Target recognition method, electronic device and computer readable storage medium

Abstract

This application discloses a target recognition method, an electronic device, and a computer-readable storage medium. The method includes: in response to receiving a target recognition instruction, acquiring currently acquired images corresponding to at least two preset supplementary lighting parameters; determining target supplementary lighting parameters from the at least two preset supplementary lighting parameters based on the image quality of each currently acquired image; and performing target recognition processing based on the target supplementary lighting parameters and the currently acquired target image corresponding to the target supplementary lighting parameters to obtain a target recognition result. The above scheme can improve the target recognition effect.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a target recognition method, an electronic device, and a computer-readable storage medium. Background Technology

[0002] In target recognition, active illumination schemes can be used to achieve stable recognition in all weather and multi-light environments. Common schemes typically use a single light source for active illumination to achieve image acquisition and recognition.

[0003] However, in practical applications, the external environment is complex. For example, the light source required in strong light environments is different from that in low light environments. If a light source with a single supplementary lighting parameter is used for target recognition, it may lead to recognition failure under complex lighting conditions. Summary of the Invention

[0004] The main technical problem addressed by this application is to provide a target recognition method, electronic device, and computer-readable storage medium that can improve the accuracy of target recognition.

[0005] To solve the above-mentioned technical problems, one technical solution adopted in this application is: to provide a target recognition method, the target recognition method comprising: in response to receiving a target recognition instruction, acquiring at least two currently acquired images corresponding to preset supplementary lighting parameters, determining target supplementary lighting parameters from the at least two preset supplementary lighting parameters based on the image quality of each currently acquired image; performing target recognition processing based on the target supplementary lighting parameters and the currently acquired images corresponding to the target supplementary lighting parameters, and obtaining a target recognition result.

[0006] To solve the above-mentioned technical problems, another technical solution adopted in this application is: to provide an electronic device, including a memory and a processor, wherein the memory stores program instructions, and the processor retrieves the program instructions from the memory to execute the above-mentioned mapping anomaly identification method.

[0007] To solve the above-mentioned technical problems, another technical solution adopted in this application is to provide a computer-readable storage medium including program data, which is used to implement the above-mentioned mapping anomaly identification method when executed by a processor.

[0008] The target recognition method of this application, in response to receiving a target recognition instruction, acquires currently captured images corresponding to at least two preset supplementary lighting parameters; determines target supplementary lighting parameters from the at least two preset supplementary lighting parameters based on the image quality of each currently captured image; and performs target recognition processing based on the target supplementary lighting parameters and the currently captured target image corresponding to the target supplementary lighting parameters to obtain the target recognition result. This scheme selects target supplementary lighting parameters by comparing the image quality of currently captured images under different preset supplementary lighting parameters, dynamically selecting the optimal supplementary lighting scheme for the current scene to improve the target recognition effect. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly described below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 This is a flowchart illustrating an exemplary embodiment of the target recognition method shown in this application; Figure 2 This is a schematic diagram of an exemplary embodiment of the target recognition device shown in this application; Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device provided in this application; Figure 4 This is a schematic diagram of an embodiment of the computer-readable storage medium provided in this application. Detailed Implementation

[0010] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It is understood that the specific embodiments described herein are only for explaining this application and not for limiting it. Furthermore, it should be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all structures. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0011] First, it's important to clarify that target recognition refers to the process of distinguishing a specific target from other targets or categories using appropriate recognition technologies. Common target recognition methods utilize visible light sensors to acquire relevant images or videos, which are then processed to obtain target recognition information. When using visible light sensors for target recognition, the intensity of ambient light can affect the recognition results; both weak and strong light can impact the accuracy. To reduce interference from the external environment, active lighting solutions can be employed. However, current active lighting solutions are mostly based on a single light source, making them difficult to adapt to the complex and ever-changing environments in real-world applications.

[0012] Based on this, embodiments of this application propose a target recognition method, electronic device, and computer-readable storage medium. The method selects target illumination parameters by comparing currently acquired images with multiple preset illumination parameters, thereby improving target recognition performance. For details, please refer to [link to relevant documentation]. Figure 1 , Figure 1 This is a flowchart illustrating an exemplary embodiment of the target recognition method shown in this application.

[0013] The execution entity of the target recognition method can be a terminal device, a server, or other processing device. The terminal device can be a user equipment (UE), computer, mobile device, user terminal, terminal, cellular phone, cordless phone, personal digital assistant (PDA), handheld device, computing device, in-vehicle device, wearable device, etc. The execution entity of the target recognition method can also be a target recognition device. In some possible implementations, the target recognition method can be implemented by a processor calling computer-readable instructions stored in memory.

[0014] Specifically, the target recognition method in this embodiment includes the following steps: S110: In response to receiving a target recognition instruction, acquire the currently acquired image corresponding to at least two preset supplementary lighting parameters.

[0015] A target recognition command is used to instruct the target recognition device to begin target recognition. For example, when a sensor detects a target entering a preset detection range, it generates a target recognition command and sends it to the target recognition device. The sensor can be a millimeter-wave radar, a PIR (Passive Infrared) pyroelectric sensor, or a capacitive proximity sensor, etc. The preset detection range can be set empirically; for example, a region separated from the target recognition device by a preset distance can be used as the preset detection range, such as 0.5 meters to 2 meters. In some embodiments, the target recognition command can be a high-level interrupt signal, with the sensor outputting a high-level interrupt signal to the external interrupt pin of the target recognition device. In other embodiments, the target recognition command can also be a low-level interrupt signal, with the sensor outputting a low-level interrupt signal to the external interrupt pin of the target recognition device.

[0016] The preset supplementary lighting parameters are pre-set supplementary lighting parameters. In some embodiments, the preset supplementary lighting parameters include supplementary lighting wavelengths, acquiring currently captured images at at least two supplementary lighting wavelengths. Since sunlight has energy peaks in the visible to near-infrared band (especially concentrated in the 700–900 nm range), traditional single-wavelength supplementary lighting sources (such as 850 nm) are easily interfered with by background light of the same frequency, resulting in low image contrast and high noise. Therefore, by actively avoiding high-energy bands of sunlight (e.g., prioritizing the 1050 nm band), ambient light interference can be effectively suppressed, improving the imaging signal-to-noise ratio and image availability in backlit scenes, thus solving the imaging failure problem caused by strong light. When the preset supplementary lighting parameters include supplementary lighting wavelengths, the light-emitting module can be a multi-channel LED driver chip, such as the TITPS61088, or a discrete MOSFET array. Multiple supplementary light wavelengths are pre-configured, and two or more can be selected from 850nm-1050nm as the corresponding supplementary light wavelengths. For example, 850nm, 940nm and 1050nm can be selected as the corresponding supplementary light wavelengths, or 850nm and 940nm can be selected as the corresponding supplementary light wavelengths; then the currently acquired image is acquired under the corresponding supplementary light wavelength.

[0017] In some embodiments, the preset supplementary lighting parameters include supplementary lighting intensity, and the currently acquired images are obtained under at least two supplementary lighting intensities. In other embodiments, the preset supplementary lighting parameters include supplementary lighting wavelength and supplementary lighting intensity, and the currently acquired images are obtained under at least two supplementary lighting wavelengths and supplementary lighting intensities. The target recognition device controls the light-emitting module to switch between different preset supplementary lighting parameters to acquire multiple currently acquired images.

[0018] The currently acquired image is an image of the current scene captured under different preset supplementary lighting parameters. For example, when the target recognition device controls the light-emitting module to switch to the preset supplementary lighting parameters, the image acquisition device correspondingly acquires an image under those preset supplementary lighting parameters. One or more images can be acquired under one preset supplementary lighting parameter. When multiple images are acquired, the image with the best quality can be selected as the currently acquired image under that preset supplementary lighting parameter. As an example, the target recognition device sends a supplementary lighting parameter switching command to the light-emitting module. The light-emitting module switches the preset supplementary lighting parameters according to a preset interval. Each time the light-emitting module lights up according to the preset supplementary lighting parameters, the target recognition device sends a frame synchronization trigger signal to the image acquisition device. Upon receiving the frame synchronization trigger signal, the image acquisition device immediately starts exposure to capture the currently acquired image under the preset supplementary lighting parameters and can store the currently acquired image in an on-chip buffer. The preset interval time is generally less than or equal to 5 milliseconds to ensure response speed. The target recognition device can send a frame synchronization trigger signal to the image acquisition device via hardware pulses (e.g., GPIO toggles) and / or software protocols (e.g., I²C commands). When using hardware pulse triggering, the pulse width output by the target recognition device is generally greater than or equal to 1 microsecond. Using software protocol delay can disable interrupt preemption and ensure a delay accuracy of less than ±1 millisecond. In some embodiments, multiple preset supplementary lighting parameters of the light-emitting module are used at different times to prevent spectral aliasing. In other embodiments, the light-emitting module can use multiple preset supplementary lighting parameters simultaneously to reduce switching latency. For example, the light-emitting module can simultaneously illuminate light sources of two supplementary lighting wavelengths to acquire images.

[0019] S120: Determine the target illumination parameter from at least two preset illumination parameters based on the image quality of each currently acquired image.

[0020] Image quality is used to measure the recognition difficulty of each currently acquired image. Higher image quality means lower recognition difficulty, and lower image quality means higher recognition difficulty. In some embodiments, a subjective evaluation method can be used to determine the image quality of each currently acquired image. In other embodiments, an objective evaluation method can be used to determine the image quality of each currently acquired image. For example, peak signal-to-noise ratio (PSNR), mean square error (MSE), etc., can be used to obtain the image quality of each currently acquired image. In still other embodiments, multiple evaluation metrics can be used to evaluate the currently acquired images, and the evaluation scores obtained from the multiple evaluation metrics can be statistically processed to obtain the image quality of each currently acquired image.

[0021] Target illumination parameters are illumination parameters used for target recognition. In some embodiments, target illumination parameters can be selected from preset illumination parameters based on the image quality of each currently acquired image. For example, the preset illumination parameter corresponding to the currently acquired image with the best image quality can be determined as the target illumination parameter, or the preset illumination parameter corresponding to the currently acquired image with an image quality greater than a preset quality threshold can be determined as the target illumination parameter. There can be one or more target illumination parameters. When multiple target illumination parameters are selected, target recognition processing can be performed using the currently acquired target images under multiple target illumination parameters to obtain the target recognition result.

[0022] S130: Perform target recognition processing based on the target illumination parameters and the currently acquired image corresponding to the target illumination parameters to obtain the target recognition result.

[0023] For example, after receiving a target recognition instruction, the target recognition device acquires multiple currently acquired images corresponding to preset supplementary lighting parameters; and selects the currently acquired image corresponding to the target supplementary lighting parameter from among the currently acquired images based on the image quality of each currently acquired image. In some embodiments, if the currently acquired image selected from among the currently acquired images meets the standard, the currently acquired image corresponding to the target supplementary lighting parameter is directly determined as the current target acquired image; target recognition is performed on the current target acquired image to obtain a target recognition result; if the currently acquired image selected from among the currently acquired images does not meet the standard, the current target acquired image is re-acquired using the target supplementary lighting parameter, and target recognition is performed on the current target acquired image to obtain a target recognition result.

[0024] It should be noted that target recognition results include both failure and success. Failure indicates that the target's identity information could not be determined, while success indicates that the target's identity information was successfully determined. When recognition is successful, the target recognition result also includes the target's location information, recognition confidence level, and identity information.

[0025] As can be seen, the target recognition method in this embodiment responds to receiving a target recognition instruction by acquiring at least two currently acquired images corresponding to preset supplementary lighting parameters; determining target supplementary lighting parameters from the at least two preset supplementary lighting parameters based on the image quality of each currently acquired image; and performing target recognition processing based on the target supplementary lighting parameters and the currently acquired target image corresponding to the target supplementary lighting parameters to obtain the target recognition result. The above scheme selects target supplementary lighting parameters by comparing the image quality corresponding to currently acquired images under different preset supplementary lighting parameters, dynamically selecting the optimal supplementary lighting scheme for the current scene to improve the target recognition effect.

[0026] In some embodiments, step S120 may include the following process: performing quality assessment processing on each currently acquired image to obtain the image quality of each currently acquired image; in response to the image quality of the currently acquired image being greater than a preset quality threshold, determining the preset supplementary lighting parameter corresponding to the currently acquired image as the target supplementary lighting parameter. Thus, by selecting the currently acquired image with better image quality, a target supplementary lighting parameter more suitable for the current scene is selected.

[0027] A preset quality threshold is used to classify the currently acquired images. When the image quality of the currently acquired image is greater than or equal to the preset quality threshold, the preset supplementary lighting parameter corresponding to the currently acquired image can be determined as the target supplementary lighting parameter; when the image quality of the currently acquired image is less than the preset quality threshold, the preset supplementary lighting parameter corresponding to the currently acquired image can be discarded. For example, the preset quality threshold can be set based on experience, or it can be obtained by statistically analyzing the image quality of each currently acquired image. For instance, the average image quality of each currently acquired image can be determined as the preset quality threshold, or the maximum image quality value among each currently acquired image can be determined as the preset quality threshold.

[0028] In other embodiments, the target recognition device performs quality assessment processing on each currently acquired image to obtain the image quality of each currently acquired image; it then selects the top N currently acquired images from those images according to their image quality from highest to lowest, and determines the preset supplementary lighting parameters corresponding to the top N currently acquired images as the target supplementary lighting parameters. Here, N can be 1, 2, or 3, etc.

[0029] The steps for determining the image quality of each currently acquired image may include: obtaining evaluation scores for at least two evaluation metrics for each currently acquired image; and weighting the evaluation scores of each evaluation metric according to the target weights corresponding to each evaluation metric to obtain the image quality of the currently acquired image. Therefore, by evaluating the currently acquired image from multiple dimensions using multiple evaluation metrics, the image quality of the currently acquired image can be assessed more accurately.

[0030] Evaluation metrics measure the quality of a currently acquired image in a corresponding dimension. For example, evaluation metrics can be determined based on the actual application scenario; alternatively, they can be selected from a preset evaluation metric library. For instance, the preset evaluation metric library may include, but is not limited to, contrast, sharpness, signal-to-noise ratio, brightness, and completeness. Two or more metrics can be selected from the preset library as evaluation metrics for the currently acquired image. In other embodiments, one metric can be selected from the preset evaluation metric library. When only one metric is selected, the evaluation score corresponding to that metric can be directly determined as the image quality of the currently acquired image.

[0031] The evaluation score refers to the numerical value obtained by evaluating the currently acquired image based on the corresponding evaluation index. Different evaluation indexes require different calculation methods to obtain the evaluation score. Since image quality is used to measure the recognition difficulty of the currently acquired image, initial detection can be performed on the currently acquired image to obtain the target region in the currently acquired image, which includes the target object; quality evaluation processing is then performed on the target region to obtain the image quality of the currently acquired image.

[0032] As an example, when the evaluation metric is contrast ratio, contrast ratio measures the intensity of the difference between light and dark areas in a target region, reflecting the visibility of the target region. Generally, a higher value indicates a more obvious contrast. The contrast ratio calculation process includes: obtaining the maximum and minimum pixel values ​​from the pixel values ​​of each pixel in the target region; obtaining the difference between the maximum and minimum pixel values, and the sum of the maximum and minimum pixel values; and determining the ratio between the difference and the sum as the contrast ratio. For example, the contrast ratio calculation satisfies the following formula:

[0033] in, Indicates contrast. Represents the set of pixel values ​​within the target region. Indicates the maximum pixel value. This represents the minimum pixel value. The value calculated using the above formula can also be normalized to obtain the contrast ratio.

[0034] As another example, when the evaluation metric is sharpness, sharpness measures the clarity of the outline and details of the target area, effectively reflecting focus quality. The sharpness calculation process can be as follows: calculate the gradient of the pixel values ​​of each pixel in the target area to obtain the image gradient of the target area; calculate the L2 norm of the image gradient of the target area; and determine the sharpness as the ratio between the L2 norm of the image gradient of the target area and the total number of pixels in the target area. For example, the sharpness calculation satisfies the following formula:

[0035] in, The image gradient can be represented by the Sobel operator. The L2 norm of the image gradient is represented. This represents the total number of pixels in the target region.

[0036] As another example, when the evaluation metric is signal-to-noise ratio (SNR), SNR is used to assess the purity of the image signal and suppress noise interference introduced by low lighting or high gain. The calculation process for SNR can be as follows: obtain the pixel average and pixel standard deviation among the pixel values ​​of each pixel in the target region; determine the ratio between the pixel average and the pixel standard deviation as the SNR. For example, the calculation of SNR satisfies the following formula:

[0037] in, Represents the average pixel value. This represents the standard deviation of pixels.

[0038] As another example, when the evaluation metric is brightness, brightness is used to measure the lightness or darkness of the currently acquired image. The brightness calculation process can be as follows: obtain the ratio between the pixel value of each pixel in the target region and a preset value; sum the ratios corresponding to each pixel in the target region to obtain a sum of ratios; and determine the brightness by dividing the sum of ratios by the total number of pixels in the target region. For example, the brightness calculation satisfies the following formula:

[0039] in, This represents the total number of pixels in the target region. This represents the pixel value of the i-th pixel. This represents the preset value. The pixel value of a pixel is between 0 and 255, so the preset value y can be set to 255.

[0040] As another example, when the evaluation metric is completeness, completeness measures the integrity of the target object. It can be determined by analyzing the position and size changes of the target region between consecutive image frames to assess target stability. If the displacement or scaling is small, the target is considered complete and stable. The calculation process for completeness can be as follows: obtain the center position offset rate of the target region in the current image relative to the target region in historical images, and obtain the area change rate of the target region in the current image relative to the target region in historical images; calculate the sum of the changes between the center position offset rate of a first preset multiple and the area change rate of a second preset multiple; determine the difference between the preset value and the sum of changes as a candidate completeness; select the completeness of the current image from the preset completeness and candidate completeness based on the relationship between the preset completeness and the candidate completeness. For example, the calculation of completeness satisfies the following formula:

[0041] in, This indicates the preset completeness level, which can be selected arbitrarily from 0.5 to 0.7; This represents the default value, which can be set to 1; Indicates the first preset multiple. This represents the center position offset rate, which can be calculated from the target region relative to the width and height of the image. Indicates the second preset multiple. This represents the rate of change of area. If the displacement is less than 5% and the area is stable, then I is 1; otherwise, it decays linearly, with the minimum being the preset integrity level.

[0042] It should be noted that the evaluation scores of the above evaluation indicators also need to be uniformly mapped to the [0,1] interval to eliminate the difference in dimensions, ensure that the quality scores of different dimensions are comparable and feasible to be integrated, and provide standardized input for subsequent processing.

[0043] After obtaining the evaluation scores of each evaluation indicator, the evaluation scores of each evaluation indicator are weighted and summed to obtain the image quality of the currently acquired image. In some embodiments, the target weights of each evaluation indicator can be fixed, for example, the target weights of each evaluation indicator can be preset based on experience. In other embodiments, the preset weights of each evaluation indicator can be adjusted based on historical recognition information to obtain the target weights of each evaluation indicator. The historical recognition information includes the historical recognition results of at least one historical recognition process and the evaluation scores of each evaluation indicator corresponding to each historical target acquired image. Specifically, the historical recognition results corresponding to at least one historical target acquired image and the evaluation scores of each evaluation indicator corresponding to each historical target acquired image are obtained, and the acquisition time of each historical target acquired image is different; the recognition impact evaluation processing of each evaluation indicator is performed based on the historical recognition results of each historical target acquired image and the evaluation scores of each evaluation indicator corresponding to each historical target acquired image to obtain the target weights corresponding to each evaluation indicator. Thus, the target weights of each evaluation indicator are determined by the historical recognition information of multiple historical target recognitions, realizing adaptive optimization of the target weights and improving the accuracy of image quality evaluation.

[0044] Historical target acquisition images refer to images identified during the historical target recognition process. For example, the acquisition process of historical target acquisition images can be as follows: at a corresponding historical time point, if a target recognition instruction is received, acquire at least two historical acquisition images corresponding to preset supplementary lighting parameters; determine target supplementary lighting parameters from the at least two preset supplementary lighting parameters based on the image quality of each historical acquisition image; if the historical acquisition image corresponding to the target supplementary lighting parameter meets the standard, then determine the historical acquisition image corresponding to the target supplementary lighting parameter as the historical target acquisition image; if the historical acquisition image corresponding to the target supplementary lighting parameter does not meet the standard, then re-acquire the historical target acquisition image based on the target supplementary lighting parameter. Multiple historical target acquisition images are acquired at different times, that is, historical target acquisition images acquired in different recognition processes. For example, the images from the Nth recognition process most recent to the current time can be determined as the historical target acquisition images in this embodiment.

[0045] Based on the recognition information generated during the recognition process of images of each historical target, the recognition impact assessment of each evaluation indicator is carried out to determine the importance of each evaluation indicator, thereby obtaining the target weight of each evaluation indicator.

[0046] In some embodiments, a target evaluation index that leads to historical recognition results can be determined from various evaluation indicators, and the preset weights of the target evaluation index can be adjusted to obtain the target weight of the target evaluation index. Specifically, in response to the historical recognition results corresponding to each historical target acquisition image indicating recognition failure, the target evaluation index that leads to recognition failure is obtained from each evaluation indicator based on the evaluation scores of each evaluation indicator in each historical target acquisition image; the preset weight of the target evaluation index is increased to obtain the target weight of the target evaluation index, and the preset weights of the other evaluation indicators besides the target evaluation index are used as the corresponding target weights. The target evaluation index that leads to recognition failure has a significant impact on image quality, so increasing its weight determines an image quality that is more in line with practical applications.

[0047] If all historical recognition results for each historical target image indicate recognition failure, it is necessary to determine which evaluation metric caused the failure. If at least one historical recognition result for each historical target image indicates successful recognition, the weight adjustment ends, and the preset weights of each evaluation metric are used as the corresponding target weights. The number of historical target images can be determined empirically, for example, 3, 4, or 5. Multiple historical target images most recently acquired from the current image can be used as the historical target images for weight optimization in this embodiment.

[0048] Furthermore, when all historical recognition results corresponding to the historical target images indicate recognition failure, the process of obtaining the target evaluation index that caused the recognition failure from each evaluation index based on the evaluation scores of each evaluation index in each historical target image can include the following steps: For each historical target image, select candidate evaluation indices whose evaluation scores are below the standard from each evaluation index; in response to the same candidate evaluation index for each historical target image and the number of candidate evaluation indices being less than or equal to a preset threshold, obtain other historical images corresponding to each historical target image. These other historical images are obtained from the same scene as the historical target images, but their supplementary lighting parameters are different; in response to the evaluation scores of the candidate evaluation indices corresponding to the other historical images being up to standard and the historical recognition results of the other historical images indicating successful recognition, the candidate evaluation index is determined as the target evaluation index. This improves the accuracy of the target evaluation index through multi-dimensional cross-validation.

[0049] Candidate evaluation indicators are those evaluation indicators that fail to meet the standards in each historical target acquisition image. For example, each evaluation indicator has corresponding compliance requirements. When the evaluation score corresponding to the evaluation indicator meets the standard, it means that the historical target acquisition image meets the requirements under the corresponding evaluation indicator; when the evaluation score corresponding to the evaluation indicator fails to meet the standard, it means that the historical target acquisition image does not meet the requirements under the corresponding evaluation indicator.

[0050] As an example, taking contrast as an evaluation metric, a higher contrast indicates a better historical target acquisition image. In some embodiments, a baseline threshold for contrast can be set. When the contrast of a historical target acquisition image is less than or equal to the baseline threshold, it indicates that the contrast of the historical target acquisition image is substandard, and contrast is used as a candidate evaluation metric. In other embodiments, a baseline threshold and an average threshold for contrast can be set. When the contrast of a historical target acquisition image is less than or equal to the baseline threshold and the average contrast of all historical target acquisition images is less than the average threshold, contrast is used as a candidate evaluation metric. In other embodiments, the average contrast and contrast standard deviation of the corresponding historical target acquisition images can be obtained from several successfully identified historical recognition results. If the contrast of a historical target acquisition image is less than a preset multiple of the average contrast and less than the contrast standard deviation, contrast is used as a candidate evaluation metric.

[0051] When the candidate evaluation metrics for each historical target image are different, the target evaluation metric causing the recognition failure cannot be accurately obtained. The weight adjustment ends, and the preset weights of each evaluation metric are used as the corresponding target weights. Different candidate evaluation metrics include situations where any historical target image does not contain any candidate evaluation metrics, or where any two historical target images have different candidate evaluation metrics. When all historical target images have the same candidate evaluation metrics, it indicates that all historical target images have the same problem. In this case, it is necessary to determine whether the number of candidate evaluation metrics for each historical target image is less than a preset threshold. If so, the determination continues; otherwise, the weight adjustment ends, and the preset weights of each evaluation metric are used as the corresponding target weights. The preset threshold can be 1, meaning each historical target image has only one candidate evaluation metric, and all historical target images have the same candidate evaluation metric. Alternatively, the preset threshold can be 2, meaning each historical target image has one or two identical candidate evaluation metrics.

[0052] When the preset threshold is 1, it means that each historical target image has only one candidate evaluation index, and all other evaluation indices meet the requirements. If the evaluation indices include contrast, sharpness, signal-to-noise ratio (SNR), brightness, and integrity, the requirements for sharpness are: sharpness greater than a preset sharpness threshold (typically 0.18); brightness greater than or equal to a first preset brightness and less than or equal to a second preset brightness (the first preset brightness can be 0.25, and the second preset brightness can be 0.75); SNR greater than a preset SNR threshold (typically 0.2); integrity greater than or equal to a preset integrity threshold (typically 0.65); and whether the angle or area of ​​the target region is abnormal.

[0053] Once it's determined that the historical target images share the same candidate evaluation metrics, it's necessary to assess whether other historical images corresponding to each target can be successfully identified. If the candidate evaluation metrics for other historical images meet the criteria and are successfully identified, environmental issues within the corresponding historical scene can be ruled out, and the reason for recognition failure can be further identified as the candidate evaluation metric. If the candidate evaluation metrics for other historical images meet the criteria but recognition fails, or vice versa, the candidate evaluation metric is excluded, as it may be due to environmental issues within the historical scene.

[0054] In other embodiments, the target evaluation index causing recognition failure can be determined based on the correlation between candidate evaluation indices and recognition confidence. Specifically, for each historical target image, candidate evaluation indices with substandard evaluation scores are selected from all evaluation indices. If the candidate evaluation indices for all historical target images are the same and the number of candidate evaluation indices is less than a preset threshold, the correlation between the evaluation scores of the candidate evaluation indices and the recognition confidence of each historical target image is obtained. If the correlation between the evaluation scores of the candidate evaluation indices and the recognition confidence of each historical target image is positively correlated, the candidate evaluation index is determined as the target evaluation index. Thus, by obtaining the correlation between candidate evaluation indices and recognition confidence, it is determined whether the candidate evaluation indices directly affect the recognition result, thereby allowing for more adaptive adjustment of the weights of the target evaluation indices.

[0055] The correlation refers to how the recognition confidence changes as the evaluation scores of candidate evaluation indicators change. For example, correlation includes positive correlation, negative correlation, and no correlation. Positive correlation means that the higher the evaluation score of the candidate evaluation indicator, the higher the recognition confidence; negative correlation means that the lower the evaluation score of the candidate evaluation indicator, the higher the recognition confidence; no correlation means that there is no relationship between the evaluation score of the candidate evaluation indicator and the recognition confidence. When the correlation between the evaluation scores of candidate evaluation indicators in historical target acquisition images and the recognition confidence corresponding to each historical target acquisition image is positively correlated, then the candidate evaluation indicator is determined to directly affect the recognition result, and the candidate evaluation indicator is identified as the target evaluation indicator.

[0056] In other embodiments, the correlation coefficient between the candidate evaluation index and the recognition confidence can be determined based on the evaluation score of the candidate evaluation index in each historical target acquisition image and the recognition confidence corresponding to each historical target acquisition image; when the correlation coefficient is greater than a preset coefficient threshold, the candidate evaluation index is determined as the target evaluation index.

[0057] In other embodiments, the recognition success rate can be obtained from historical recognition information when other evaluation indicators are the same but the candidate evaluation indicator meets the standard. If the recognition success rate is higher than a preset success rate threshold, the candidate evaluation indicator is determined as the target evaluation indicator. In other embodiments, it can be inferred from historical recognition information whether the historical target image can be successfully recognized if the candidate evaluation indicator meets the standard. If so, the candidate evaluation indicator is determined as the target evaluation indicator.

[0058] It should be noted that the target recognition device can arbitrarily combine the above-mentioned target evaluation index determination methods, and this application does not limit this. For ease of understanding, taking the combination of the above multiple methods as an example, we determine whether contrast is a candidate target evaluation index. First, the contrast of each historical target acquisition image is judged to meet the standard, including absolute threshold verification: the contrast of each historical target acquisition image is less than the benchmark threshold and the average contrast of each historical target acquisition image is less than the average threshold; relative historical comparison: the contrast of each historical target acquisition image is less than the average contrast of the corresponding historical target acquisition images in the successfully recognized historical recognition results, and the contrast of each historical target acquisition image deviates from the standard deviation of the contrast of the corresponding historical target acquisition images in the successfully recognized historical recognition results.

[0059] Then, other evaluation indicators besides contrast are checked for compliance, such as sharpness greater than the preset sharpness threshold; brightness greater than or equal to the first preset brightness and less than or equal to the second preset brightness; signal-to-noise ratio greater than the preset signal-to-noise ratio threshold; integrity greater than or equal to the preset integrity threshold, and whether the angle or area of ​​the target area is abnormal, etc.

[0060] If the contrast of each historical target image meets the standard but other evaluation indicators do not, then it is determined whether the corresponding historical images of each historical target image are successfully recognized. If the historical images are successfully recognized and the contrast meets the standard, then the correlation between contrast and recognition confidence is further determined. If the historical images fail to be recognized but the contrast meets the standard, then it is confirmed that contrast is not a target evaluation indicator and may be due to external environmental factors.

[0061] If other historical images are successfully recognized and their contrast meets the standard, the correlation between the contrast of each historical target image and the corresponding recognition confidence is obtained. If the evaluation score of the contrast in each historical target image is positively correlated with the recognition confidence of each historical target image, then the contrast is determined as the target evaluation index.

[0062] In some embodiments, after obtaining the results of each of the above judgment processes, the confidence level is calculated based on the results of each judgment process. As an example, when the contrast of each historical target acquisition image is less than the first contrast threshold, other evaluation indicators meet the standards, other historical images are successfully identified, and the correlation coefficient between contrast and recognition confidence is greater than the first correlation coefficient, the confidence level of the contrast is determined to be high confidence, and the confidence level is greater than the first confidence threshold; when the contrast of each historical target acquisition image is greater than or equal to the first contrast threshold and less than the second contrast threshold, at most one other evaluation indicator fails to meet the standards but is slightly abnormal, other historical images are successfully identified, and the correlation coefficient between contrast and recognition confidence is greater than the first correlation coefficient, the confidence level of the contrast is determined to be medium confidence, and the confidence level is less than or equal to the first confidence threshold and greater than the second confidence threshold; when the contrast of each historical target acquisition image is less than the first contrast threshold, at least two other evaluation indicators fail to meet the standards, other historical images have recognition failures, or the correlation coefficient between contrast and recognition confidence is less than the first correlation coefficient, the confidence level of the contrast is determined to be low confidence, and the confidence level is less than the second confidence threshold. The credibility level determines whether contrast is a target evaluation indicator. The higher the credibility level, the greater the likelihood that contrast is a target evaluation indicator.

[0063] Taking historical data with three failed recognition attempts as an example, the contrast ratios were 0.11, 0.1, and 0.09 (all below the first contrast threshold); the sharpness was 0.25, 0.26, and 0.24 (all normal); and the brightness was 0.45, 0.43, and 0.44 (normal). Other historical images from the same acquisition time showed a contrast ratio of 0.35, indicating successful recognition. The above data indicates high confidence, and contrast is the target evaluation indicator. Contrast ratios of 0.14, 0.13, and 0.15 (slightly below the threshold); sharpness of 0.12, 0.14, and 0.11 (all significantly low); and brightness of 0.42, 0.41, and 0.43 (normal) indicate low confidence, and contrast is not the target evaluation indicator.

[0064] In other embodiments, when selecting historical target images, images acquired under the same supplementary lighting parameters can also be selected. In other embodiments, if each historical target image is successfully identified even when its evaluation score for a certain evaluation indicator is significantly lower than the corresponding threshold, the preset weight of that evaluation indicator can be reduced to obtain the target weight of that evaluation indicator.

[0065] In other embodiments, environmental parameters and timestamps corresponding to the currently acquired image can also be obtained, and the current scene type can be determined based on these parameters and timestamps. The target weights corresponding to each evaluation indicator can be obtained from a preset mapping relationship based on the current scene type. This preset mapping relationship includes the correspondence between preset scene types and preset weights. For example, preset scene types may include strong backlight during the day, weak light at night, uniform indoor lighting, and cloudy outdoor lighting. Preset weights are set for each evaluation indicator under different preset scene types. The preset mapping relationship can be stored in flash memory and updated via OTA (Over-the-Air Technology).

[0066] In other embodiments, the weights of each evaluation metric can be dynamically adjusted through reinforcement learning. See Table 1 for details.

[0067] Table 1 The specific process can be as follows: the agent observes the current state; selects the corresponding action according to the strategy, such as increasing the preset weight of contrast; executes and observes the target recognition result in the next cycle; receives the reward and updates the network parameters; thereby continuously accumulating experience, obtaining the target weights of each evaluation index, and forming the optimal control strategy.

[0068] After obtaining the target weights corresponding to each evaluation indicator, the evaluation scores of each indicator are weighted and summed according to their respective target weights to obtain the image quality of the currently acquired image. For example, if the evaluation indicators include contrast, sharpness, signal-to-noise ratio, brightness, and integrity, the calculation of the image quality of the currently acquired image satisfies the following formula:

[0069] in, This indicates the image quality of the currently acquired image. This represents the target weight corresponding to the contrast. Indicates contrast. This represents the target weight corresponding to the clarity. Indicates sharpness, This represents the target weight corresponding to the signal-to-noise ratio. Indicates the signal-to-noise ratio. This represents the target weight corresponding to brightness. Indicates brightness. This represents the target weight corresponding to the completeness. Indicates completeness.

[0070] In other embodiments, after obtaining the image quality of the currently acquired image, the image quality of each currently acquired image can be adjusted based on historical recognition results to obtain an adjusted image quality. A target illumination parameter is determined from at least two preset illumination parameters based on the adjusted image quality of each currently acquired image. Target recognition processing is then performed based on the target illumination parameter and the currently acquired image corresponding to the target illumination parameter to obtain a target recognition result. For example, if the recognition success rate of a preset illumination parameter within a historical preset time period is greater than a first preset success rate, the image quality of the currently acquired image under the corresponding preset illumination parameter is increased to obtain an adjusted image quality, for example, by adding an additional preset adjustment value. If the recognition success rate of a preset illumination parameter within a historical preset time period is less than a second preset success rate, the image quality of the currently acquired image under the corresponding preset illumination parameter is decreased to obtain an adjusted image quality, for example, by subtracting an additional preset adjustment value.

[0071] Furthermore, when determining the preset supplementary lighting parameters, the preset supplementary lighting parameters for this identification can be selected from the preset supplementary lighting parameter library in various ways. In some embodiments, historical identification information of the recorded target can be obtained, and the preset supplementary lighting parameters can be determined based on the historical identification information of the recorded target. For example, if the recorded target is wearing glasses, and the historical identification information determines that the identification effect is best when the supplementary lighting wavelength is 1050nm, then 1050nm is selected for target identification. In other embodiments, the current environmental parameters (such as illumination, time, temperature, and humidity) can also be analyzed, and preset supplementary lighting parameters with a success rate greater than the preset success rate under the current environmental parameters can be matched from the historical scene database. In other embodiments, the ambient light intensity and spectral distribution of the current scene can also be detected in real time, and different preset supplementary lighting parameters can be selected for strong light environments and weak light environments. For example, a preset supplementary lighting parameter with a high supplementary lighting wavelength can be selected for a strong light environment, and a preset supplementary lighting parameter with a short supplementary lighting wavelength can be selected for a weak light environment. In other embodiments, the preset supplementary lighting parameters can be selected in a hierarchical manner according to the above methods. When all of the above information is lacking, any preset supplementary lighting parameter can be selected.

[0072] It is understandable that the number of preset supplementary lighting parameters can be dynamically adjusted according to the current scene. When a previously recorded target is detected in the current scene or the current scene is the same as a historically detected scene, the corresponding preset supplementary lighting parameter is directly selected, and target recognition can be completed in a single cycle. When a part of the current scene matches a historically detected scene, the currently acquired image corresponding to one of the preset supplementary lighting parameters is acquired. If the target recognition result indicates recognition failure, another preset supplementary lighting parameter is used to acquire the currently acquired image for target recognition until the target recognition result indicates successful recognition. Thus, by first using one preset supplementary lighting parameter for target recognition, if recognition is successful, the target recognition result can be obtained quickly, and at least in one attempt. In some other embodiments, at least two currently acquired images corresponding to preset supplementary lighting parameters can be acquired, and the target supplementary lighting parameter is determined from the at least two preset supplementary lighting parameters based on the image quality of each currently acquired image. Target recognition processing is performed based on the target supplementary lighting parameter and the currently acquired image corresponding to the target supplementary lighting parameter to obtain the target recognition result. In response to the target recognition result indicating recognition failure, a supplementary image in the current scene is acquired using supplementary lighting parameters. Target recognition processing is performed on the supplementary image to obtain the target recognition result. For example, you can first use a light source with a supplementary light wavelength of 850nm and 940nm to acquire the currently acquired image. If the recognition fails or the image quality is insufficient, you can then acquire the currently acquired image at 1050nm to perform target recognition.

[0073] Target recognition refers to the process performed under selected target illumination parameters, including target alignment, feature extraction, and comparison with a preset target database. Feature extraction can utilize lightweight models such as FaceNet and ArcFace, and the comparison with the preset target database supports 1:N recognition. When the target recognition result indicates successful recognition, the corresponding unlocking command is executed; when the target recognition result indicates failure, a log is recorded, including the corresponding target illumination parameters, image quality, and failure code. All target recognition results are written to local non-volatile storage (such as Flash or EEPROM) for subsequent analysis and model optimization. After recognition is complete, all light sources are turned off, the image acquisition device enters sleep mode, and the target recognition device enters a low-power standby state (e.g., current less than 5μA). If the target has not left and target recognition commands continue to be triggered, an energy-saving mode is activated, using only one preset illumination parameter for rapid recognition, reducing the number of scans with multiple preset illumination parameters, lowering power consumption, and supporting an automatic backoff mechanism: recognition is paused for 10 seconds after three consecutive recognition failures.

[0074] To further illustrate the application of this application in target recognition methods, the details are as follows: When the target recognition device detects that a target has entered the preset detection range, it enters the scanning preparation stage with multiple preset supplementary lighting parameters. The target recognition device sends a supplementary lighting parameter switching command to the light-emitting module, and after the light-emitting module is lit, it sends a frame synchronization trigger signal to the image acquisition device to simultaneously acquire the current acquisition image corresponding to multiple preset supplementary lighting parameters.

[0075] Initial detection is performed on the currently acquired image to determine the target region where the target is located. Quality assessment is then performed on the target region to obtain the image quality of the currently acquired image. This assessment involves a comprehensive evaluation of multiple evaluation metrics, obtaining evaluation scores for each metric, and then using the target weights of these metrics to perform a weighted summation, ultimately yielding the image quality of the currently acquired image. The target weights for each evaluation metric can be fixed or adjusted based on historical recognition results and the evaluation scores of each metric during the historical recognition process. For example, if a preset illumination parameter fails to be evaluated N times consecutively due to a particular evaluation metric, the weight of that metric is temporarily increased until the next successful recognition. The target weights for each evaluation metric can also be determined based on the environmental parameters and timestamp corresponding to the currently acquired image. Furthermore, the target weights for each evaluation metric can be optimized using reinforcement learning methods.

[0076] Based on the image quality of each currently acquired image, target supplementary lighting parameters are selected from the preset supplementary lighting parameters corresponding to each currently acquired image; the selected target supplementary lighting parameters are written into the control register for target recognition; if recognition is successful, an unlock command is output; if recognition fails, a rejection command is output and the reason for failure is recorded; the target supplementary lighting parameters, the image quality of each currently acquired image, and the target recognition result are written into the historical database for subsequent use; after recognition is completed, the light-emitting module is turned off and enters a low-power standby mode.

[0077] Please see Figure 2 , Figure 2 This is a schematic diagram of an exemplary embodiment of the target recognition device shown in this application. The target recognition device 200 includes an acquisition module 210, a determination module 220, and a recognition module 230. The acquisition module 210 is used to acquire currently acquired images corresponding to at least two preset supplementary lighting parameters in response to receiving a target recognition instruction. The determination module 220 is used to determine target supplementary lighting parameters from at least two preset supplementary lighting parameters based on the image quality of each currently acquired image. The recognition module 230 is used to perform target recognition processing based on the target supplementary lighting parameters and the currently acquired images corresponding to the target supplementary lighting parameters to obtain a target recognition result.

[0078] In the above scheme, the target recognition device, upon receiving a target recognition command, acquires at least two currently captured images corresponding to preset supplementary lighting parameters; determines target supplementary lighting parameters from the at least two preset supplementary lighting parameters based on the image quality of each currently captured image; and performs target recognition processing based on the target supplementary lighting parameters and the corresponding currently captured target images to obtain the target recognition result. This scheme improves target recognition performance by dynamically selecting the optimal supplementary lighting scheme for the current scene by comparing the image quality of currently captured images under different preset supplementary lighting parameters.

[0079] The functions of each module can be found in the target recognition method implementation examples, and will not be repeated here.

[0080] To implement the target recognition method of the above embodiments, this application proposes another electronic device, please refer to [link / reference needed]. Figure 3 , Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device provided in this application.

[0081] Electronic device 300 includes memory 310 and processor 320, wherein memory 310 and processor 320 are coupled together.

[0082] The memory 310 is used to store program data, and the processor 320 is used to execute the program data to implement the target recognition method of the above embodiment.

[0083] In this embodiment, processor 320 can also be referred to as CPU (Central Processing Unit). Processor 320 may be an integrated circuit chip with signal processing capabilities. Processor 320 can also be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component. The general-purpose processor can be a microprocessor, or processor 320 can be any conventional processor.

[0084] This application also provides a computer-readable storage medium, such as Figure 4 As shown, the computer-readable storage medium 400 is used to store program data 410, which, when executed by a processor, is used to implement the target recognition method as described in the method embodiments of this application.

[0085] The methods involved in the target identification method embodiments of this application, when implemented as software functional units and sold or used as independent products, can be stored in a device, such as a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0086] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0087] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The term "and / or" is merely a description of the association of related objects, indicating that three relationships can exist; for example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects are in an "or" relationship. Furthermore, "many" in this document means two or more. In addition, the term "at least one" in this document means any combination of at least two of any one or more of a plurality of elements, such as including at least one of A, B, and C, and may mean including any one or more elements selected from the set consisting of A, B, and C.

Claims

1. A target recognition method, characterized in that, The target recognition method includes: In response to receiving a target recognition command, acquire the currently acquired image corresponding to at least two preset supplementary lighting parameters; The target supplementary lighting parameters are determined from the at least two preset supplementary lighting parameters based on the image quality of each currently acquired image; Target recognition processing is performed based on the target illumination parameters and the currently acquired image corresponding to the target illumination parameters to obtain the target recognition result.

2. The target recognition method according to claim 1, characterized in that, The step of determining the target illumination parameter from the at least two preset illumination parameters based on the image quality of each currently acquired image includes: Each currently acquired image is subjected to quality assessment processing to obtain the image quality of each currently acquired image; If the image quality of the currently acquired image is greater than or equal to a preset quality threshold, then the preset supplementary lighting parameter corresponding to the currently acquired image is determined as the target supplementary lighting parameter.

3. The target recognition method according to claim 2, characterized in that, The step of performing quality assessment processing on each currently acquired image to obtain the image quality of each currently acquired image includes: For each currently acquired image, obtain evaluation scores for at least two evaluation metrics; The evaluation scores of each evaluation indicator are weighted according to the target weights corresponding to each evaluation indicator to obtain the image quality of the currently acquired image.

4. The target recognition method according to claim 3, characterized in that, Before the step of weighting the evaluation scores of each evaluation indicator according to the target weights corresponding to each evaluation indicator to obtain the image quality of the currently acquired image, the method further includes: Obtain the historical recognition results corresponding to at least one historical target image and the evaluation scores of each evaluation index corresponding to each historical target image. The acquisition time of each historical target image is different. Based on the historical recognition results of each historical target image and the evaluation scores of each evaluation indicator corresponding to each historical target image, the impact assessment of each evaluation indicator is carried out to obtain the target weight corresponding to each evaluation indicator.

5. The target recognition method according to claim 4, characterized in that, The step of performing identification impact assessment processing on each evaluation indicator based on the historical recognition results of each historical target image and the evaluation scores of each evaluation indicator corresponding to each historical target image, to obtain the target weight corresponding to each evaluation indicator, includes: In response to the historical recognition results corresponding to each historical target image indicating recognition failure, the target evaluation index that caused the recognition failure is obtained from each evaluation index based on the evaluation score of each evaluation index in each historical target image. The preset weights of the target evaluation indicators are increased to obtain the target weights of the target evaluation indicators, and the preset weights of the other evaluation indicators besides the target evaluation indicators are used as the corresponding target weights.

6. The target recognition method according to claim 5, characterized in that, The step of obtaining the target evaluation index that caused the recognition failure from each evaluation index based on the evaluation score of each evaluation index in each historical target acquired image includes: For each historical target image, candidate evaluation indicators with substandard evaluation scores are selected from various evaluation indicators. In response to the fact that the candidate evaluation indicators of each historical target acquisition image are the same and the number of indicators of the candidate evaluation indicators is less than or equal to a preset threshold, other historical images corresponding to each historical target acquisition image are obtained. The other historical images are acquired based on the same scene as the historical target acquisition images, and the supplementary lighting parameters of the historical target acquisition images and the other historical images are different. If the evaluation score of the candidate evaluation index corresponding to the other historical images meets the standard and the historical recognition result of the other historical images indicates successful recognition, then the candidate evaluation index is determined as the target evaluation index.

7. The target recognition method according to claim 5, characterized in that, The historical recognition results include recognition confidence scores. The step of obtaining the target evaluation index that caused the recognition failure from each evaluation index based on the evaluation scores of each evaluation index in the historical target acquisition images includes: For each historical target image, candidate evaluation indicators with substandard evaluation scores are selected from various evaluation indicators. In response to the fact that the candidate evaluation indicators of each historical target acquisition image are the same and the number of the candidate evaluation indicators is less than a preset threshold, the correlation between the evaluation score of the candidate evaluation indicator in each historical target acquisition image and the recognition confidence level corresponding to each historical target acquisition image is obtained. If the evaluation scores of candidate evaluation indicators in each historical target acquisition image are positively correlated with the recognition confidence level of each historical target acquisition image, then the candidate evaluation indicator is determined as the target evaluation indicator.

8. The target recognition method according to claim 1, characterized in that, After the step of performing target recognition processing based on the target illumination parameters and the current target acquisition image corresponding to the target illumination parameters to obtain the target recognition result, the method further includes: If the target recognition result indicates that the recognition has failed, supplementary lighting parameters are used to acquire a supplementary image of the current scene. The supplementary image is subjected to target recognition processing to obtain the target recognition result.

9. An electronic device, characterized in that, include: A memory and a processor, wherein the memory stores program instructions, and the processor retrieves the program instructions from the memory to perform the method as claimed in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, include: The system stores program data, which, when executed by a processor, is used to implement the method as described in any one of claims 1-8.

Images