure mode, obtain the reference brightness of the target exposure image;

[0110] Based on the current exposure of the target exposure mode, the target brightness, and the reference brightness, obtain the target exposure of the target exposure mode;

[0111] Update the exposure parameters of the target exposure mode based on the target exposure amount of the target exposure mode.

[0112] Specifically, in this embodiment, the reference brightness can be customized according to the actual application. For example, the average pixel value of all pixels in the target exposure image can be used as the reference brightness. Furthermore, a reference brightness can be provided for long exposure images. For short-exposure images, there is a reference brightness. Based on this, taking a long exposure mode as the target exposure mode as an example, we can first calculate the exposure adjustment amount. After that, the current exposure It can be updated to Taking a short exposure mode as an example, the exposure adjustment amount can be calculated first. After that, the current exposure It can be updated to .

[0113] Furthermore, in this embodiment, the exposure parameters include exposure gain and exposure time. When the scene is classified as a high dynamic range (HMR) motion scene, the target exposure mode includes a long exposure mode and a short exposure mode, and the exposure gain of the long and short exposure modes is updated preferentially to reduce motion blur caused by the increased exposure time. When the scene is classified as a high dynamic range stationary scene, the target exposure mode includes a long exposure mode and a short exposure mode, and the exposure time of the long and short exposure modes is updated preferentially to obtain a better signal-to-noise ratio.

[0114] If a triple exposure mode is used, the target exposure for each mode in various scenarios is determined by the ratio of the updated long exposure mode's exposure to the short exposure mode's exposure. For example, it could be: .

[0115] In this embodiment, to achieve finer exposure adjustment, an incremental adjustment method can be used, adjusting one exposure parameter at a time until an exposure parameter that meets the preset requirements is obtained. Here, an exposure adjustment step size can be introduced. ,in, To control the exposure adjustment speed variable, The larger the exposure value, the slower the exposure adjustment speed, until exposure convergence. In this embodiment, exposure convergence refers to... ,in, The table shows the collection threshold.

[0116] Furthermore, to meet the hardware requirements of the image sensor's internal exposure and readout timing, specific exposure constraints apply to the exposure parameters of the multiple exposure modes. That is, the exposure parameters of the target exposure mode obtained in step S24 also need to be updated according to these constraints. Specifically, the exposure parameters may include the image sensor's exposure time. Exposure gain of the imaging system etc., among which, exposure gain Including the gain of the image sensor and system digital gain If the exposure parameters set for multiple exposures do not meet the corresponding constraints, it may lead to abnormalities such as loss of image sensor output data or even failure to output data. Therefore, multiple exposure adjustments need to be performed while meeting the exposure constraints.

[0117] Typically, exposure constraints include: the longest and shortest exposure times, and the maximum and minimum image sensor gain. The ratio of maximum to minimum exposure, maximum frame rate, and minimum frame rate, etc., are among the factors considered. The calculation method for the exposure ratio in multiple exposures is as follows: The specific adjustment method is to prioritize ensuring that the exposure time meets the constraints, that is... Then, image sensor gain AG compensation is used. When the image sensor gain AG does not meet the exposure constraint, i.e. or Digital gain (DG) is used for compensation to maintain constant exposure. Finally, the exposure ratio is calculated. If satisfied Then, the exposure parameters of the target exposure mode in the multiple exposure mode can be set to the digital imaging device. If the requirements are not met, then... or Then it needs to be increased accordingly. or reduce To meet the constraint of the exposure ratio.

[0118] Furthermore, the constraints may also include the requirement that the set exposure time be able to eliminate light source frequency flicker; specifically, this can be achieved by obtaining light source statistics through an additional photoelectric sensor. When a photoelectric sensor detects a periodically flickering light source in the environment, it can output the light source frequency, for example, based on light source statistics. When a light source flickering at a frequency of 50Hz is detected in the current scene, the exposure time can be adjusted to an integer multiple of 10ms first, and then the exposure time adjusted by gain compensation can be used to maintain the exposure constant.

[0119] In this embodiment, dynamic control of multiple exposure modes under different shooting scenarios during video shooting can be realized. Specifically, during video shooting, automatic exposure control of the target exposure mode can be realized by combining the brightness information and category of the current scene, improving the effect of multiple exposure, and realizing efficient and stable intelligent video enhancement processing of online real-time video streams. By utilizing real-time control of multiple exposures, an automatic exposure control scheme with controllable dynamic range, adaptable to multiple scene switching, and meeting the exposure requirements of various image sensors can be realized. In addition, it can also be used for real-time processing of video streams from various commonly used mobile terminal cameras, security equipment, etc.

[0120] Example 3

[0121] This embodiment provides a chip that may include a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it can implement the scene recognition method provided in Embodiment 1 or the automatic exposure control method provided in Embodiment 2.

[0122] Example 4

[0123] This embodiment provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the program implements the steps of the scene recognition method provided in Embodiment 1 or the steps of the automatic exposure control method provided in Embodiment 2. The readable storage medium may include, but is not limited to, portable disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0124] In possible implementations, the present invention can also be implemented as a program product comprising program code. When the program product is run on a terminal device, the program code causes the terminal device to perform the steps of the scene recognition method provided in Embodiment 1 or the automatic exposure control method provided in Embodiment 2. The program code for executing the present invention can be written in any combination of one or more programming languages. The program code can be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on a remote device.

[0125] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these are merely illustrative examples, and the scope of protection of the present invention is defined by the appended claims. Those skilled in the art can make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, but all such changes and modifications fall within the scope of protection of the present invention.

Claims

1. A scene recognition method, characterized in that, The method includes: Based on the exposure image corresponding to the current scene in multiple exposure mode, obtain the exposure statistics information corresponding to the current scene, wherein different exposure modes correspond to different exposure amounts; Based on the exposure image and the prior exposure image corresponding to the current scene in the reference exposure mode, motion statistics information corresponding to the current scene is obtained, wherein the reference exposure mode is one of multiple exposure modes, and the exposure image corresponding to the reference exposure mode is at least one exposure cycle apart from the prior exposure image; Based on the exposure statistics and the motion statistics, the category of the current scene is identified, wherein the category includes at least one of a non-high dynamic scene, a high dynamic motion scene, and a high dynamic static scene.

2. The scene recognition method as described in claim 1, characterized in that, The multiple exposure modes include long exposure mode and short exposure mode. The step of obtaining exposure statistics information corresponding to the current scene based on the exposure image corresponding to the current scene in the multiple exposure modes includes: Based on the long exposure image corresponding to the current scene in the long exposure mode, a long exposure statistical factor is obtained, which is used to characterize the bright and dark features of the long exposure image. Based on the short exposure image corresponding to the current scene in the short exposure mode, a short exposure statistical factor is obtained, which is used to characterize the highlight features of the short exposure image; The exposure statistics information is obtained based on the long exposure statistics factor and the short exposure statistics factor.

3. The scene recognition method as described in claim 2, characterized in that, The long exposure statistical factors include: The first bright feature is used to characterize the average pixel value of all pixels in the long exposure image whose pixel values ​​fall within the first range; The second bright part feature is used to characterize the proportion of the number of pixels in the long exposure image whose pixel values ​​fall within the first range to the total number of pixels in the long exposure image; A first dark area feature is used to characterize the average pixel value of all pixels in the long exposure image whose pixel values ​​fall within a second range, wherein the maximum value of the second range is less than the minimum value of the first range. The short exposure statistical factors include: The third highlight feature is used to characterize the proportion of the number of pixels in the short exposure image whose pixel values ​​fall within the third range to the total number of pixels in the short exposure image.

4. The scene recognition method as described in claim 3, characterized in that, The step of obtaining the exposure statistics information based on the long exposure statistics factor and the short exposure statistics factor includes: The exposure statistics are obtained using the following formula: in, , , Characterize weights, and , Characterizing the bit width of the exposed image, Characterizing the first bright part features, Characterize the first dark area feature. Characterizing the second bright part feature, Characterizes the third bright part feature.

5. The scene recognition method as described in claim 1, characterized in that, The step of identifying the category of the current scene based on the exposure statistics and the motion statistics includes: If the exposure statistics are not greater than the high dynamic range threshold, then the current scene is identified as a non-high dynamic range scene. If the exposure statistics are greater than the high dynamic range threshold and the motion statistics are greater than the motion threshold, then the current scene is identified as a high dynamic range motion scene. If the exposure statistics are greater than the high dynamic range threshold and the motion statistics are not greater than the motion threshold, then the current scene is identified as the high dynamic range static scene. And / or, After obtaining the exposure statistics information corresponding to the current scene, the method further includes: The exposure statistics are then subjected to temporal smoothing.

6. An automatic exposure control method, characterized in that, The method is used to automatically update the exposure parameters of at least one of multiple exposure modes, wherein the multiple exposure modes include a long exposure mode and a short exposure mode, and the method includes: Obtain the brightness information and category of the current scene, wherein the category includes at least one of a non-high dynamic scene, a high dynamic motion scene, and a high dynamic static scene, and the category is obtained according to the scene recognition method as described in any one of claims 1-5; Determine the target exposure mode among the multiple exposure modes according to the category; The target brightness of the target exposure mode is determined based on the brightness information; The exposure parameters of the target exposure mode are updated based on the target brightness of the target exposure mode.

7. The automatic exposure control method as described in claim 6, characterized in that, Updating the exposure parameters of the target exposure mode based on the target brightness of the target exposure mode includes: Based on the target exposure image corresponding to the current scene in the target exposure mode, obtain the reference brightness of the target exposure image; The target exposure amount of the target exposure mode is obtained based on the current exposure amount of the target exposure mode, the target brightness, and the reference brightness; The exposure parameters of the target exposure mode are updated based on the target exposure amount of the target exposure mode.

8. The automatic exposure control method as described in claim 7, characterized in that, The determination of the target exposure mode among the multiple exposure modes according to the category includes at least one of the following: When the category is the non-high dynamic range scene, the target exposure mode includes the long exposure mode; When the category is the high dynamic range motion scene, the target exposure mode includes the long exposure mode and the short exposure mode; When the category is the high dynamic range static scene, the target exposure mode includes the long exposure mode and the short exposure mode; And / or, The exposure parameters include exposure gain and exposure time, and updating the exposure parameters of the target exposure mode according to the target exposure amount of the target exposure mode includes at least one of the following: When the category is the high dynamic range motion scene, the exposure gain of the long exposure mode and the short exposure mode is updated preferentially; When the category is the high dynamic range static scene, the exposure time of the long exposure mode and the short exposure mode are updated preferentially; And / or, After updating the exposure parameters of the target exposure mode according to the target exposure amount of the target exposure mode, the method further includes: The exposure parameters of the target exposure mode are updated according to the constraints.

9. A chip comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the scene recognition method as described in any one of claims 1-5 or the automatic exposure control method as described in any one of claims 6-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the scene recognition method as described in any one of claims 1-5 or the steps of the automatic exposure control method as described in any one of claims 6-8.

Images