Method and apparatus for performing a recognition operation on a target in a region

Abstract

This invention relates to the field of target recognition technology, specifically disclosing a method and apparatus for performing target recognition operations in a region. By using the acquisition time of data collected by one type of data collector and the acquisition time of data collected by another type of data collector, the time error of the data collected by the other data collector is corrected. This can reduce the timestamp error of different types of data sources, accurately obtain the acquisition data of different data collectors at the same acquisition time, thereby improving the information fusion effect of different types of data sources and improving the accuracy of target recognition.

Description

Technical Field

[0001] This invention relates to the field of target recognition technology, and in particular to a method and apparatus for performing target recognition operations in a region. Background Technology

[0002] In real life, various needs, such as safe driving and traffic optimization, often require the identification of moving targets in order to track them. To achieve target identification, current methods often involve fusing data collected by different types of data acquisition devices. The target is then identified based on the fusion result, thus enabling target tracking. For example, data from image acquisition devices and radar can be fused to identify vehicles and track them.

[0003] However, in practice, it has been found that due to unstable frame rates of different types of data collectors and network instability during transmission, the data acquired by various types of data collectors can easily have large timestamp errors, thereby reducing the data fusion effect and consequently reducing the accuracy of target recognition.

[0004] Therefore, there is an urgent need to propose a technical solution to reduce timestamp errors from different types of data sources, thereby improving the information fusion effect of different types of data sources and enhancing the accuracy of target identification. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method and apparatus for performing target identification operations in a region, which can reduce the timestamp error of different types of data sources, thereby improving the information fusion effect of different types of data sources and improving the accuracy of target identification.

[0006] To address the aforementioned technical problems, the first aspect of the present invention discloses a method for performing a target identification operation in a region, the method comprising:

[0007] A baseline correction object matching the current scene of a first region is determined. The first region is provided with the baseline correction object and the object to be corrected. The baseline correction object and the object to be corrected are data acquisition devices of different types. The baseline correction object is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the baseline correction object and the acquisition time corresponding to the acquisition data. The object to be corrected is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the object to be corrected and the acquisition time corresponding to the acquisition data.

[0008] Using the acquisition time corresponding to the benchmark correction object as a benchmark, and based on the acquisition time corresponding to the object being corrected, acquisition data matching the acquisition time corresponding to the benchmark correction object is obtained from the acquisition data corresponding to the object being corrected; wherein, the acquisition data matching the acquisition time corresponding to the benchmark correction object includes acquisition data whose acquisition time is later than the acquisition time corresponding to the benchmark correction object and acquisition data whose acquisition time is earlier than the acquisition time corresponding to the benchmark correction object.

[0009] Based on the acquisition time corresponding to the benchmark correction object, the acquisition data that matches the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, a correction operation is performed on the acquisition data corresponding to the object to be corrected to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object.

[0010] The corrected data collected for the object being corrected is used as the basis for performing identification operations on the target in the first region.

[0011] As an optional implementation, in the first aspect of the present invention, the acquisition data corresponding to the reference correction object consists of at least one first sub-acquisition data, and the acquisition time corresponding to the reference correction object consists of the first sub-acquisition times corresponding to all the first sub-acquisition data.

[0012] The collected data corresponding to the object being corrected consists of at least one second sub-collected data, and the collection time corresponding to the object being corrected consists of the second sub-collected times corresponding to all the second sub-collected data.

[0013] As an optional implementation, in the first aspect of the present invention, the step of performing a correction operation on the acquisition data corresponding to the object to be corrected based on the acquisition time corresponding to the reference correction object, the acquisition data matching the acquisition time corresponding to the reference correction object, and the acquisition time corresponding to the acquisition data, to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the reference correction object, includes:

[0014] For any first sub-acquisition time corresponding to the benchmark correction object, based on the first sub-acquisition time, from all second sub-acquisition times, select the first target sub-acquisition time whose time is later than the first sub-acquisition time and the second target sub-acquisition time whose time is earlier than the first sub-acquisition time;

[0015] Based on the first sub-acquisition time, the first target sub-acquisition time, and the second target sub-acquisition time, determine the time correction parameter corresponding to the object to be corrected;

[0016] Based on the time correction parameter corresponding to the object being corrected, the second sub-collection data corresponding to the first target sub-collection time, and the second sub-collection data corresponding to the second target sub-collection time, the collection data matching the first sub-collection time is determined;

[0017] The corrected acquisition data corresponding to the object being corrected includes all acquisition data that matches the first sub-acquisition time.

[0018] As an optional implementation, in the first aspect of the present invention, the method further includes:

[0019] From the corrected acquisition data corresponding to the object being corrected and the acquisition data corresponding to the benchmark corrected object, the first target acquisition data that needs to be converted is selected;

[0020] Perform a parameter transformation operation on the first target acquisition data that matches the first target acquisition data to obtain the parameter-transformed second target acquisition data;

[0021] Based on the first target acquisition data and the third target acquisition data after parameter transformation, a fusion matching operation is performed on the targets in the first region to obtain the target matching result;

[0022] Based on the target matching result, perform feature assignment operations on the targets within the first region;

[0023] The third target acquisition data is the other acquisition data excluding the first target acquisition data from the corrected acquisition data corresponding to the corrected object and the acquisition data corresponding to the benchmark corrected object.

[0024] As an optional implementation, in a first aspect of the present invention, the target matching result includes a first target matching result and / or a second target matching result, wherein:

[0025] The first target matching result is used to represent all first targets that are successfully matched based on the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object;

[0026] The second target matching result is used to represent all second targets that did not successfully match the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object.

[0027] As an optional implementation, in a first aspect of the present invention, the step of performing a feature assignment operation on the target within the first region based on the target matching result includes:

[0028] When the target matching result includes the first target matching result, for any first target, the features of the first target identified by the benchmark correction object and the features of the first target identified by the corrected object are both added to the first target;

[0029] When the target matching result includes the second target matching result, a reference identification parameter matching the current scene of the first region is determined. The reference identification parameter includes a reference distance, which is a distance determined based on the location of the reference correction object or the location of the object being corrected.

[0030] Based on the reference identification parameters, a second region is determined, wherein the range of the second region is smaller than the range of the first region;

[0031] For any second target within the second region, determine the type of the collector that successfully identified the second target. When the type of the collector includes a type that matches the second region, determine the target collector that matches the type that matches the second region, and add the features identified by the target collector to the second target.

[0032] Wherein, when the area of ​​the second region is less than or equal to the area formed by the distance between the reference distance and the reference distance, the target collector is the reference correction object; when the area of ​​the second region is greater than the area formed by the distance between the reference distance and the reference distance, the target collector is the object to be corrected.

[0033] As an optional implementation, in a first aspect of the present invention, performing a parameter transformation operation on the first target acquisition data that matches the first target acquisition data to obtain the parameter-transformed second target acquisition data includes:

[0034] When the first target acquisition data is the corrected acquisition data corresponding to the corrected object, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters matched with the benchmark corrected object to obtain the parameter-converted second target acquisition data;

[0035] When the first target acquisition data is the acquisition data corresponding to the benchmark correction object, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters matched with the correction object to obtain the parameter-converted second target acquisition data.

[0036] A second aspect of the present invention discloses an apparatus for performing a target identification operation in a region, the apparatus comprising:

[0037] The determination module is used to determine a reference correction object that matches the current scene of the first region. The first region is provided with the reference correction object and the object to be corrected. The reference correction object and the object to be corrected are data acquisition devices of different types. The reference correction object is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the reference correction object and the acquisition time corresponding to the acquisition data. The object to be corrected is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the object to be corrected and the acquisition time corresponding to the acquisition data.

[0038] The acquisition module is used to acquire, based on the acquisition time corresponding to the benchmark correction object, the acquisition data that matches the acquisition time corresponding to the acquisition time corresponding to the object being corrected from the acquisition data corresponding to the object being corrected; wherein, the acquisition data that matches the acquisition time corresponding to the benchmark correction object includes acquisition data whose acquisition time is later than the acquisition time corresponding to the benchmark correction object and acquisition data whose acquisition time is earlier than the acquisition time corresponding to the benchmark correction object.

[0039] The correction module is used to perform a correction operation on the acquisition data corresponding to the object to be corrected based on the acquisition time corresponding to the benchmark correction object, the acquisition data that matches the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, so as to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object.

[0040] The corrected data collected for the object being corrected is used as the basis for performing identification operations on the target in the first region.

[0041] As an optional implementation, in the second aspect of the present invention, the acquisition data corresponding to the reference correction object consists of at least one first sub-acquisition data, and the acquisition time corresponding to the reference correction object consists of the first sub-acquisition times corresponding to all the first sub-acquisition data.

[0042] The collected data corresponding to the object being corrected consists of at least one second sub-collected data, and the collection time corresponding to the object being corrected consists of the second sub-collected times corresponding to all the second sub-collected data.

[0043] As an optional implementation, in a second aspect of the present invention, the correction module performs a correction operation on the acquisition data corresponding to the object to be corrected based on the acquisition time corresponding to the reference correction object, the acquisition data matching the acquisition time corresponding to the reference correction object, and the acquisition time corresponding to the acquisition data, to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the reference correction object. The specific method for obtaining this corrected acquisition data includes:

[0044] For any first sub-acquisition time corresponding to the benchmark correction object, based on the first sub-acquisition time, from all second sub-acquisition times, select the first target sub-acquisition time whose time is later than the first sub-acquisition time and the second target sub-acquisition time whose time is earlier than the first sub-acquisition time;

[0045] Based on the first sub-acquisition time, the first target sub-acquisition time, and the second target sub-acquisition time, determine the time correction parameter corresponding to the object to be corrected;

[0046] Based on the time correction parameter corresponding to the object being corrected, the second sub-collection data corresponding to the first target sub-collection time, and the second sub-collection data corresponding to the second target sub-collection time, the collection data matching the first sub-collection time is determined;

[0047] The corrected acquisition data corresponding to the object being corrected includes all acquisition data that matches the first sub-acquisition time.

[0048] As an optional implementation, in a second aspect of the invention, the apparatus further includes:

[0049] The filtering module is used to filter out the first target data that needs to be converted from the correction data corresponding to the object being corrected and the data corresponding to the benchmark correction object.

[0050] The conversion module is used to perform a parameter conversion operation on the first target acquisition data that matches the first target acquisition data, so as to obtain the second target acquisition data after parameter conversion;

[0051] The fusion matching module is used to perform a fusion matching operation on the targets in the first region based on the parameter-transformed first target acquisition data and third target acquisition data to obtain target matching results.

[0052] The feature addition module is used to perform feature assignment operations on targets within the first region based on the target matching result;

[0053] The third target acquisition data is the other acquisition data excluding the first target acquisition data from the corrected acquisition data corresponding to the corrected object and the acquisition data corresponding to the benchmark corrected object.

[0054] As an optional implementation, in a second aspect of the invention, the target matching result includes a first target matching result and / or a second target matching result, wherein:

[0055] The first target matching result is used to represent all first targets that are successfully matched based on the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object;

[0056] The second target matching result is used to represent all second targets that did not successfully match the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object.

[0057] As an optional implementation, in a second aspect of the present invention, the specific manner in which the feature adding module performs feature assignment operation on the target within the first region based on the target matching result includes:

[0058] When the target matching result includes the first target matching result, for any first target, the features of the first target identified by the benchmark correction object and the features of the first target identified by the corrected object are both added to the first target;

[0059] When the target matching result includes the second target matching result, a reference identification parameter matching the current scene of the first region is determined. The reference identification parameter includes a reference distance, which is a distance determined based on the location of the reference correction object or the location of the object being corrected.

[0060] Based on the reference identification parameters, a second region is determined, wherein the range of the second region is smaller than the range of the first region;

[0061] For any second target within the second region, determine the type of the collector that successfully identified the second target. When the type of the collector includes a type that matches the second region, determine the target collector that matches the type that matches the second region, and add the features identified by the target collector to the second target.

[0062] Wherein, when the area of ​​the second region is less than or equal to the area formed by the distance between the reference distance and the reference distance, the target collector is the reference correction object; when the area of ​​the second region is greater than the area formed by the distance between the reference distance and the reference distance, the target collector is the object to be corrected.

[0063] As an optional implementation, in a second aspect of the present invention, the conversion module performs a parameter conversion operation on the first target acquisition data that matches the first target acquisition data to obtain the parameter-converted second target acquisition data in the following specific ways:

[0064] When the first target acquisition data is the corrected acquisition data corresponding to the corrected object, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters matched with the benchmark corrected object to obtain the parameter-converted second target acquisition data;

[0065] When the first target acquisition data is the acquisition data corresponding to the benchmark correction object, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters matched with the correction object to obtain the parameter-converted second target acquisition data.

[0066] A third aspect of the present invention discloses another apparatus for performing a target identification operation in a region, the apparatus comprising:

[0067] Memory containing executable program code;

[0068] A processor coupled to the memory;

[0069] The processor calls the executable program code stored in the memory to execute the method for performing target identification operations in a region disclosed in the first aspect of the present invention.

[0070] The fourth aspect of the present invention discloses a computer-readable storage medium storing computer instructions, which, when invoked, are used to execute the method for performing a target identification operation in a region disclosed in the first aspect of the present invention.

[0071] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0072] In this embodiment of the invention, a reference correction object matching the current scene of a first region is determined. The first region is provided with a reference correction object and a correction target, which are data acquisition devices of different types. The reference correction object is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the reference correction object and the acquisition time corresponding to the acquisition data. The correction target is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the correction target and the acquisition time corresponding to the acquisition data. Using the acquisition time corresponding to the reference correction object as a benchmark, and based on the acquisition time corresponding to the correction target, the data corresponding to the reference correction object is obtained from the acquisition data corresponding to the correction target. The data collected at the time of acquisition is matched with the acquisition time of the reference correction object. The data collected at the time of acquisition of the reference correction object includes data collected at a time later than the acquisition time of the reference correction object and data collected at a time earlier than the acquisition time of the reference correction object. Based on the acquisition time of the reference correction object, the data collected at the time of acquisition of the reference correction object, and the acquisition time corresponding to that data, a correction operation is performed on the acquisition data corresponding to the object to be corrected, resulting in corrected acquisition data of the object to be corrected at the acquisition time of the reference correction object. The corrected acquisition data corresponding to the object to be corrected is used as the basis for performing a target identification operation in the first region. Therefore, this invention corrects the time error of data collected by one type of data collector and the acquisition data and acquisition time of another type of data collector by using the acquisition time of the acquisition data collected by one type of data collector and the acquisition data and acquisition time of another data collector. This reduces timestamp errors between different types of data sources, accurately obtains acquisition data from different data collectors at the same acquisition time, thereby improving the information fusion effect of different types of data sources and improving the accuracy of target identification. Attached Figure Description

[0073] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0074] Figure 1 This is a flowchart illustrating a method for performing target identification operations in a region, as disclosed in an embodiment of the present invention.

[0075] Figure 2 This is a flowchart illustrating another method for performing vehicle identification operations in a region, as disclosed in an embodiment of the present invention.

[0076] Figure 3This is a schematic diagram illustrating the fusion and matching of radar data and image data as disclosed in an embodiment of the present invention;

[0077] Figure 4 This is a schematic diagram of a scenario for fusing and matching radar data and image data, as disclosed in an embodiment of the present invention.

[0078] Figure 5 This is a schematic diagram of the structure of a device for performing target identification operations in a region, as disclosed in an embodiment of the present invention;

[0079] Figure 6 This is a schematic diagram of another device for performing target identification operations in a region, as disclosed in an embodiment of the present invention;

[0080] Figure 7 This is a schematic diagram of the structure of another device for performing target identification operations in a region, as disclosed in an embodiment of the present invention. Detailed Implementation

[0081] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0082] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0083] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0084] This invention discloses a method and apparatus for performing target identification operations in a region. By using the acquisition time of data collected by one type of data collector and the acquisition time of data collected by another type of data collector, the time error of the data collected by the second data collector is corrected. This reduces timestamp errors between different types of data sources, accurately obtains data collected by different data collectors at the same acquisition time, thereby improving the information fusion effect of different types of data sources and enhancing the accuracy of target identification. Detailed descriptions follow.

[0085] Example 1

[0086] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for performing target identification operations in a region, as disclosed in an embodiment of the present invention. Figure 1 The described method can be applied to any scenario requiring the identification of moving targets, such as vehicle monitoring. Figure 1 As shown, the method may include the following operations:

[0087] 101. Determine the baseline correction object that matches the current scene in the first region.

[0088] In this embodiment of the invention, optionally, a first region is provided with a reference correction object and a correction object. The reference correction object is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the reference correction object and the acquisition time corresponding to the acquisition data. The correction object is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the correction object and the acquisition time corresponding to the acquisition data. The reference correction object and the correction object are different types of data acquisition devices. For example, when the reference correction object is an image acquisition device, the correction object is a radar acquisition device; when the reference correction object is a radar acquisition device, the correction object is an image acquisition device. Optionally, the image acquisition device can be a monocular image acquisition device or a multi-view image acquisition device. Further optionally, the acquisition data will be different depending on the different types of current scenarios. For example, the acquisition data acquired by the radar acquisition device and the image acquisition device can be location data.

[0089] In this embodiment of the invention, optionally, the data collected corresponding to the reference correction object consists of at least one first sub-collected data, and the collection time corresponding to the reference correction object consists of the first sub-collected times corresponding to all the first sub-collected data; the data collected corresponding to the object being corrected consists of at least one second sub-collected data, and the collection time corresponding to the object being corrected consists of the second sub-collected times corresponding to all the second sub-collected data.

[0090] In this embodiment of the invention, optionally, based on the performance data of each data collector within a preset time period (within three days), the performance data of each data collector is analyzed to obtain the target performance of each data collector within the preset time period. The target performance of the two data collectors is compared, and the one with better target performance is selected as the benchmark correction object. The performance data includes one or more of the following: network status data, data acquisition stability data, and data acquisition noise data. For example, the more stable the network and / or the more stable the data acquisition and / or the lower the noise, the better the performance. By comparing and analyzing the performance of each data collector, the accuracy of determining the benchmark correction object can be improved, thereby improving the accuracy of time difference correction of the corresponding collected data, obtaining accurate and reliable correction data, and further improving the accuracy of data fusion and matching.

[0091] Optionally, the matching accuracy of each data collector during a preset time period, when used as a benchmark correction object, is obtained, and the mean matching accuracy of all matching accuracies within the preset time period is calculated. Based on each matching accuracy and the mean matching accuracy, the standard deviation matching accuracy corresponding to each data collector is calculated, and the standard deviation matching accuracy of the two is compared. The target performance corresponding to each data collector is quantified to obtain quantifiable target performance. Based on the current scene object occlusion situation in the first area and the type (e.g., vehicle, person) and / or movement (e.g., direction, speed) of the identified target, corresponding weights are assigned to the quantified target performance and the standard deviation matching accuracy. Based on the quantified target performance and the corresponding weights, the corresponding correction effect value is calculated. The correction value corresponding to each data collector is compared, and the one with the larger correction effect value is used as the benchmark correction object. The larger the correction effect value, the better the correction effect is obtained when the other data is corrected based on it as the benchmark correction object. By further combining the matching accuracy and performance of each data acquisition device as a benchmark correction object, and comprehensively analyzing the benchmark correction object to be used, the accuracy of the benchmark correction object determination can be further improved.

[0092] 102. Using the acquisition time corresponding to the benchmark correction object as the benchmark, based on the acquisition time corresponding to the object being corrected, acquire acquisition data that matches the acquisition time corresponding to the benchmark correction object from the acquisition data corresponding to the object being corrected; wherein, the acquisition data that matches the acquisition time corresponding to the benchmark correction object includes acquisition data whose acquisition time is later than the acquisition time corresponding to the benchmark correction object and acquisition data whose acquisition time is earlier than the acquisition time corresponding to the benchmark correction object.

[0093] 103. Based on the acquisition time corresponding to the benchmark correction object, the acquisition data matching the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, perform a correction operation on the acquisition data corresponding to the object to be corrected to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object; wherein, the corrected acquisition data corresponding to the object to be corrected is used as the basis for performing the identification operation on the target in the first region.

[0094] As can be seen, by using the acquisition time of the data collected by one type of data collector and the acquisition time of the data collected by another type of data collector to correct the time error of the data collected by the other type of data collector, the timestamp error of different data sources can be reduced, and the data collected by different data collectors at the same acquisition time can be accurately obtained, thereby improving the information fusion effect of different data sources and improving the accuracy of target recognition.

[0095] In this embodiment of the invention, optionally, based on the acquisition time corresponding to the benchmark correction object, the acquisition data matching the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, a correction operation is performed on the acquisition data corresponding to the object to be corrected to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object, including:

[0096] For any first sub-acquisition time corresponding to the benchmark correction object, based on the first sub-acquisition time, select the first target sub-acquisition time that is later than the first sub-acquisition time and the second target sub-acquisition time that is earlier than the first sub-acquisition time from all the second sub-acquisition times;

[0097] Based on the first sub-acquisition time, the first target sub-acquisition time, and the second target sub-acquisition time, determine the time correction parameters corresponding to the object to be corrected;

[0098] Based on the time correction parameter corresponding to the object being corrected, the second sub-collection data corresponding to the first target sub-collection time, and the second sub-collection data corresponding to the second target sub-collection time, determine the collection data that matches the first sub-collection time;

[0099] In this optional embodiment, the corrected acquisition data corresponding to the object to be corrected includes all acquisition data matching the first sub-acquisition time, wherein the acquisition data matching each first sub-acquisition time is the corrected acquisition data of the object to be corrected after time difference correction at that first sub-acquisition time. Optionally, the number of first target sub-acquisition times or second target sub-acquisition times can be one or more.

[0100] As can be seen, this embodiment can improve the efficiency and accuracy of time difference correction by filtering the sub-collection times before and after the sub-collection time corresponding to the benchmark correction object from the sub-collection times of the object being corrected, and performing time difference correction on the collected data of the object being corrected based on the sub-collection times before and after the sub-collection times and the collected data corresponding to the sub-collection times and the sub-collection time corresponding to the benchmark correction object. This is conducive to further improving the accuracy of acquiring the collected data of the object being corrected.

[0101] In an optional embodiment, the method may further include the following steps:

[0102] From the correction data collected for the object being corrected and the data collected for the baseline correction object, select the first target data that needs to be converted into parameters.

[0103] Perform a parameter transformation operation on the first target acquisition data to obtain the parameter-transformed second target acquisition data;

[0104] Based on the parameter-transformed first target acquisition data and third target acquisition data, a fusion matching operation is performed on the targets in the first region to obtain the target matching results;

[0105] Based on the target matching results, perform feature assignment operations on the targets within the first region;

[0106] Among them, the third target acquisition data is the other acquisition data excluding the first target acquisition data from the correction acquisition data corresponding to the corrected object and the acquisition data corresponding to the benchmark correction object.

[0107] In this optional embodiment, the target matching result may include a first target matching result and / or a second target matching result, wherein: the first target matching result is used to represent all first targets that are successfully matched based on the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object; the second target matching result is used to represent all second targets that are not successfully matched based on the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object. Specifically, for any first target, when the corrected object is identified and the benchmark correction object is also identified, this target is defined as a first target; for any second target, when the benchmark correction object is identified but the corrected object is not identified, or when the benchmark correction object is not identified but the corrected object is identified, this target is defined as a second target.

[0108] In this optional embodiment, optionally, a parameter transformation operation matching the first target acquisition data is performed on the first target acquisition data to obtain the parameter-transformed second target acquisition data, including:

[0109] When the first target acquisition data is the corrected acquisition data corresponding to the object being corrected, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters that match the benchmark corrected object to obtain the parameter-converted second target acquisition data;

[0110] When the first target acquisition data is the acquisition data corresponding to the benchmark correction object, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters that match the correction object to obtain the parameter-converted second target acquisition data.

[0111] In this optional embodiment, the parameter transformation operations will differ depending on the type of current scene, the type of data collected by different data acquisition devices, or the data collected by different data acquisition devices. For example, for location data acquired by an image acquisition device, the corresponding transformation parameter can be a coordinate system transformation, and the parameter transformation operation is a location coordinate transformation operation.

[0112] In this optional embodiment, different data fusion and matching methods can be used to fuse and match the collected data from different data collectors, depending on the data type. For example, for location data, the Hungarian matching algorithm can be used for fusion and matching. Further optionally, different types of data collectors or different targets require different features to be added. For example, for image collectors, visual features such as color and shape need to be added, while for radar collectors, non-visual features such as distance and speed need to be added.

[0113] In this optional embodiment, based on the first target acquisition data and the third target acquisition data after parameter transformation, a fusion matching operation is performed on the target in the first region to obtain the target matching result. Specifically, for any sub-acquisition time, the acquisition data and the corrected acquisition data corresponding to the same sub-acquisition time are fused and matched to obtain the matching result of the same sub-acquisition time. The target matching result is composed of multiple matching results of the same sub-acquisition time.

[0114] As can be seen, this optional embodiment transforms the collected data of one data collector by using the transformation parameters corresponding to one data collector, and then fuses and matches it with the collected data of one data collector that does not require transformation. This can improve the accuracy of data fusion and matching, thereby improving the accuracy of target recognition. Then, based on different matching results, corresponding features are added to the target in the region to achieve accurate completion of missing target data.

[0115] In this optional embodiment, optionally, based on the target matching result, a feature assignment operation is performed on the target within the first region, including:

[0116] When the target matching result includes the first target matching result, for any first target, the features of the first target identified by the benchmark correction object and the features of the first target identified by the corrected object are both added to the first target;

[0117] When the target matching result includes the second target matching result, determine the reference identification parameters that match the current scene of the first region. The reference identification parameters include the reference distance, which is the distance determined based on the location of the reference correction object or the location of the object being corrected, such as 150 meters.

[0118] Based on the baseline identification parameters, a second region is determined, and the range of the second region is smaller than that of the first region.

[0119] For any second target within the second region, determine the type of the collector that successfully identified the second target. When the type of the collector includes a type that matches the second region, determine the target collector that matches the type that matches the second region, and add the features identified by the target collector to the second target.

[0120] Specifically, when the area of ​​the second region is the area formed by the distance less than or equal to the reference distance and the reference distance, the target collector is the reference correction object; when the area of ​​the second region is the area formed by the distance greater than the reference distance and the reference distance, the target collector is the object to be corrected.

[0121] As can be seen, this optional embodiment can accurately filter out the corresponding features through different target matching results, thereby achieving accurate completion of missing target data and improving completion efficiency.

[0122] Example 2

[0123] Please see Figure 2 , Figure 2 This is a flowchart illustrating another method for performing vehicle identification operations in a region, as disclosed in an embodiment of the present invention. Figure 2 This is an example embodiment using the current scenario as a vehicle monitoring scenario, with the reference correction object as the image acquisition device, the object to be corrected as the radar acquisition device, the acquisition data corresponding to the reference correction object as image data, and the acquisition data corresponding to the object to be corrected as radar data. Figure 2 As shown, the method may include the following operations:

[0124] 201. Based on the acquisition time corresponding to the image acquisition device in the first region, according to the acquisition time corresponding to the radar acquisition device, obtain the radar data that matches the acquisition time corresponding to the image acquisition device from the radar data corresponding to the radar acquisition device; wherein, the radar data that matches the acquisition time corresponding to the image acquisition device includes radar data whose acquisition time is later than the acquisition time corresponding to the image acquisition device and radar data whose acquisition time is earlier than the acquisition time corresponding to the image acquisition device.

[0125] In this embodiment of the invention, optionally, an image acquisition device and a radar acquisition device are provided in the first area. The image acquisition device is used to perform data acquisition operations on the first area to obtain image data corresponding to the image acquisition device and the acquisition time corresponding to the image data. The radar acquisition device is used to perform data acquisition operations on the first area to obtain radar data corresponding to the radar acquisition device and the acquisition time corresponding to the radar data. Optionally, the image acquisition device can be a monocular image acquisition device or a multi-view image acquisition device.

[0126] In this embodiment of the invention, optionally, the image data corresponding to the image acquisition device consists of at least one sub-image data, and the acquisition time corresponding to the image acquisition device consists of a first sub-acquisition time corresponding to all the sub-image data; the radar data corresponding to the radar acquisition device consists of at least one sub-radar data, and the acquisition time corresponding to the radar acquisition device consists of a second sub-acquisition time corresponding to all the sub-radar data. Based on the acquisition time corresponding to the image acquisition device in the first region, and according to the acquisition time corresponding to the radar acquisition device, radar data matching the acquisition time corresponding to the image acquisition device is obtained from the radar data corresponding to the radar acquisition device. Specifically, for any first sub-acquisition time corresponding to the image acquisition device, based on that first sub-acquisition time, and according to all the second sub-acquisition times corresponding to the radar acquisition device, sub-radar data matching the first sub-acquisition time is obtained from the radar data corresponding to the radar acquisition device.

[0127] 202. Based on the acquisition time corresponding to the image acquisition device, the radar data matching the acquisition time corresponding to the image acquisition device, and the acquisition time corresponding to the radar data, a correction operation is performed on the radar data corresponding to the radar acquisition device to obtain the corrected radar data of the radar acquisition device at the acquisition time corresponding to the image acquisition device; wherein, the corrected radar data corresponding to the radar acquisition device is used as the basis for performing vehicle identification operation in the first area.

[0128] As can be seen, by implementing the present invention, the time error of the radar data collected by the radar collector is corrected by the acquisition time corresponding to the image collector and the radar data and acquisition time collected by the radar collector. This can reduce the timestamp error of different types of data sources, accurately obtain the data collected by different data collectors at the same acquisition time, thereby improving the information fusion effect of different types of data sources and improving the accuracy of vehicle recognition.

[0129] In this embodiment of the invention, optionally, based on the acquisition time corresponding to the image acquisition device, the radar data matching the acquisition time corresponding to the image acquisition device, and the acquisition time corresponding to the radar data, a correction operation is performed on the radar data corresponding to the radar acquisition device to obtain the corrected radar data of the radar acquisition device at the acquisition time corresponding to the image acquisition device, including:

[0130] For any first sub-acquisition time corresponding to the image acquisition device, based on the first sub-acquisition time, select the first target sub-acquisition time whose time is later than the first sub-acquisition time and the second target sub-acquisition time whose time is earlier than the first sub-acquisition time from all the second sub-acquisition times;

[0131] Based on the first sub-acquisition time, the first target sub-acquisition time, and the second target sub-acquisition time, determine the time correction parameters corresponding to the radar acquisition device;

[0132] Based on the time correction parameters corresponding to the radar acquisition device, the sub-radar data corresponding to the first target sub-acquisition time, and the sub-radar data corresponding to the second target sub-acquisition time, determine the sub-radar data that matches the first sub-acquisition time.

[0133] In this optional embodiment, the corrected radar data corresponding to the radar collector includes all sub-radar data that match the first sub-collection time, wherein the sub-radar data that matches each first sub-collection time is the corrected radar data of the radar collector at that first sub-collection time after time difference correction.

[0134] In embodiments of the present invention, such as Figure 3 As shown, Figure 3 This is a schematic diagram illustrating the fusion and matching of radar data and image data disclosed in an embodiment of the present invention. Figure 3The diagram shown illustrates the process before radar data correction. The horizontal axis represents the time acquisition moment t. Before data fusion and matching, the acquisition moment of the image acquisition device is used as the reference. That is, at the acquisition moments when image data (video frame data) is acquired at Ki-2, Ki-1, Ki, etc., the radar data with the nearest neighbor timestamp before that acquisition moment is synchronously selected for data fusion and matching. Taking Ki-2 acquisition moment as an example, image data is acquired at this acquisition moment, and radar data acquired by the radar acquisition device at Kj-3 acquisition moment is simultaneously acquired and fused with the image data. At Ki-1 acquisition moment, radar data from Kj-1 acquisition moment is acquired and fused. However, at Ki acquisition moment, since no radar data is acquired between Ki-1 and Ki acquisition moments, image data is used solely as the input for fusion and matching to update the features of the fused vehicle. Figure 3 The correction of radar data is explained in detail below. The radar data needs to be matched with the image data acquired at time Ki-2. The radar data actually acquired by the radar collector is the radar data between time Kj-3 and Kj-2. The radar correction data M at time Ki-2 is obtained by linear interpolation using the data acquired by the radar collector at time Kj-3 and time Kj-2: M = a*(1-t) + b*t. Therefore, at this time, 'a' represents the radar data acquired by the radar collector at time Kj-3, 'b' represents the radar data acquired by the radar collector at time Kj-2, '1-t' represents (Kj-2-Ki-2) / (Kj-2-Kj-3), and 't' represents (Ki-2-Kj-3) / (Kj-2-Kj-3). Then, the radar correction data M is fused and matched with the image data acquired by the image collector at time Ki-2.

[0135] As can be seen, this embodiment improves the efficiency and accuracy of time difference correction by filtering the sub-acquisition times before and after the sub-acquisition time corresponding to the image acquisition time from the sub-acquisition times of the radar acquisition device, and performing time difference correction on the acquisition data acquired by the radar acquisition device based on the sub-acquisition times before and after, the acquisition data corresponding to the sub-acquisition time, and the sub-acquisition time corresponding to the image acquisition device. This is beneficial to further improve the accuracy of the acquisition data of the radar acquisition device.

[0136] In an optional embodiment, the method may further include the following steps:

[0137] From the corrected radar data corresponding to the radar acquisition device and the image data corresponding to the image acquisition device, the first target data that needs to be converted into parameters is selected.

[0138] Perform a parameter transformation operation on the first target data that matches the first target data to obtain the parameter-transformed second target data;

[0139] Based on the first and third target data after parameter transformation, a fusion matching operation is performed on the vehicles in the first region to obtain the vehicle matching results;

[0140] Based on the vehicle matching results, perform feature assignment operations on the vehicles in the first region;

[0141] The third target data is the corrected radar data corresponding to the radar collector and the image data corresponding to the image collector, excluding the first target data.

[0142] In this optional embodiment, the vehicle matching result may include a first vehicle matching result and / or a second vehicle matching result, wherein: the first vehicle matching result represents all first vehicles that have successfully matched based on the image data corresponding to the image collector and the corrected radar data corresponding to the radar collector; the second vehicle matching result represents all second vehicles that have not successfully matched based on the image data corresponding to the image collector and the corrected radar data corresponding to the radar collector. Specifically, for any first vehicle, if the radar collector identifies it and it is also identified by the image collector, this target is defined as a first vehicle; for any second vehicle, if the image collector identifies it but the radar collector does not, or if the image collector does not identify it but the radar collector does, this target is defined as a second vehicle.

[0143] In this optional embodiment, optionally, a parameter transformation operation matching the first target data is performed on the first target data to obtain the parameter-transformed second target data, including:

[0144] When the first target data is the corrected radar data corresponding to the radar collector, a parameter conversion operation is performed on the first target data based on the conversion parameters matched with the image collector to obtain the parameter-converted second target data;

[0145] When the first target data is the image data corresponding to the image acquisition device, a parameter conversion operation is performed on the first target data based on the conversion parameters matched with the radar acquisition device to obtain the parameter-converted second target data.

[0146] In this optional embodiment, optionally, for the location data acquired by the image acquisition device, the corresponding transformation parameter can be a coordinate system transformation, and the parameter transformation operation is a location coordinate transformation operation. Further optionally, it is preferred to transform the radar correction data corresponding to the radar acquisition device to the coordinate system corresponding to the image acquisition device. Furthermore, before performing the location coordinate transformation, the coordinate system has been calibrated. The calibration method is based on fitting multiple sets of radar point coordinates (world coordinates) with their corresponding image point coordinates to obtain the corresponding coordinate system, or other methods are not limited. Also optionally, for the location data, the Hungarian matching algorithm can be used for fusion matching. Optionally, Figure 4 This is a schematic diagram illustrating a scenario where radar data and image data are fused and matched, as disclosed in an embodiment of the present invention. Figure 4 As shown, the radar data from the radar collector, after correction and conversion from 3D coordinates to image coordinates, is represented by the red dots in the image. The green rectangles in the image represent the 2D vehicle position information detected from the image. Therefore, it is necessary to match the targets detected by the radar with those detected by the image, i.e., matching the red dot targets with the green rectangle targets. The distance between each red dot and each green rectangle is calculated as follows: if the red dot is inside the green rectangle, the distance is recorded as 0; if the red dot is not inside the green rectangle, the distance from the red dot to the nearest block distance to the green rectangle is calculated (if this block distance is greater than the width or height of the green rectangle, the distance is set to a maximum value, indicating that the two targets are too far apart, and the subsequent fusion matching algorithm will not identify the two targets as the same target). If there are M red dots and N green rectangles, an M*N distance matrix can be generated and input into the Hungarian matching algorithm for matching to obtain the vehicle matching results. For the green rectangle closest to the red dot and the distance is less than or equal to the width or height of the green rectangle, the two are successfully matched and are the same target; otherwise, the match is unsuccessful and they are not the same target.

[0147] In this optional embodiment, based on the parameter-converted first target data and third target data, a fusion matching operation is performed on the vehicles in the first region to obtain the vehicle matching result. Specifically, for any sub-acquisition time, the acquisition data and corrected radar data corresponding to the same sub-acquisition time are fused and matched to obtain the matching result of the same sub-acquisition time. The vehicle matching result consists of multiple matching results of the same sub-acquisition time.

[0148] As can be seen, this optional embodiment can improve the accuracy of location data fusion and matching by converting the radar correction data corresponding to the radar collector into the coordinates corresponding to the image collector, and then fusing and matching the converted location data with the location data collected by the image collector at the same time. This improves the accuracy of vehicle identification. Furthermore, based on different matching results, corresponding features are added to vehicles in the area to accurately fill in the missing vehicle data.

[0149] In this optional embodiment, optionally, based on the vehicle matching result, a feature assignment operation is performed on the vehicles within the first region, including:

[0150] When the vehicle matching result includes the first vehicle matching result, for any first vehicle, the features of the first vehicle identified by the image collector and the features of the first vehicle identified by the radar collector are both added to the first vehicle.

[0151] When the vehicle matching result includes the second vehicle matching result, determine the reference recognition parameters that match the current scene of the first area. The reference recognition parameters include the reference distance, which is the distance determined based on the location of the image collector or the location of the radar collector, such as 150 meters.

[0152] Based on the baseline identification parameters, a second region is determined, and the range of the second region is smaller than that of the first region.

[0153] For any second vehicle within the second area, determine the type of the collector that successfully identified the second vehicle. When the type of the collector includes a type that matches the second area, determine the target collector that matches the type that matches the second area, and add the features identified by the target collector to the second vehicle.

[0154] Specifically, when the area of ​​the second region is the area formed by the distance less than or equal to the reference distance and the reference distance, the target acquisition device is an image acquisition device; when the area of ​​the second region is the area formed by the distance greater than the reference distance and the reference distance, the target acquisition device is a radar acquisition device.

[0155] In this optional embodiment, for any successfully matched first vehicle, visual features (such as vehicle color, vehicle shape, license plate, vehicle type, etc.) and non-visual features (such as speed, distance, acceleration, angle, etc.) are added to the first vehicle. For a second vehicle that does not match successfully, if the image acquisition device detects the second vehicle within a specific distance range (e.g., within 150 meters), but the second vehicle does not match a radar target, the fusion process only retains visual feature information (such as vehicle color, vehicle shape, license plate, vehicle type, etc.) and does not assign any radar features; if the second vehicle is not detected visually, regardless of whether the radar acquires the second vehicle, the fusion process considers the second vehicle not to have been detected. Outside a specific range (e.g., 150 meters), due to issues such as occlusion, the visual effect cannot be guaranteed. For a second vehicle that is not successfully matched, if the radar detects the second vehicle but it is not matched with any visual second vehicle, the fusion process only retains radar feature information (such as speed, distance, acceleration, angle, etc.) and does not assign any visual features. If the second vehicle is not detected, regardless of whether the visual system acquires the second vehicle, the fusion process will consider it as if the second vehicle was not detected.

[0156] As can be seen, this optional embodiment can accurately filter out the corresponding features through different vehicle matching results, thereby achieving accurate completion of the target missing data and improving the completion efficiency.

[0157] It should be noted that for descriptions of other technical content, please refer to the detailed description of other content in Embodiment 1, which will not be repeated here.

[0158] Example 3

[0159] Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of a device for performing target identification operations in a region, as disclosed in an embodiment of the present invention. Figure 5 The described device can be applied to any scenario requiring target recognition, such as vehicle monitoring. Figure 5 As shown, the device may include:

[0160] The determining module 301 is used to determine a reference correction object that matches the current scene of the first region. The first region is provided with a reference correction object and a correction target. The reference correction object is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the reference correction object and the acquisition time corresponding to that data. The correction target is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the correction target and the acquisition time corresponding to that data. The reference correction object and the correction target are different types of data acquisition devices. For example, when the reference correction object is an image acquisition device, the correction target is a radar acquisition device; when the reference correction object is a radar acquisition device, the correction target is an image acquisition device. Optionally, the image acquisition device can be a monocular image acquisition device or a multi-view image acquisition device. Further optionally, the acquisition data may differ depending on the different types of current scenes. For example, the acquisition data acquired by the radar acquisition device and the image acquisition device may be location data.

[0161] The acquisition module 302 is used to acquire, based on the acquisition time corresponding to the benchmark correction object, the acquisition data that matches the acquisition time corresponding to the benchmark correction object from the acquisition data corresponding to the object being corrected; wherein, the acquisition data that matches the acquisition time corresponding to the benchmark correction object includes acquisition data whose acquisition time is later than the acquisition time corresponding to the benchmark correction object and acquisition data whose acquisition time is earlier than the acquisition time corresponding to the benchmark correction object.

[0162] The correction module 303 is used to perform a correction operation on the acquisition data corresponding to the object to be corrected based on the acquisition time corresponding to the benchmark correction object, the acquisition data matching the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object; wherein, the corrected acquisition data corresponding to the object to be corrected is used as the basis for performing the identification operation on the target in the first region.

[0163] In this embodiment of the invention, optionally, the data collected corresponding to the reference correction object consists of at least one first sub-collected data, and the collection time corresponding to the reference correction object consists of the first sub-collected times corresponding to all the first sub-collected data; the data collected corresponding to the object being corrected consists of at least one second sub-collected data, and the collection time corresponding to the object being corrected consists of the second sub-collected times corresponding to all the second sub-collected data.

[0164] It is evident that implementation Figure 5 The described device corrects the time error of the data collected by one type of data collector by using the acquisition time of the data collected by one type of data collector and the acquisition time of the data collected by another type of data collector. This reduces the timestamp error of different data sources, accurately obtains the data collected by different data collectors at the same acquisition time, thereby improving the information fusion effect of different data sources and improving the accuracy of target recognition.

[0165] In this embodiment of the invention, optionally, the correction module 301 performs a correction operation on the acquisition data corresponding to the object to be corrected based on the acquisition time corresponding to the benchmark correction object, the acquisition data matching the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object. The specific methods for obtaining the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object include:

[0166] For any first sub-acquisition time corresponding to the benchmark correction object, based on the first sub-acquisition time, select the first target sub-acquisition time that is later than the first sub-acquisition time and the second target sub-acquisition time that is earlier than the first sub-acquisition time from all the second sub-acquisition times;

[0167] Based on the first sub-acquisition time, the first target sub-acquisition time, and the second target sub-acquisition time, determine the time correction parameters corresponding to the object to be corrected;

[0168] Based on the time correction parameter corresponding to the object being corrected, the second sub-collection data corresponding to the first target sub-collection time, and the second sub-collection data corresponding to the second target sub-collection time, determine the collection data that matches the first sub-collection time;

[0169] In this optional embodiment, the corrected acquisition data corresponding to the object to be corrected includes all acquisition data matching the first sub-acquisition time, wherein the acquisition data matching each first sub-acquisition time is the corrected acquisition data of the object to be corrected after time difference correction at that first sub-acquisition time. Optionally, the number of first target sub-acquisition times or second target sub-acquisition times can be one or more.

[0170] As can be seen, this embodiment can improve the efficiency and accuracy of time difference correction by filtering the sub-collection times before and after the sub-collection time corresponding to the benchmark correction object from the sub-collection times of the object being corrected, and performing time difference correction on the collected data of the object being corrected based on the sub-collection times before and after the sub-collection times and the collected data corresponding to the sub-collection times and the sub-collection time corresponding to the benchmark correction object. This is conducive to further improving the accuracy of acquiring the collected data of the object being corrected.

[0171] In an optional embodiment, such as Figure 6 As shown, Figure 6 This is a schematic diagram of another device for performing target identification operations in a region, as disclosed in an embodiment of the present invention. Figure 6 As shown, the device also includes:

[0172] The filtering module 304 is used to filter out the first target data that needs to be converted from the correction data corresponding to the object being corrected and the data corresponding to the benchmark correction object.

[0173] The conversion module 305 is used to perform a parameter conversion operation on the first target acquisition data to match the first target acquisition data, so as to obtain the second target acquisition data after parameter conversion;

[0174] The fusion matching module 306 is used to perform a fusion matching operation on the targets in the first region based on the parameter-transformed first target acquisition data and third target acquisition data to obtain the target matching result.

[0175] The feature addition module 307 is used to perform feature assignment operations on targets within the first region based on the target matching results;

[0176] Among them, the third target acquisition data is the other acquisition data excluding the first target acquisition data from the correction acquisition data corresponding to the corrected object and the acquisition data corresponding to the benchmark correction object.

[0177] In this optional embodiment, the target matching result may include a first target matching result and / or a second target matching result, wherein: the first target matching result is used to represent all first targets that are successfully matched based on the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object; the second target matching result is used to represent all second targets that are not successfully matched based on the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object. Specifically, for any first target, when the corrected object is identified and the benchmark correction object is also identified, this target is defined as a first target; for any second target, when the benchmark correction object is identified but the corrected object is not identified, or when the benchmark correction object is not identified but the corrected object is identified, this target is defined as a second target.

[0178] In this optional embodiment, the conversion module 305 may perform a parameter conversion operation on the first target acquisition data to obtain the parameter-converted second target acquisition data in the following ways:

[0179] When the first target acquisition data is the corrected acquisition data corresponding to the object being corrected, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters that match the benchmark corrected object to obtain the parameter-converted second target acquisition data;

[0180] When the first target acquisition data is the acquisition data corresponding to the benchmark correction object, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters that match the correction object to obtain the parameter-converted second target acquisition data.

[0181] In this optional embodiment, the parameter transformation operations will differ depending on the type of current scene, the type of data collected by different data acquisition devices, or the data collected by different data acquisition devices. For example, for location data acquired by an image acquisition device, the corresponding transformation parameter can be a coordinate system transformation, and the parameter transformation operation is a location coordinate transformation operation.

[0182] In this optional embodiment, different data fusion and matching methods can be used to fuse and match the collected data from different data collectors, depending on the data type. For example, for location data, the Hungarian matching algorithm can be used for fusion and matching. Further optionally, different types of data collectors or different targets require different features to be added. For example, for image collectors, visual features such as color and shape need to be added, while for radar collectors, non-visual features such as distance and speed need to be added.

[0183] In this optional embodiment, based on the first target acquisition data and the third target acquisition data after parameter transformation, a fusion matching operation is performed on the target in the first region to obtain the target matching result. Specifically, for any sub-acquisition time, the acquisition data and the corrected acquisition data corresponding to the same sub-acquisition time are fused and matched to obtain the matching result of the same sub-acquisition time. The target matching result is composed of multiple matching results of the same sub-acquisition time.

[0184] As can be seen, this optional embodiment transforms the collected data of one data collector by using the transformation parameters corresponding to one data collector, and then fuses and matches it with the collected data of one data collector that does not require transformation. This can improve the accuracy of data fusion and matching, thereby improving the accuracy of target recognition. Then, based on different matching results, corresponding features are added to the target in the region to achieve accurate completion of missing target data.

[0185] In this optional embodiment, the feature addition module 307 may optionally perform feature assignment operations on targets within the first region based on the target matching result in the following specific ways:

[0186] When the target matching result includes the first target matching result, for any first target, the features of the first target identified by the benchmark correction object and the features of the first target identified by the corrected object are both added to the first target;

[0187] When the target matching result includes the second target matching result, determine the reference identification parameters that match the current scene of the first region. The reference identification parameters include the reference distance, which is the distance determined based on the location of the reference correction object or the location of the object being corrected, such as 150 meters.

[0188] Based on the baseline identification parameters, a second region is determined, and the range of the second region is smaller than that of the first region.

[0189] For any second target within the second region, determine the type of the collector that successfully identified the second target. When the type of the collector includes a type that matches the second region, determine the target collector that matches the type that matches the second region, and add the features identified by the target collector to the second target.

[0190] Specifically, when the area of ​​the second region is the area formed by the distance less than or equal to the reference distance and the reference distance, the target collector is the reference correction object; when the area of ​​the second region is the area formed by the distance greater than the reference distance and the reference distance, the target collector is the object to be corrected.

[0191] As can be seen, this optional embodiment can accurately filter out the corresponding features through different target matching results, thereby achieving accurate completion of missing target data and improving completion efficiency.

[0192] Example 3

[0193] Please see Figure 7 , Figure 7 This is a schematic diagram of another device disclosed in an embodiment of the present invention for performing target identification operations in a region. This device can be applied to any scenario requiring target identification, such as vehicle monitoring. Figure 7 As shown, the device may include:

[0194] Memory 401 storing executable program code;

[0195] Processor 402 coupled to memory 401;

[0196] The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the method for performing target identification operations in the region as described in Embodiment 1 or Embodiment 2 of the present invention.

[0197] Example 4

[0198] This invention discloses a computer-readable storage medium storing computer instructions that, when invoked, execute steps in the method for performing a target identification operation in a region as described in Embodiment 1 or Embodiment 2 of this invention.

[0199] Example 5

[0200] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps of the method for performing a target identification operation in a region as described in Embodiment 1 or Embodiment 2.

[0201] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0202] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0203] Finally, it should be noted that the method and apparatus for performing target identification operations in a region disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for performing target identification operations in a region, characterized in that, The method includes: A baseline correction object matching the current scene of a first region is determined. The first region is provided with the baseline correction object and the object to be corrected. The baseline correction object and the object to be corrected are data acquisition devices of different types. The baseline correction object is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the baseline correction object and the acquisition time corresponding to the acquisition data. The object to be corrected is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the object to be corrected and the acquisition time corresponding to the acquisition data. Using the acquisition time corresponding to the benchmark correction object as a benchmark, and based on the acquisition time corresponding to the object being corrected, acquisition data matching the acquisition time corresponding to the benchmark correction object is obtained from the acquisition data corresponding to the object being corrected; wherein, the acquisition data matching the acquisition time corresponding to the benchmark correction object includes acquisition data whose acquisition time is later than the acquisition time corresponding to the benchmark correction object and acquisition data whose acquisition time is earlier than the acquisition time corresponding to the benchmark correction object. Based on the acquisition time corresponding to the benchmark correction object, the acquisition data that matches the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, a correction operation is performed on the acquisition data corresponding to the object to be corrected to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object. The corrected data collected for the object being corrected is used as the basis for performing identification operations on the target in the first region. The method further includes: From the corrected acquisition data corresponding to the object being corrected and the acquisition data corresponding to the benchmark corrected object, the first target acquisition data that needs to be converted is selected; Perform a parameter transformation operation on the first target acquisition data that matches the first target acquisition data to obtain the parameter-transformed second target acquisition data; Based on the first target acquisition data and the third target acquisition data after parameter transformation, a fusion matching operation is performed on the targets in the first region to obtain the target matching result; Based on the target matching result, perform feature assignment operations on the targets within the first region; The third target acquisition data is the other acquisition data excluding the first target acquisition data from the corrected acquisition data corresponding to the corrected object and the acquisition data corresponding to the benchmark corrected object.

2. The method for performing target identification operations in a region according to claim 1, characterized in that, The data collected corresponding to the benchmark correction object consists of at least one first sub-collected data, and the acquisition time corresponding to the benchmark correction object consists of the first sub-collected times corresponding to all the first sub-collected data. The collected data corresponding to the object being corrected consists of at least one second sub-collected data, and the collection time corresponding to the object being corrected consists of the second sub-collected times corresponding to all the second sub-collected data.

3. The method for performing target identification operations in a region according to claim 2, characterized in that, The step of performing a correction operation on the acquisition data corresponding to the object to be corrected, based on the acquisition time corresponding to the benchmark correction object, the acquisition data matching the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object, includes: For any first sub-acquisition time corresponding to the benchmark correction object, based on the first sub-acquisition time, from all second sub-acquisition times, select the first target sub-acquisition time whose time is later than the first sub-acquisition time and the second target sub-acquisition time whose time is earlier than the first sub-acquisition time; Based on the first sub-acquisition time, the first target sub-acquisition time, and the second target sub-acquisition time, determine the time correction parameter corresponding to the object to be corrected; Based on the time correction parameter corresponding to the object being corrected, the second sub-collection data corresponding to the first target sub-collection time, and the second sub-collection data corresponding to the second target sub-collection time, the collection data matching the first sub-collection time is determined; The corrected acquisition data corresponding to the object being corrected includes all acquisition data that matches the first sub-acquisition time.

4. The method for performing target identification operations in a region according to any one of claims 1-3, characterized in that, The target matching result includes a first target matching result and / or a second target matching result, wherein: The first target matching result is used to represent all first targets that are successfully matched based on the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object; The second target matching result is used to represent all second targets that did not successfully match the collected data corresponding to the benchmark correction object and the corrected collected data corresponding to the corrected object.

5. The method for performing target identification operations in a region according to claim 4, characterized in that, The step of assigning features to targets within the first region based on the target matching result includes: When the target matching result includes the first target matching result, for any first target, the features of the first target identified by the benchmark correction object and the features of the first target identified by the corrected object are both added to the first target; When the target matching result includes the second target matching result, a reference identification parameter matching the current scene of the first region is determined. The reference identification parameter includes a reference distance, which is a distance determined based on the location of the reference correction object or the location of the object being corrected. Based on the reference identification parameters, a second region is determined, wherein the range of the second region is smaller than the range of the first region; For any second target within the second region, determine the type of the collector that successfully identified the second target. When the type of the collector includes a type that matches the second region, determine the target collector that matches the type that matches the second region, and add the features identified by the target collector to the second target. Wherein, when the area of ​​the second region is less than or equal to the area formed by the distance between the reference distance and the reference distance, the target collector is the reference correction object; when the area of ​​the second region is greater than the area formed by the distance between the reference distance and the reference distance, the target collector is the object to be corrected.

6. The method for performing target identification operations in a region according to any one of claims 1-3, characterized in that, The step of performing a parameter transformation operation on the first target acquisition data that matches the first target acquisition data to obtain the parameter-transformed second target acquisition data includes: When the first target acquisition data is the corrected acquisition data corresponding to the object to be corrected, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters that match the benchmark corrected object to obtain the second target acquisition data after parameter conversion. When the first target acquisition data is the acquisition data corresponding to the benchmark correction object, a parameter conversion operation is performed on the first target acquisition data based on the conversion parameters that match the correction object to obtain the parameter-converted second target acquisition data.

7. An apparatus for performing a target identification operation in a region, characterized in that, The apparatus is used to implement the method for performing a target identification operation in a region as described in any one of claims 1-6, and the apparatus comprises: The determination module is used to determine a reference correction object that matches the current scene of the first region. The first region is provided with the reference correction object and the object to be corrected. The reference correction object and the object to be corrected are data acquisition devices of different types. The reference correction object is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the reference correction object and the acquisition time corresponding to the acquisition data. The object to be corrected is used to perform data acquisition operations on the first region to obtain the acquisition data corresponding to the object to be corrected and the acquisition time corresponding to the acquisition data. The acquisition module is used to acquire, based on the acquisition time corresponding to the benchmark correction object, the acquisition data that matches the acquisition time corresponding to the acquisition time corresponding to the object being corrected from the acquisition data corresponding to the object being corrected; wherein, the acquisition data that matches the acquisition time corresponding to the benchmark correction object includes acquisition data whose acquisition time is later than the acquisition time corresponding to the benchmark correction object and acquisition data whose acquisition time is earlier than the acquisition time corresponding to the benchmark correction object. The correction module is used to perform a correction operation on the acquisition data corresponding to the object to be corrected based on the acquisition time corresponding to the benchmark correction object, the acquisition data that matches the acquisition time corresponding to the benchmark correction object, and the acquisition time corresponding to the acquisition data, so as to obtain the corrected acquisition data of the object to be corrected at the acquisition time corresponding to the benchmark correction object. The corrected data collected for the object being corrected is used as the basis for performing identification operations on the target in the first region.

8. An apparatus for performing a target identification operation in a region, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the method for performing target identification operations in the region as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, which, when invoked, are used to perform the method for performing a target identification operation in the region as described in any one of claims 1-6.

Images