A target identification method and device, and an identification and storage medium

Abstract

Embodiments of the present application disclose a target identification method, device and equipment and a storage medium. The method comprises: performing feature extraction and target identification on an obtained initial image to be adjusted to determine at least one target image region; determining a target image region that does not meet a preset image processing condition in each target image region as an image region to be processed, and determining an exposure consistency type of each image region to be processed; adjusting an exposure parameter according to the exposure consistency type and each image region to be processed, and obtaining a target image under the adjusted exposure parameter; performing feature extraction and target identification on the target image to determine a target identification result. The technical solution of the embodiments of the present application does not need to adjust the visual sensor hardware of the collected image for obtaining the target image, and the target identification based on the target image obtained after adjusting the exposure parameter improves the confidence of the target identification result obtained by identification, does not need to improve the image transmission bandwidth, and reduces the target identification cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of visual perception, and in particular to a target identification method, device, equipment and storage medium. BACKGROUND

[0002] With the development of the times, people's living quality is continuously improved, and cars gradually enter thousands of households and become one of the main means of transportation. In the application of intelligent driving function, visual sensors gradually become an indispensable sensor type. In the process of high-speed driving of the vehicle, the images collected by the visual sensor often do not meet the needs of the back-end algorithm processing.

[0003] The visual sensor provided in the existing vehicle often relies on the exposure adjustment power provided in the visual sensor itself, adjusts the exposure parameters according to the light amount of the entire visual sensor, and can increase the high dynamic range mode to make the color and details of the image collected in the mode more outstanding.

[0004] However, simply relying on the light amount adjustment of the exposure parameter is difficult to finely adjust the overall exposure of the image obtained by the visual sensor, and using the high dynamic range mode for image acquisition puts higher requirements on the image transmission bandwidth of the visual sensor, making it difficult to balance the image acquisition quality and cost. SUMMARY

[0005] The present application provides a target identification method, device, equipment and storage medium, which dynamically adjusts the output image of the visual sensor without increasing the hardware cost, improves the confidence of target identification based on the adjusted image, and further improves the performance of visual perception, the accuracy of target identification, and the user driving experience.

[0006] In a first aspect, the present application embodiment provides a target identification method, comprising:

[0007] performing feature extraction and target identification on the obtained initial to-be-adjusted image to determine at least one target image region;

[0008] determining a target image region in each target image region that does not meet the preset image processing condition as a to-be-processed image region, and determining the exposure consistency type of each to-be-processed image region;

[0009] adjusting the exposure parameter according to the exposure consistency type and each to-be-processed image region, and obtaining a target image under the adjusted exposure parameter;

[0010] performing feature extraction and target identification on the target image to determine a target identification result.

[0011] In a second aspect, the present application embodiment further provides a target identification device, comprising:

[0012] a target region determination module configured to perform feature extraction and target recognition on the obtained initial image to be adjusted, and determine at least one target image region;

[0013] a consistency determination module configured to determine, as an image region to be processed, a target image region that does not satisfy a preset image processing condition in each target image region, and determine an exposure consistency type of each image region to be processed;

[0014] a target image acquisition module configured to adjust an exposure parameter according to the exposure consistency type and each image region to be processed, and acquire a target image under the adjusted exposure parameter;

[0015] a recognition result determination module configured to perform feature extraction and target recognition on the target image, and determine a target recognition result.

[0016] In a third aspect, an embodiment of the present application further provides a target recognition device, comprising:

[0017] at least one processor; and

[0018] a memory in communication connection with the at least one processor; wherein

[0019] the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the target recognition method of any embodiment of the present application.

[0020] In a fourth aspect, an embodiment of the present application further provides a storage medium containing computer executable instructions, which, when executed by a computer processor, enable the computer processor to execute the target recognition method of any embodiment of the present application.

[0021] The target identification method, device, equipment and storage medium provided by the embodiment of the application determine at least one target image region by performing feature extraction and target identification on the obtained initial to-be-adjusted image; determine the target image regions that do not meet the preset image processing condition as to-be-processed image regions, and determine the exposure consistency type of each to-be-processed image region; adjust the exposure parameter according to the exposure consistency type and each to-be-processed image region, and obtain the target image under the adjusted exposure parameter; perform feature extraction and target identification on the target image, and determine the target identification result. By using the above technical solution, after the feature extraction and target identification on the obtained initial to-be-adjusted image are performed, the identification result is not directly transmitted to the subsequent data processing module for use, but whether there is an image region that does not meet the preset image processing condition in each target image region identified is judged, the exposure parameter of the visual sensor is adjusted according to the exposure consistency type of the to-be-processed image region that does not meet the preset image processing condition, and the target image is reacquired based on the adjusted exposure parameter, and then the target identification of the target image is performed to obtain the corresponding target identification result. The acquisition of the target image does not need to adjust the visual sensor hardware of the collected image, that is, the image transmission bandwidth does not need to be improved, the target identification cost is reduced, the confidence of the target identification result obtained by the target identification based on the target image acquired after the exposure parameter is adjusted is improved, the performance of the visual perception is improved, and then the user driving experience is improved.

[0022] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the application, nor is it intended to limit the scope of the application. Other features of the application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0024] Figure 1 The flowchart of a target identification method provided for the first embodiment of the application;

[0025] Figure 2 The flowchart of another target identification method provided for the second embodiment of the application;

[0026] Figure 3 The structural schematic diagram of a target identification device provided for the third embodiment of the application;

[0027] Figure 4A structural schematic diagram of a target identification device provided for the fourth embodiment of the present application. DETAILED DESCRIPTION

[0028] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] Embodiment One

[0031] Figure 1 A flowchart of a target identification method provided for the first embodiment of the present application, the present embodiment can be applicable to adjusting a visual sensor in an intelligent driving process to obtain an image more suitable for target identification for target identification, the method can be executed by a target identification device, the target identification device can be realized in the form of hardware and / or software, the target identification device can be configured in target identification, and the target identification device can be a notebook, a desktop computer, a smart tablet or a vehicle, etc., and the present application does not limit this.

[0032] As shown in Figure 1 A target identification method provided by the present application, specifically comprising the following steps:

[0033] S101, performing feature extraction and target identification on the obtained initial to-be-adjusted image to determine at least one target image region.

[0034] In the embodiment, the initial image to be adjusted can be understood as an image directly collected by the visual sensor after adjusting the exposure parameter according to its own characteristics, and the image is only subjected to simple preprocessing such as noise reduction and format conversion and then input for subsequent data processing such as target recognition. The target image region can be understood as the position region of a target object identified from the initial image to be adjusted. It can be understood that multiple target objects can exist in the collected image, such as multiple vehicles and road barriers in the image collected during road driving, wherein each vehicle and road barrier can be a target object, and the region occupied in the image after target recognition is the target image region.

[0035] Specifically, during vehicle driving, the visual sensor arranged inside or outside the vehicle can collect surrounding environment information, the image collected after the visual sensor automatically adjusts the exposure parameter according to the light amount is determined as the initial image to be adjusted, the image features of the initial image to be adjusted are extracted, the target recognition is performed on the feature map obtained after the feature extraction, multiple target objects existing in the initial image to be adjusted are determined, and the region of each target object in the initial image to be adjusted is determined as the target image region. It can be understood that there are multiple selectable implementation methods for the feature extraction and target recognition of the image in the prior art, and a suitable image processing method can be selected according to the actual situation to complete the feature extraction and target recognition of the initial image to be adjusted, and the embodiment of the present application does not limit this.

[0036] S102, determining each target image region in which the target image region does not satisfy the preset image processing condition as a to-be-processed image region, and determining the exposure consistency type of each to-be-processed image region.

[0037] In the embodiment, the preset image processing condition can be understood as a condition for indicating whether the image state of the determined target image region is sufficient to ensure the target recognition accuracy, which is set in advance according to the actual situation. It should be noted that the authenticity of the target image recognized can be determined by the confidence, but the confidence of target recognition is related to multiple factors, such as the low dynamic range of a certain target image region in the image, which can lead to low target recognition confidence of the region. However, the low dynamic range does not necessarily lead to low target recognition confidence, because the dynamic range required for target recognition is uncertain for different scenes and objects, and therefore an image processing condition can be set to determine whether the image collected from each target image region is sufficient to meet the target recognition accuracy. The to-be-processed image region can be understood as an image region whose current image state is difficult to ensure the target recognition accuracy and needs to be processed again. The exposure consistency type can be understood as a type parameter for indicating whether the exposure states of different image regions are consistent.

[0038] Specifically, if there is a target image region that does not meet the preset image processing condition in all target image regions determined from the initial image to be adjusted, the target image region is regarded as a to-be-processed image region. It can be considered that the accuracy of the target object identified by the to-be-processed image region cannot be guaranteed, and the to-be-processed image region does not meet the input to the downstream intelligent driving function. At this time, the exposure state of each to-be-processed image region can be confirmed, that is, it is determined whether each to-be-processed image region has overexposure or underexposure relative to the overall exposure state of the initial image to be adjusted. Then, according to whether the exposure states are consistent or not, the exposure consistency type of each to-be-processed image region is determined.

[0039] S103, adjusting the exposure parameter according to the exposure consistency type and each to-be-processed image region, and obtaining a target image under the adjusted exposure parameter.

[0040] Specifically, according to the exposure consistency type and each to-be-processed image region, it is determined that there is a problem of overexposure or underexposure in each to-be-processed image region, that is, each to-be-processed image region can be divided into an overexposed region and an underexposed region. At the same time, it can be clear that the underexposed region needs to increase the exposure amount, and the overexposed region needs to reduce the exposure amount. The exposure parameter of the vision sensor is adjusted according to the required increase in exposure amount and the required decrease in exposure amount, and the image collected by the vision sensor after adjusting the exposure parameter is determined as the target image.

[0041] S104, performing feature extraction and target recognition on the target image to determine a target recognition result.

[0042] In this embodiment, the target recognition result can be understood as a set of target objects whose confidence meets the subsequent use of the intelligent driving function after target recognition of the target image.

[0043] Specifically, the target image obtained after adjusting the exposure parameter is subjected to feature extraction and target recognition to determine the image regions where the identified multiple targets are located. Then, the confidence of the identified targets in each image region is judged, and the set of image regions whose confidence meets the preset demand condition is determined as the target recognition result. The target recognition result can be input into the downstream intelligent driving function module for use in the form of path planning and obstacle avoidance.

[0044] The technical scheme of the embodiment is characterized in that: feature extraction and target recognition are performed on the obtained initial image to be adjusted to determine at least one target image region; a target image region that does not meet the preset image processing condition in each target image region is determined as an image region to be processed, and the exposure consistency type of each image region to be processed is determined; the exposure parameter is adjusted according to the exposure consistency type and each image region to be processed, and a target image under the adjusted exposure parameter is obtained; feature extraction and target recognition are performed on the target image to determine a target recognition result. By using the above technical scheme, after feature extraction and target recognition are performed on the obtained initial image to be adjusted, the recognition result is not directly transmitted to the subsequent data processing module for use, but whether there is an image region that does not meet the preset image processing condition in each target image region recognized is judged, the exposure parameter of the visual sensor is adjusted according to the exposure consistency type of the image region to be processed that does not meet the preset image processing condition, and the target image is reacquired based on the adjusted exposure parameter, and then the target image can be recognized to obtain the corresponding target recognition result. The acquisition of the target image does not require adjustment of the visual sensor hardware of the collected image, that is, the image transmission bandwidth does not need to be improved, the target recognition cost is reduced, the confidence of the target recognition result obtained by target recognition based on the target image acquired after adjustment of the exposure parameter is improved, the performance of visual perception is improved, and the user driving experience is improved.

[0045] Embodiment Two

[0046] Figure 2 The flowchart of another target recognition method provided by Embodiment Two of the application, the technical scheme of the embodiment is further optimized on the basis of the above-mentioned optional technical schemes, whether each target image region meets the preset image processing condition is determined based on the image dynamic range and target recognition confidence of each target image region, and in the case where it is clear that image reacquisition is needed, the exposure consistency of each image region to be processed is determined by the brightness difference between the image region to be processed and the initial image to be adjusted, the exposure parameter is adjusted based on the exposure consistency and each brightness difference, and the target image under the corresponding adjusted exposure parameter is obtained, only the image of the part of the target image that needs to be adjusted is used for corresponding target recognition, and the target recognition result is determined by comprehensively recognizing the target objects in each target image. The influence of the image dynamic range on target recognition is fully considered, high-quality images are obtained by comprehensively adjusting the exposure parameter and reacquiring the target image without hardware upgrade, the confidence of the target recognition result determined based on the high-quality target image is improved, the target recognition cost is reduced, and the performance of visual perception is improved.

[0047] As Figure 2As shown, the target recognition method provided in the second embodiment of the present application specifically comprises the following steps:

[0048] S201, performing feature extraction and target recognition on the obtained initial image to be adjusted to determine at least one target image region.

[0049] S202, determining the corresponding image dynamic range and target recognition confidence for each target image region.

[0050] In this embodiment, the image dynamic range can be specifically understood as a ratio between the brightest and darkest shades that can be recorded in the target image region, which can be used to indicate the range of tone information that the visual sensor can capture in the target image region. The target recognition confidence can be specifically understood as a possibility of determining whether there is an object in the target image region. The higher the target recognition confidence, the higher the possibility of the target image region containing an object that can be used by the incoming downstream intelligent driving function, and the target image region should be adopted.

[0051] Specifically, according to each target image region, the channel information of each pixel point in the corresponding image thereof can be determined, i.e., the hue, saturation, and brightness of each pixel point can be determined, and then the ratio between the brightest and darkest shades in the target image region can be determined to obtain the image dynamic range corresponding to the target image region. At the same time, since the target recognition algorithm for the image can output the confidence of the detection frame obtained by target recognition, which indicates the possibility of recognizing that there is a target to be recognized in the detection frame, the confidence of the detection frame corresponding to the target image region obtained by the target recognition algorithm can be determined as the target recognition confidence of the target image region.

[0052] S203, if the image dynamic range is less than a preset dynamic range threshold and the target recognition confidence is less than a preset confidence threshold, the target image region is determined as an image region to be processed.

[0053] In this embodiment, the preset dynamic range threshold can be specifically understood as a range threshold that is set in advance according to actual conditions and is used to determine whether the dynamic range will cause the confidence in the target image region to decrease. The preset confidence threshold can be specifically understood as a confidence value that is set in advance according to actual conditions and is used to determine whether there is a target to be recognized in the target image region.

[0054] Specifically, for each target image region, if the image dynamic range corresponding to the target image region is less than the preset dynamic range threshold and the target recognition confidence is less than the preset confidence threshold, it can be considered that the target image region contains a low possibility of target that can be used for downstream intelligent driving function, and it can be considered that the target image region does not meet the preset image processing condition, that is, the clarity of the target image region collected under the original exposure parameter is insufficient, and the exposure parameter should be adjusted again to collect a new image for target recognition. At this time, the target image region that does not meet the preset dynamic range threshold and the preset confidence threshold is determined as the image region to be processed.

[0055] S204, determine the brightness difference value of each image region to be processed and the initial image to be adjusted.

[0056] Specifically, based on the brightness value of each pixel point in the initial image to be adjusted, the overall brightness value of the initial image to be adjusted is determined, and based on the brightness value of each pixel point in each image region to be processed, the region brightness value of the corresponding image of each image region to be processed is determined. For each image region to be processed, the region brightness value is subtracted from the overall brightness value to determine the brightness difference value of the image region to be processed. It can be understood that when the brightness difference value is positive, it can be considered that the image in the image region to be processed has an overexposure problem, and when the brightness difference value is negative, it can be considered that the image in the image region to be processed has an underexposure problem.

[0057] S205, determine whether each brightness difference value is greater than zero or less than zero, if yes, execute step S206; if no, execute step S215.

[0058] Specifically, when each brightness difference value is greater than zero or less than zero, it can be considered that all image regions to be processed are in the same exposure state, that is, overexposure or underexposure. At this time, step S206 is executed. If there are both greater than zero and less than zero in each brightness difference value, it can be considered that there are both image regions with overexposure problems and regions with underexposure problems in each image region to be processed, that is, all image regions to be processed are not in the same exposure state. At this time, step S215 is executed.

[0059] S206, determine that the exposure consistency type of each image region to be processed is consistent, and when the brightness difference value corresponding to each image region to be processed is greater than zero, execute step S207; when the brightness difference value corresponding to each image region to be processed is less than zero, execute step S211.

[0060] Specifically, when the exposure consistency type of each to-be-processed image region is consistent, and the luminance difference corresponding to each to-be-processed image region is greater than zero, it can be considered that each to-be-processed image region has an image overexposure problem, and step S207 is performed. When the exposure consistency type of each to-be-processed image region is consistent, and the luminance difference corresponding to each to-be-processed image region is less than zero, it can be considered that each to-be-processed image region has an image underexposure problem, and step S211 is performed.

[0061] S207, determining the positive luminance difference target image obtained after adjusting the exposure parameter according to the maximum luminance difference in the luminance difference as the target image.

[0062] Specifically, the maximum positive luminance difference in the luminance difference is determined as the maximum positive luminance difference. In order to adjust the overexposure problem of the visual sensor, the exposure parameter in the visual sensor is adjusted according to the maximum positive luminance difference, and the target image obtained by the visual sensor after the exposure parameter adjustment is determined as the positive luminance difference target image.

[0063] S208, performing feature extraction and target recognition on the positive luminance difference target image to determine at least one first target recognition result and a first target recognition confidence of each first target recognition result.

[0064] Specifically, the positive luminance difference target image is subjected to feature extraction and target recognition to determine the image regions in which a plurality of recognized targets are located. The target type and target position information contained in each image region are determined as the first target recognition result, and the confidence of the detection frame corresponding to the first target recognition result is determined as the first target recognition confidence.

[0065] S209, determining the first target recognition confidence greater than the preset confidence threshold as the first reserved confidence.

[0066] Specifically, each first target recognition confidence is compared with the preset confidence threshold. If the first target recognition confidence is greater than the preset confidence threshold, it can be considered that the image region corresponding to the first target recognition confidence contains a target that is more likely to be used by a downstream intelligent driving function. At this time, the first target recognition confidence is determined as the first reserved confidence.

[0067] S210, determining the set of first target recognition results corresponding to each first reserved confidence as the target recognition result.

[0068] Specifically, the object recognized in the image region corresponding to the first reserved confidence is determined as the first target recognition result, and the set of each first target recognition result corresponding to each first reserved confidence is determined as the target recognition result.

[0069] S211、determine the negative brightness difference target image obtained after adjusting the exposure parameter according to the maximum brightness difference value in each brightness difference value as the target image.

[0070] Specifically, the negative brightness difference value with the largest absolute value in each brightness difference value is determined as the maximum negative brightness difference value. In order to adjust the underexposure problem in the visual sensor, the exposure parameter in the visual sensor is adjusted according to the maximum negative brightness difference value, and the target image obtained by the visual sensor after the exposure parameter adjustment is determined as the negative brightness difference target image.

[0071] S212, perform feature extraction and target recognition on the negative brightness difference target image, determine at least one second target recognition result, and a second target recognition confidence of each second target recognition result.

[0072] Specifically, the negative brightness difference target image is subjected to feature extraction and target recognition, and the image regions where the multiple targets recognized are determined. The target type and target position information contained in each image region are determined as the second target recognition result, and the confidence of the detection box corresponding to the second target recognition result is determined as the second target recognition confidence.

[0073] S213, determine the second target recognition confidence greater than the preset confidence threshold as the second reserved confidence.

[0074] Specifically, each second target recognition confidence is compared with the preset confidence threshold. If the second target recognition confidence is greater than the preset confidence threshold, it can be considered that the image region corresponding to the second target recognition confidence contains a target with a higher possibility of being used by the downstream intelligent driving function. At this time, the second target recognition confidence is determined as the second reserved confidence.

[0075] S214, determine the set of second target recognition results corresponding to each second reserved confidence as the target recognition result.

[0076] Specifically, the object recognized in the image region corresponding to the second reserved confidence is determined as the second target recognition result, and the set of each second target recognition result corresponding to each second reserved confidence is determined as the target recognition result.

[0077] S215, determine that the exposure consistency type of each to-be-processed image region is not consistent.

[0078] Specifically, when there are both greater than zero and less than zero in each brightness difference value, it can be considered that the exposure states of different to-be-processed image regions are different, that is, part of the to-be-processed image regions have overexposure problems, and another part of the to-be-processed image regions have underexposure problems. At this time, the exposure consistency type is determined as not consistent.

[0079] It can be understood that it is difficult to meet the requirement of normally completing target recognition for image regions to be processed with different exposure problems by adjusting exposure parameters only once, that is, when the exposure consistency type is not consistent, it is necessary to complete exposure parameter adjustment and target image acquisition for overexposure and underexposure respectively, and the specific execution mode is shown in the following steps.

[0080] S216, determining the maximum positive luminance difference value and the maximum negative luminance difference value according to the luminance difference values corresponding to each image region to be processed.

[0081] Specifically, for the positive luminance difference values in each luminance difference value, the positive luminance difference value with the largest absolute value is determined as the maximum positive luminance difference value. For the negative luminance difference values in each luminance difference value, the negative luminance difference value with the largest absolute value is determined as the maximum negative luminance difference value. The maximum positive luminance difference value and the maximum negative luminance difference value can be used to indicate the luminance information of the maximum overexposure and the maximum underexposure problems in the initial image to be adjusted, respectively.

[0082] S217, determining the positive luminance difference target image obtained after adjusting the exposure parameter according to the maximum positive luminance difference value and the negative luminance difference target image obtained after adjusting the exposure parameter according to the maximum negative luminance difference value as the target image.

[0083] Specifically, the image obtained by the vision sensor after adjusting the exposure parameter in the vision sensor according to the maximum positive luminance difference value is determined as the positive luminance difference target image; and the image obtained by the vision sensor after adjusting the exposure parameter in the vision sensor according to the maximum negative luminance difference value is determined as the negative luminance difference target image. The positive luminance difference target image and the negative luminance difference target image are both images obtained after adjusting the exposure parameter, and are also images that can be used for subsequent secondary target recognition. Therefore, they are unified as the target image corresponding to the image region to be processed.

[0084] S218, determining the positive luminance image range in the positive luminance difference target image and the negative luminance image range in the negative luminance difference target image according to the luminance difference values corresponding to each image region to be processed.

[0085] Specifically, since the obtained positive brightness difference target image has the same size as the initial to-be-adjusted image, the positions of each to-be-processed image region in the initial to-be-adjusted image are the same as the positions in the positive brightness difference target image. The positions corresponding to the to-be-processed image region with a positive brightness difference value are determined as the positive brightness image range in the positive brightness difference target image, that is, the image in the positive brightness image range needs to be subjected to secondary feature extraction and target recognition to replace the target recognition result in the initial to-be-adjusted image with a confidence that does not meet the requirement. The positions corresponding to the to-be-processed image region with a negative brightness difference value are determined as the negative brightness image range in the negative brightness difference target image, that is, the image in the negative brightness image range needs to be subjected to secondary feature extraction and target recognition to replace the target recognition result in the initial to-be-adjusted image with a confidence that does not meet the requirement.

[0086] S219, performing feature extraction and target recognition on the image in the positive brightness image range in the positive brightness difference target image to determine at least one positive brightness target recognition result and a first confidence group corresponding to each positive brightness target recognition result.

[0087] Specifically, the feature extraction and target recognition are performed on the image in the positive brightness image range in the positive brightness difference target image to obtain the positive brightness target recognition result corresponding to each to-be-processed image region in the positive brightness image range. Each positive brightness target recognition result should include the corresponding target type and target position and the like. The confidence of the detection frame corresponding to the positive brightness target recognition result is determined as the first confidence, and the set of each first confidence is determined as the first confidence group.

[0088] S220, performing feature extraction and target recognition on the image in the negative brightness image range in the negative brightness difference target image to determine at least one negative brightness target recognition result and a second confidence group corresponding to each negative brightness target recognition result.

[0089] Specifically, the feature extraction and target recognition are performed on the image in the negative brightness image range in the negative brightness difference target image to obtain the negative brightness target recognition result corresponding to each to-be-processed image region in the negative brightness image range. Each negative brightness target recognition result should include the corresponding target type and target position and the like. The confidence of the detection frame corresponding to the negative brightness target recognition result is determined as the second confidence, and the set of each second confidence is determined as the second confidence group.

[0090] S221, determining the first confidence and the second confidence in the first confidence group and the second confidence group that are greater than a preset confidence threshold as a reserved confidence.

[0091] Specifically, each first confidence in the first confidence group is compared with a preset confidence threshold respectively, if the first confidence is greater than the preset confidence threshold, it can be considered that the image region corresponding to the first confidence contains a target with a higher possibility of being used by the downstream intelligent driving function, at this time, the first confidence is determined as a reserved confidence. Similarly, each second confidence in the second confidence group is compared with a preset confidence threshold respectively, if the second confidence is greater than the preset confidence threshold, it can be considered that the image region corresponding to the second confidence contains a target with a higher possibility of being used by the downstream intelligent driving function, at this time, the second confidence is determined as a reserved confidence.

[0092] S222, the set of positive brightness target recognition results and negative brightness target recognition results corresponding to each reserved confidence is determined as a target recognition result.

[0093] Specifically, the objects recognized in each reserved confidence corresponding image region, that is, the set of positive brightness target recognition results recognized in the positive brightness difference target image and the negative brightness target recognition results recognized in the negative brightness difference target image, is determined as a target recognition result.

[0094] The technical scheme of the embodiment is based on the image dynamic range of each target image region and the target recognition confidence to determine whether each target image region meets the preset image processing condition, and in the case where it is clear that image reacquisition needs to be performed, the adjustment of the exposure parameter is completed based on the maximum positive brightness difference value and the maximum negative brightness difference value corresponding to each target image region, and the target image corresponding to the adjusted exposure parameter is obtained respectively, only the image of the part of the target image that needs to be adjusted is used for corresponding target recognition, and the target recognition result is determined by comprehensively considering the target objects recognized in each target image and the confidence corresponding to each target object, the influence of the image dynamic range on target recognition is fully considered, the exposure parameter is adjusted and the target image is reacquired, the higher quality image is obtained without hardware upgrade, and the confidence of the target recognition result determined based on the high-quality target image is improved, the target recognition cost is reduced, and the visual perception performance is improved.

[0095] Embodiment three

[0096] Figure 3 A structure schematic diagram of a target recognition device provided by the third embodiment of the present application is provided, which comprises a target region determination module 31, a consistency determination module 32, a target image acquisition module 33 and an identification result determination module 34.

[0097] The target region determination module 31 is configured to perform feature extraction and target recognition on the obtained initial image to be adjusted to determine at least one target image region.

[0098] The technical scheme of the embodiment is not directly transmitted to the subsequent data processing module for use after performing feature extraction and target recognition on the obtained initial image to be adjusted, but determines whether there is an image region that does not meet the preset image processing condition in each target image region obtained by recognition, adjusts the exposure parameter of the visual sensor according to the exposure consistency type of the image region to be processed that does not meet the preset image processing condition, reacquires the target image under the adjusted exposure parameter, and then performs target recognition on the target image to obtain the corresponding target recognition result. The target image is acquired without adjusting the visual sensor hardware for collecting the image, that is, without increasing the image transmission bandwidth, thereby reducing the target recognition cost. The target recognition based on the target image acquired after adjusting the exposure parameter improves the confidence of the target recognition result obtained by recognition, improves the performance of visual perception, and further improves the user driving experience.

[0099] Optionally, the consistency determination module 32 comprises:

[0100] The image parameter determination unit is configured to determine, for each target image region, the corresponding image dynamic range and target recognition confidence.

[0101] The to-be-processed region determination unit is configured to determine the target image region as a to-be-processed image region if the image dynamic range is less than a preset dynamic range threshold and the target recognition confidence is less than a preset confidence threshold.

[0102] The brightness difference determination unit is configured to determine the brightness difference between each to-be-processed image region and the initial image to be adjusted.

[0103] The consistency determination unit is configured to determine that the exposure consistency type of each to-be-processed image region is consistent if each brightness difference is greater than zero or less than zero, or otherwise, determine that the exposure consistency type of each to-be-processed region is inconsistent.

[0104] Optionally, the target image acquisition module 33 is specifically configured to:

[0105] If the exposure consistency type is consistent, and the luminance difference values corresponding to each of the image regions to be processed are all greater than zero, a positive luminance difference target image obtained by adjusting the exposure parameter according to the maximum luminance difference value among the luminance difference values is determined as the target image.

[0106] If the exposure consistency type is consistent, and the luminance difference values corresponding to each of the image regions to be processed are all less than zero, a negative luminance difference target image obtained by adjusting the exposure parameter according to the maximum luminance difference value among the luminance difference values is determined as the target image.

[0107] If the exposure consistency type is inconsistent, the maximum positive luminance difference value and the maximum negative luminance difference value are determined according to the luminance difference values corresponding to each of the image regions to be processed, and a positive luminance difference target image obtained by adjusting the exposure parameter according to the maximum positive luminance difference value and a negative luminance difference target image obtained by adjusting the exposure parameter according to the maximum negative luminance difference value are determined as the target image.

[0108] Optionally, when the target image is the positive luminance difference target image, the recognition result determination module 34 is specifically configured to:

[0109] perform feature extraction and target recognition on the positive luminance difference target image, determine at least one first target recognition result and a first target recognition confidence of each first target recognition result, determine a first reserved confidence from among the first target recognition confidences that is greater than a preset confidence threshold, and determine a set of first target recognition results corresponding to each first reserved confidence as the target recognition result.

[0110] Optionally, when the target image is the negative luminance difference target image, the recognition result determination module 34 is specifically configured to:

[0111] perform feature extraction and target recognition on the negative luminance difference target image, determine at least one second target recognition result and a second target recognition confidence of each second target recognition result, determine a second reserved confidence from among the second target recognition confidences that is greater than a preset confidence threshold, and determine a set of second target recognition results corresponding to each second reserved confidence as the target recognition result.

[0112] Optionally, when the target image is the positive luminance difference target image and the negative luminance difference target image, the recognition result determination module 34 is specifically configured to:

[0113] The positive brightness image range is determined in the positive brightness difference target image according to the brightness difference corresponding to each to-be-processed image region, and the negative brightness image range is determined in the negative brightness difference target image; feature extraction and target recognition are performed on the image in the positive brightness difference target image located in the positive brightness image range, at least one positive brightness target recognition result and a first confidence group corresponding to each positive brightness target recognition result are determined; feature extraction and target recognition are performed on the image in the negative brightness difference target image located in the negative brightness image range, at least one negative brightness target recognition result and a second confidence group corresponding to each negative brightness target recognition result are determined; the first confidence and the second confidence greater than a preset confidence threshold in the first confidence group and the second confidence group are determined as reserved confidences; and a set of the positive brightness target recognition result and the negative brightness target recognition result corresponding to each reserved confidence is determined as a target recognition result.

[0114] The target recognition device provided by the embodiments of the present application can execute the target recognition method provided by any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0115] Embodiment four

[0116] Figure 4 A structural schematic diagram of a target recognition device provided by Embodiment Four of the present application. The target recognition device 40 can be an electronic device, which is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present application described and / or claimed in this document.

[0117] As shown in Figure 4 The target recognition device 40 includes at least one processor 41, and a memory, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., which is in communication connection with the at least one processor 41, wherein the memory stores a computer program executable by the at least one processor. The processor 41 can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 into the random access memory (RAM) 43. In the RAM 43, various programs and data required for the operation of the target recognition device 40 can also be stored. The processor 41, the ROM 42, and the RAM 43 are connected to each other through a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0118] A plurality of components in the target recognition device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a magnetic disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the target recognition device 40 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0119] The processor 41 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 41 performs various methods and processes described above, such as the target recognition method.

[0120] In some embodiments, the target recognition method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 48. In some embodiments, part or all of the computer program can be loaded and / or installed onto the target recognition device 40 via the ROM 42 and / or the communication unit 49. When the computer program is loaded onto the RAM 43 and executed by the processor 41, one or more steps of the target recognition method described above can be performed. Alternatively, in other embodiments, the processor 41 can be configured to perform the target recognition method by any other appropriate means, such as by means of firmware.

[0121] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a programmable logic device (PLD), a computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0122] Computer programs for implementing the methods of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer program, when executed, enables the functions / acts specified in the flowcharts and / or block diagrams to be implemented. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine as a standalone software package and partially on a remote machine or entirely on a remote machine or server.

[0123] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of electrical connections, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0124] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0125] The systems and techniques described herein can be implemented in a computing system that includes a back end component, e.g., as a data server, or that includes a middleware component, e.g., an application server, or that includes a front end component, e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described herein, or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication, e.g., a communication network. Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0126] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. A server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0127] It should be understood that the various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the scope of the present disclosure. For example, the steps recited in the present disclosure can be performed in parallel, in series, or in a different order, and the present disclosure is not limited in this regard.

[0128] The specific embodiments described above are not intended to limit the scope of the present disclosure. Those skilled in the art will understand that various modifications, combinations, sub-combinations, and alternatives can be made to the specific embodiments without departing from the spirit and principles of the present disclosure. Any further modifications, equivalents, and / or alternatives come within the scope of the present disclosure as recited by the claims.

Claims

1. A target recognition method, characterized in that, include: Feature extraction and target recognition are performed on the acquired initial image to be adjusted to determine at least one target image region; For each target image region, determine the corresponding image dynamic range and target recognition confidence level; If the dynamic range of the image is less than a preset dynamic range threshold and the target recognition confidence is less than a preset confidence threshold, then the target image region is determined as the image region to be processed. Determine the brightness difference between each of the image regions to be processed and the initial image to be adjusted; If all the brightness differences are greater than zero or less than zero, then the exposure consistency type of each of the image regions to be processed is determined to be consistent. Otherwise, the exposure consistency type of each of the image regions to be processed is determined to be inconsistent; The exposure parameters are adjusted according to the exposure consistency type and each of the image regions to be processed, and the target image under the adjusted exposure parameters is obtained. Feature extraction and target recognition are performed on the target image to determine the target recognition result.

2. The method according to claim 1, characterized in that, The step of adjusting the exposure parameters according to the exposure consistency type and each of the image regions to be processed, and obtaining the target image under the adjusted exposure parameters, includes: If the exposure consistency type is consistent and the brightness difference corresponding to each of the image regions to be processed is greater than zero, then the positive brightness difference target image obtained after adjusting the exposure parameters according to the maximum brightness difference among the brightness differences will be determined as the target image. If the exposure consistency type is consistent and the brightness difference corresponding to each of the image regions to be processed is less than zero, then the negative brightness difference target image obtained after adjusting the exposure parameters according to the maximum brightness difference among the brightness differences will be determined as the target image. If the exposure consistency type is inconsistent, then the maximum positive brightness difference and the maximum negative brightness difference are determined according to the brightness difference corresponding to each of the image regions to be processed, and the target image with positive brightness difference obtained after adjusting the exposure parameters according to the maximum positive brightness difference and the target image with negative brightness difference obtained after adjusting the exposure parameters according to the maximum negative brightness difference are determined as the target image.

3. The method according to claim 2, characterized in that, When the target image is the positive brightness difference target image, the step of performing feature extraction and target recognition on the target image to determine the target recognition result includes: Feature extraction and target recognition are performed on the positive brightness difference target image to determine at least one first target recognition result and a first target recognition confidence level for each first target recognition result; The first target identification confidence scores that are greater than a preset confidence threshold are determined as the first retained confidence scores; The set of first target recognition results corresponding to each of the first retained confidence levels is determined as the target recognition result.

4. The method according to claim 2, characterized in that, When the target image is the negative brightness difference target image, the step of performing feature extraction and target recognition on the target image to determine the target recognition result includes: Feature extraction and target recognition are performed on the negative brightness difference target image to determine at least one second target recognition result and a second target recognition confidence level for each second target recognition result; The second target identification confidence scores that are greater than a preset confidence threshold are determined as the second retained confidence scores. The set of second target identification results corresponding to each second retained confidence level is determined as the target identification result.

5. The method according to claim 2, characterized in that, When the target image is the positive brightness difference target image and the negative brightness difference target image, the step of performing feature extraction and target recognition on the target image to determine the target recognition result includes: Based on the brightness difference corresponding to each of the image regions to be processed, a positive brightness image range is determined in the positive brightness difference target image, and a negative brightness image range is determined in the negative brightness difference target image; Feature extraction and target recognition are performed on the images located within the range of the positive brightness image in the positive brightness difference target image to determine at least one positive brightness target recognition result and a first confidence group corresponding to each positive brightness target recognition result; Feature extraction and target recognition are performed on the images located within the negative brightness image range in the negative brightness difference target image to determine at least one negative brightness target recognition result and a second confidence group corresponding to each negative brightness target recognition result; The first confidence level and the second confidence level in the first confidence level group and the second confidence level group that are greater than the preset confidence level threshold are determined as the retained confidence levels; The set of positive brightness target recognition results and negative brightness target recognition results corresponding to each of the aforementioned confidence levels is determined as the target recognition result.

6. A target recognition device, characterized in that, include: The target region determination module is used to extract features and identify targets from the acquired initial image to be adjusted, and to determine at least one target image region. The consistency determination module is used to determine the target image regions in each target image region that do not meet the preset image processing conditions as image regions to be processed, and to determine the exposure consistency type of each image region to be processed. The target image acquisition module is used to adjust the exposure parameters according to the exposure consistency type and each of the image regions to be processed, and to acquire the target image under the adjusted exposure parameters; The recognition result determination module is used to perform feature extraction and target recognition on the target image and determine the target recognition result; The consistency determination module includes: The image parameter determination unit is used to determine the corresponding image dynamic range and target recognition confidence for each target image region; The region to be processed determination unit is used to determine the target image region as the image region to be processed if the dynamic range of the image is less than a preset dynamic range threshold and the target recognition confidence is less than a preset confidence threshold. A brightness difference determination unit is used to determine the brightness difference between each of the image regions to be processed and the initial image to be adjusted; The consistency determination unit is used to determine that the exposure consistency type of each of the image regions to be processed is consistent if all the brightness differences are greater than zero or less than zero; otherwise, it determines that the exposure consistency type of each of the image regions to be processed is inconsistent.

7. A target recognition device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the target recognition method according to any one of claims 1-5.

8. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the target recognition method as described in any one of claims 1-5.

Images