Tunnel scene recognition method, device, equipment, storage medium and product

Abstract

The present application discloses a tunnel scene recognition method, device, equipment, storage medium and product, which relates to the field of sensor technology, including: detecting the scene to be recognized based on millimeter wave radar to obtain detection data frames; obtaining target detection features according to the detection data frames; target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces and the proportion of close-range high signal-to-noise ratio point traces; classifying the scene to be recognized based on the target detection features to obtain scene classification results; determining the single-frame tunnel probability of the scene to be recognized according to the scene classification results. Since multiple point trace distribution features are used as parameters, the distribution features of the stationary point traces at far and near can be fully considered, and a classification model is used to distinguish tunnel scenes from non-tunnel scenes. On the one hand, the influence of pitch accuracy on the recognition results is weakened, and on the other hand, there is no need to find the boundary between the tunnel scene and non-tunnel scene features, thereby improving the recognition efficiency.

Description

Technical Field

[0001] The present application relates to the field of sensor technology, and in particular to tunnel scene recognition methods, devices, equipment, storage media and products. Background Art

[0002] In recent years, with the development of automobile intelligence, radar sensors are increasingly used in automobiles. Among them, millimeter-wave radar has attracted the attention of major automobile manufacturers due to its advantages such as easy installation, no weather influence, strong anti-interference ability in complex environments, and low cost. However, it is easy to generate multipath and ghost images when detecting targets in tunnel scenes, which has a certain impact on radar detection performance.

[0003] One solution in the prior art is based on millimeter-wave radar, which counts specific areas in front of the vehicle for tunnel scene recognition. This solution only uses the feature that there are many and scattered static points in front of the vehicle in the tunnel scene, and it makes recognition errors in certain scenarios (such as turning the vehicle or under an overpass). Another solution in the prior art is based on 4D millimeter-wave radar point cloud, which sets two areas, far and near, in front of the vehicle for detection. However, the judgment condition of the far area in this solution is heavily dependent on the height information of the static points, and the detection accuracy of the millimeter-wave radar in the height dimension is low. And as the number of scene samples increases, the features described in this solution are difficult to show a clear boundary in the tunnel / non-tunnel scene, and it is difficult to distinguish the tunnel / non-tunnel by a specific threshold. Summary of the invention

[0004] The main purpose of the present application is to provide a tunnel scene recognition method, device, equipment, storage medium and product, aiming to solve the technical problem of low detection accuracy when performing tunnel scene detection in the existing solution method.

[0005] To achieve the above objectives, the present application proposes a tunnel scene recognition method, which includes:

[0006] Detect the scene to be identified based on the millimeter wave radar to obtain a detection data frame;

[0007] Acquire target detection features according to the detection data frame; the target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces;

[0008] Classifying the scene to be identified based on the target detection feature to obtain a scene classification result;

[0009] The single-frame tunnel probability of the scene to be identified is determined according to the scene classification result.

[0010] In one implementation, the step of classifying the scene to be identified based on the target detection feature to obtain a scene classification result includes:

[0011] Perform queue data processing based on the target detection features to obtain target detection feature values ​​corresponding to various detection features in each frame of data;

[0012] Determine a target detection sample according to the target detection feature value;

[0013] The target detection samples are classified by a classification model to obtain a scene classification result.

[0014] In one implementation, the step of performing queue data processing based on the target detection features to obtain target detection feature values ​​corresponding to various detection features in each frame of data includes:

[0015] Storing the target detection features of each single frame in the detection data frame into a frame data queue;

[0016] The target detection features corresponding to the valid frames in the frame data queue are averaged to determine the target detection feature values ​​of the frame data queue.

[0017] In one implementation, the step of acquiring target detection features according to the detection data frame includes:

[0018] Traversing all the traces of a single frame according to the detection data frame, and counting the trace parameters of the stationary traces;

[0019] The target detection features of a single frame are determined based on the point trace parameters.

[0020] In one implementation, the step of determining the single-frame tunnel probability of the scene to be identified according to the scene classification result includes:

[0021] Get the adjustable parameters for probability calculation;

[0022] Determining an initial single-frame tunnel probability of the scene to be identified based on the scene classification result and the adjustable parameter;

[0023] The initial single-frame tunnel probability is filtered by a preset filtering method to obtain an updated single-frame tunnel probability.

[0024] In one implementation, the method further includes:

[0025] Get preset driving conditions;

[0026] When the driving condition in the target detection data frame does not satisfy the preset driving condition, the target detection data frame is marked as an invalid frame.

[0027] In addition, to achieve the above-mentioned purpose, the present application also proposes a tunnel scene recognition device, the tunnel scene recognition device comprising:

[0028] A data detection module, used to detect the scene to be identified based on the millimeter wave radar and obtain a detection data frame;

[0029] A feature acquisition module is used to acquire the target detection features of a single frame according to the detection data frame; the target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces;

[0030] A scene classification module, used to classify the scene to be identified based on the target detection feature to obtain a scene classification result;

[0031] A probability determination module is used to determine the single-frame tunnel probability of the scene to be identified according to the scene classification result.

[0032] In addition, to achieve the above-mentioned purpose, the present application also proposes a tunnel scene recognition device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the tunnel scene recognition method described above.

[0033] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the tunnel scene recognition method described above are implemented.

[0034] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the tunnel scene recognition method described above are implemented.

[0035] One or more technical solutions proposed in this application have at least the following technical effects:

[0036] The present application detects the scene to be identified based on millimeter wave radar to obtain a detection data frame; obtains target detection features based on the detection data frame; target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces; classifies the scene to be identified based on the target detection features to obtain a scene classification result; determines the single-frame tunnel probability of the scene to be identified based on the scene classification result. Since a variety of point trace distribution features are used as parameters, the distribution features of the stationary point traces at a distance and near can be fully considered, and a classification model is used to distinguish between tunnel scenes and non-tunnel scenes. On the one hand, the influence of pitch accuracy on the recognition results is weakened, and on the other hand, there is no need to find the boundary between the features of the tunnel scene and the non-tunnel scene, thereby improving the efficiency of recognition. In addition, the present application also constructs a single-frame tunnel probability calculation model based on the scene classification results, which improves the accuracy of the single-frame tunnel calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0039] Figure 1 A schematic diagram of a flow chart provided for the first embodiment of the tunnel scene recognition method of the present application;

[0040] Figure 2 A schematic diagram of a flow chart provided for the second embodiment of the tunnel scene recognition method of the present application;

[0041] Figure 3 A schematic diagram of a flow chart provided for Embodiment 3 of the tunnel scene recognition method of the present application;

[0042] Figure 4 This is a schematic diagram of the module structure of the tunnel scene recognition device according to an embodiment of the present application;

[0043] Figure 5 Schematic diagram of the device structure of the hardware operating environment involved in the tunnel scene recognition method in the embodiment of the present application.

[0044] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0045] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0046] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0047] The main solution of the embodiment of the present application is: to detect the scene to be identified based on the millimeter wave radar to obtain the detection data frame; to obtain the target detection features of a single frame according to the detection data frame; to classify the scene to be identified based on the target detection features to obtain the scene classification result; and to determine the single-frame tunnel probability of the scene to be identified according to the scene classification result.

[0048] One of the existing technologies currently proposed a scheme for identifying tunnel scenes based on millimeter-wave radar. The specific approach is to set a specific area in front of the vehicle and determine whether to enter the tunnel by counting the number of static traces and the length of the static traces in the area. This scheme only uses the feature that there are many and scattered static traces in front of the vehicle in the tunnel scene, and it may make mistakes in some scenes (such as when the vehicle turns or under an overpass).

[0049] Another implementation method of the current prior art proposes a tunnel scene recognition solution based on 4D millimeter-wave radar point cloud. The specific approach is to set up two areas, far and near, in front of the vehicle. In the far area, the Z-direction cluster length and variance of the stationary point traces are calculated; in the near area, the probability density of the Y-direction mean of the stationary point traces is calculated. Combine the calculation results of the far and near areas to determine whether to enter the tunnel. This solution has two problems. One is that the judgment conditions in the far area are heavily dependent on the height information of the stationary point traces, while the detection accuracy of the millimeter-wave radar in the height dimension is low. Second, as the number of scene samples increases, the features described in this solution are difficult to have a clear boundary in the tunnel / non-tunnel scenario, and it is difficult to distinguish the tunnel / non-tunnel through a specific threshold.

[0050] This application provides a solution that aims to fully utilize the multiple distribution features of static point traces in the process of millimeter radar environmental perception to identify tunnel scenes in real time, and to improve the accuracy of the recognition results by constructing a single-frame tunnel probability calculation model and outputting a single-frame tunnel probability filtering process, thereby optimizing the target detection function of the millimeter-wave radar tunnel scene. Generally speaking, the detection performance of millimeter-wave radar in the height dimension is low, and it is impossible to accurately identify the tunnel scene by relying solely on the height information of the target; in addition, as the number of scene samples increases, it is difficult for the various features mentioned in the existing solutions to have a clear boundary, and it is difficult to directly distinguish between tunnels and non-tunnels.

[0051] The present application makes full use of the various features of the static point trace distribution and adopts a classification model to identify tunnel scenes. On the one hand, this solution can weaken the influence of the pitch detection accuracy, and on the other hand, there is no need to find the boundary of the tunnel / non-tunnel scene features; in addition, a single-frame tunnel probability calculation model is constructed to improve the accuracy of the single-frame tunnel probability calculation; further, the final output single-frame tunnel probability is filtered, which also reduces the impact of unstable radar detection performance.

[0052] It should be noted that the execution subject of this embodiment may be a computing service device with data processing, network communication and program running functions, such as a vehicle-mounted terminal, a server, etc., or an electronic device, a virtual device, etc. capable of realizing the above functions. The following takes a tunnel scene recognition device (hereinafter referred to as the recognition device) as an example to illustrate this embodiment and the following embodiments.

[0053] Based on this, the embodiment of the present application provides a tunnel scene recognition method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the tunnel scene recognition method of the present application.

[0054] In this embodiment, the tunnel scene recognition method includes steps S10 to S40:

[0055] Step S10, detecting the scene to be identified based on the millimeter wave radar to obtain a detection data frame;

[0056] Step S20, obtaining target detection features according to the detection data frame; the target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces;

[0057] It is understandable that millimeter wave radar is a radar sensor that works in the millimeter wave band of 30 to 300 GHz and a wavelength of 1 to 10 mm. In tunnels, due to the walls and top structures, electromagnetic waves are prone to multipath reflection, causing the radar to receive multiple echoes, making it difficult to accurately obtain information such as the target's true position and speed. Especially at the tunnel entrance or the bend of the tunnel, the multipath reflection phenomenon is particularly obvious, which may produce false mirror targets and seriously affect the detection and recognition effect of the radar in the tunnel scene.

[0058] It is understandable that when using millimeter wave radar for detection, the radar can emit electromagnetic waves and receive echo signals reflected by the target to detect the target. After amplification, filtering and other processing, these echo signals are converted into a series of data points, that is, point traces. Stationary point traces are data points formed by radar electromagnetic wave signals reflected by stationary targets.

[0059] It should be noted that in the tunnel scene, the four features of the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces are different from those in the non-tunnel scene, but there is no obvious boundary that can directly distinguish the tunnel scene from the non-tunnel scene. Therefore, in the embodiment of the present application, these four features are selected as parameters, and a classification model is designed to distinguish between the tunnel scene and the non-tunnel scene.

[0060] Specifically, in step S30, the scene to be identified is classified based on the target detection feature to obtain a scene classification result.

[0061] It is understandable that in the embodiment of the present application, the target detection features can be input into the above classification model for classification, and then the scene classification result can be determined. By analyzing the scene classification result, the single-frame tunnel probability that the scene to be identified is a tunnel scene can be determined. Specifically, step S40 determines the single-frame tunnel probability of the scene to be identified based on the scene classification result.

[0062] The embodiment of the present application detects the scene to be identified based on millimeter wave radar to obtain a detection data frame; obtains target detection features based on the detection data frame; the target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces; classifies the scene to be identified based on the target detection features to obtain a scene classification result; determines the single-frame tunnel probability of the scene to be identified based on the scene classification result. Since a variety of point trace distribution features are used as parameters, the distribution features of the stationary point traces at a distance and near can be fully considered, and a classification model is used to distinguish between tunnel scenes and non-tunnel scenes. On the one hand, the influence of pitch accuracy on the recognition results is weakened, and on the other hand, there is no need to find the boundary between the features of the tunnel scene and the non-tunnel scene, thereby improving the efficiency of recognition. In addition, the present application also constructs a single-frame tunnel probability calculation model based on the scene classification results to improve the accuracy of the single-frame tunnel calculation.

[0063] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 , the step of acquiring target detection features according to the detection data frame comprises:

[0064] Step S21, traversing all the traces of a single frame according to the detection data frame, and counting the trace parameters of the static traces;

[0065] Step S22, determining the target detection features of a single frame based on the point trace parameters.

[0066] It can be understood that the above-mentioned point trace parameters can also be used to determine the parameters related to the point trace characteristics of the stationary point trace, such as the spatial position of each point trace, the point trace pitch angle, the positional relationship between the point trace and the vehicle, etc., and the embodiments of the present application are not limited to this.

[0067] It should be noted that in the embodiment of the present application, four features of a single-frame static point trace can be counted, which may include the variance of the static point trace pitch angle , the proportion of static points in front of the vehicle 、Number of close-range points And the proportion of close-range high signal-to-noise ratio traces , where the number of close-range points is Can be directly counted, the variance of the pitch angle of the stationary point trace , the proportion of static points in front of the vehicle And the proportion of close-range high signal-to-noise ratio traces The calculation method can be as follows:

[0068] ;

[0069] ;

[0070] ;

[0071] in, is the number of points, is the pitch angle of each stationary point detected by the millimeter-wave radar, is the average value of the pitch angles of these stationary points; is the number of stationary points directly in front of the vehicle, is the number of overall stationary traces; is the number of stationary traces with high signal-to-noise ratio at close range.

[0072] It is understandable that, in practical applications, a signal-to-noise ratio threshold and a distance threshold may be set to determine the close-range stationary point traces among the stationary point traces and the number of stationary point traces with a high close-range signal-to-noise ratio.

[0073] In an implementation of the embodiment of the present application, for each detection data frame, it is possible to determine whether it is an invalid frame according to a preset driving condition. Specifically, in the embodiment of the present application, the method further includes:

[0074] Get preset driving conditions;

[0075] When the driving condition in the target detection data frame does not satisfy the preset driving condition, the target detection data frame is marked as an invalid frame.

[0076] For example, the above-mentioned preset driving conditions may include: (1) the vehicle speed is greater than 10 km / h; (2) the yaw rate is less than 0.25 rad / s. In practical applications, other vehicle speeds or yaw rates may be used as preset driving conditions, and the present application embodiment does not limit this.

[0077] It should be noted that when the preset driving conditions are not met, the single-frame tunnel probability can be directly set The single frame statistics are set to 0 and inserted into the queue, and the single frame is marked as an invalid frame.

[0078] The embodiment of the present application traverses all the traces of a single frame according to the detection data frame, and counts the trace parameters of the static traces; the target detection features of the single frame are determined based on the trace parameters. Since the trace features are determined by determining the trace parameters corresponding to all the traces in the detection data frame, the distribution features of the static traces are fully utilized, and the accuracy of tunnel recognition is improved.

[0079] Based on the first embodiment and / or the second embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those of the above-mentioned first embodiment and / or the second embodiment can refer to the above introduction, and will not be repeated later. Figure 3 The step of classifying the scene to be identified based on the target detection feature to obtain a scene classification result includes:

[0080] Step S31, performing queue data processing based on the target detection features to obtain target detection feature values ​​corresponding to various detection features in each frame of data;

[0081] Step S32, determining a target detection sample according to the target detection feature value;

[0082] Step S33, classifying the target detection samples through a classification model to obtain a scene classification result.

[0083] It should be noted that the detection data frame may include multiple data frames, and when counting the above-mentioned features of the stationary point traces, the structure of a single data frame (single frame for short) is very different. Therefore, in the embodiment of the present application, the feature data of multiple data frames can be continuously counted, and the mean of the feature data can be used as the basis for classification.

[0084] Specifically, in the embodiment of the present application, a queue capable of storing M frames of data can be set, and each frame of data in the queue can record four characteristic values ​​of a data frame, which can be recorded as the variance of the pitch angle of the static point trace, , the proportion of static points in front of the vehicle 、Number of close-range points , the proportion of close-range high signal-to-noise ratio points In one implementation of the embodiment of the present application, the process of processing the data in the queue may be: for valid frames in the queue data, the feature means corresponding to each type of detection feature are respectively counted. Specifically, the number of valid frames in the queue may be N, and the feature means for each type of detection feature may be calculated as follows:

[0085] ;

[0086] ;

[0087] ;

[0088] ;

[0089] in, It is used to represent the target detection feature value corresponding to the pitch angle variance of the stationary point trace. It is used to represent the target detection feature value corresponding to the proportion of the static point trace directly in front of the vehicle. It is used to represent the target detection feature value corresponding to the number of close-range points. The target detection feature value corresponding to the ratio of close-range high signal-to-noise ratio points is used to represent the target detection feature value. That is, the step of performing queue data processing based on the target detection feature to obtain the target detection feature value corresponding to each type of detection feature in each frame of data includes:

[0090] Storing the target detection features of each single frame in the detection data frame into a frame data queue;

[0091] The target detection features corresponding to the valid frames in the frame data queue are averaged to determine the target detection feature values ​​of the frame data queue.

[0092] In the embodiment of the present application, a support vector machine (SVM) classification method can be used to establish a classification model.

[0093] Specifically, the target detection sample can be determined based on the above target detection feature values. ,in, Used to indicate storage i The four target detection feature values ​​of samples, that is, ,in It is a label, and its value can be 1 or -1.

[0094] Furthermore, the classification hyperplane equation can be constructed:

[0095] ;

[0096] in, , is the normal vector of the hyperplane, is the bias value.

[0097] Furthermore, constraints can be set for it:

[0098] ;

[0099] in, γ is the distance between the support vector and the hyperplane. and It can be scaled arbitrarily without affecting the hyperplane, so γ can be taken as 1.

[0100] Furthermore, the model can be optimized, specifically:

[0101] max , st 1;

[0102] The above formula means: under the constraint condition Next, so that The largest value.

[0103] It should be noted that the above optimization model can be solved by using the Lagrange multiplier method or other methods, and the result obtained is and .

[0104] It can be understood that by optimizing these two parameters, the classification model can be optimized.

[0105] In the specific implementation, the target detection samples can be input into the classification model for classification and the classification results can be obtained:

[0106] ;

[0107] Further, when the classification result is determined, the single-frame tunnel probability can be calculated based on the classification result. Specifically, the single-frame tunnel probability is calculated. The calculation method can be as follows:

[0108] ;

[0109] in, m is an adjustable parameter for probability calculation.

[0110] That is, the step of determining the single-frame tunnel probability of the scene to be identified according to the scene classification result includes:

[0111] Get the adjustable parameters for probability calculation;

[0112] Determining an initial single-frame tunnel probability of the scene to be identified based on the scene classification result and the adjustable parameter;

[0113] The initial single-frame tunnel probability is filtered by a preset filtering method to obtain an updated single-frame tunnel probability.

[0114] It is understandable that in order to improve the accuracy of recognition, the obtained initial single-frame tunnel probability may be filtered. The filtering method may be an α~β filtering method or other methods, which are not limited in the embodiment of the present application.

[0115] In one implementation of the embodiment of the present application, the method of filtering the initial single-frame tunnel probability may be as follows:

[0116]

[0117] in, is the updated tunnel probability, is the tunnel probability before updating, A custom weight.

[0118] The embodiment of the present application processes queue data based on target detection features to obtain target detection feature values ​​corresponding to various detection features in each frame of data; determines target detection samples based on the target detection feature values; and classifies the target detection samples through a classification model to obtain scene classification results. By building a classification model to classify target detection samples, accurate detection of tunnel probability is achieved.

[0119] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the tunnel scene recognition method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0120] This application also provides a tunnel scene recognition device, please refer to Figure 4 , the tunnel scene recognition device comprises:

[0121] A data detection module 10 is used to detect the scene to be identified based on the millimeter wave radar and obtain a detection data frame;

[0122] The feature acquisition module 20 is used to acquire the target detection features of a single frame according to the detection data frame; the target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces;

[0123] A scene classification module 30, used to classify the scene to be identified based on the target detection feature to obtain a scene classification result;

[0124] The probability determination module 40 is used to determine the single-frame tunnel probability of the scene to be identified according to the scene classification result.

[0125] The tunnel scene recognition device provided by the present application adopts the tunnel scene recognition method in the above embodiment, which can solve the technical problem of low detection accuracy when the existing method detects tunnel scenes. Compared with the prior art, the beneficial effects of the tunnel scene recognition device provided by the present application are the same as the beneficial effects of the tunnel scene recognition method provided by the above embodiment, and the other technical features in the tunnel scene recognition device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0126] The present application provides a tunnel scene recognition device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the tunnel scene recognition method in the above-mentioned embodiment 1.

[0127] Reference below Figure 5 , which shows a schematic diagram of the structure of a tunnel scene recognition device suitable for implementing the embodiment of the present application. The tunnel scene recognition device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The tunnel scene recognition device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0128] like Figure 5As shown, the tunnel scene recognition device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the tunnel scene recognition device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the tunnel scene recognition device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a tunnel scene recognition device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided instead.

[0129] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0130] The tunnel scene recognition device provided by the present application adopts the tunnel scene recognition method in the above embodiment, which can solve the technical problem of low detection accuracy when the existing method detects tunnel scenes. Compared with the prior art, the beneficial effects of the tunnel scene recognition device provided by the present application are the same as the beneficial effects of the tunnel scene recognition method provided by the above embodiment, and the other technical features in the tunnel scene recognition device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0131] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0132] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0133] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the tunnel scene recognition method in the above-mentioned embodiment.

[0134] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0135] The computer-readable storage medium may be included in the tunnel scene recognition device; or may exist independently without being assembled into the tunnel scene recognition device.

[0136] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the tunnel scene recognition device, the tunnel scene recognition device:

[0137] Detect the scene to be identified based on the millimeter wave radar to obtain a detection data frame;

[0138] Acquire target detection features according to the detection data frame; the target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces;

[0139] Classifying the scene to be identified based on the target detection feature to obtain a scene classification result;

[0140] The single-frame tunnel probability of the scene to be identified is determined according to the scene classification result.

[0141] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0142] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0143] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0144] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned tunnel scene recognition method, and can solve the technical problem of low detection accuracy when performing tunnel scene detection in the existing solution method. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the present application are the same as the beneficial effects of the tunnel scene recognition method provided in the above-mentioned embodiment, and will not be repeated here.

[0145] The present application also provides a computer program product, including a computer program, which implements the steps of the tunnel scene recognition method as described above when executed by a processor.

[0146] The computer program product provided by the present application can solve the technical problem of low detection accuracy when the existing method detects tunnel scenes. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the tunnel scene recognition method provided by the above embodiment, which will not be repeated here.

[0147] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A tunnel scene recognition method, characterized in that: The method comprises: Detecting the scene to be identified based on the millimeter wave radar to obtain a detection data frame; the detection data frame includes multiple data frames; Acquire target detection features according to the detection data frame; the target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces; Classifying the scene to be identified based on the target detection feature to obtain a scene classification result; The single-frame tunnel probability of the scene to be identified is determined according to the scene classification result.

2. The tunnel scene recognition method according to claim 1, characterized in that: The step of classifying the scene to be identified based on the target detection feature to obtain a scene classification result includes: Perform queue data processing based on the target detection features to obtain target detection feature values ​​corresponding to various detection features in the detection data frame; Determine a target detection sample according to the target detection feature value; The target detection samples are classified by a classification model to obtain a scene classification result.

3. The tunnel scene recognition method according to claim 2, characterized in that: The step of performing queue data processing based on the target detection feature to obtain target detection feature values ​​corresponding to various detection features in the detection data frame includes: Storing the target detection feature of each of the detection data frames into a frame data queue; The target detection features corresponding to the valid frames in the frame data queue are averaged to determine the target detection feature values ​​corresponding to each type of detection feature in the detection data frame.

4. The tunnel scene recognition method according to claim 1, characterized in that: The step of acquiring target detection features according to the detection data frame comprises: Traversing all the traces of the data frame according to the detection data frame, and counting the trace parameters of the stationary traces; The target detection features of the data frame are determined based on the point trace parameters.

5. The tunnel scene recognition method according to claim 1, characterized in that: The step of determining the single-frame tunnel probability of the scene to be identified according to the scene classification result comprises: Get the adjustable parameters for probability calculation; Determining an initial single-frame tunnel probability of the scene to be identified based on the scene classification result and the adjustable parameter; The initial single-frame tunnel probability is filtered by a preset filtering method to obtain an updated single-frame tunnel probability.

6. The tunnel scene recognition method according to claim 1, characterized in that: The method further comprises: Get preset driving conditions; When the driving condition in the target detection data frame does not satisfy the preset driving condition, the target detection data frame is marked as an invalid frame.

7. A tunnel scene recognition device, characterized in that: The tunnel scene recognition device comprises: A data detection module, used to detect the scene to be identified based on the millimeter wave radar and obtain a detection data frame; the detection data frame includes multiple data frames; A feature acquisition module is used to acquire the target detection features of a single frame according to the detection data frame; the target detection features include: the variance of the pitch angle of the stationary point trace, the proportion of the stationary point trace directly in front of the vehicle, the number of close-range point traces, and the proportion of close-range high signal-to-noise ratio point traces; A scene classification module, used to classify the scene to be identified based on the target detection feature to obtain a scene classification result; A probability determination module is used to determine the single-frame tunnel probability of the scene to be identified according to the scene classification result.

8. A tunnel scene recognition device, characterized in that: The device comprises: a memory, a processor, and a tunnel scene recognition program stored in the memory and executable on the processor, wherein the tunnel scene recognition program is configured to implement the steps of the tunnel scene recognition method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium stores a tunnel scene recognition program, and when the tunnel scene recognition program is executed by the processor, the steps of the tunnel scene recognition method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the tunnel scene recognition method according to any one of claims 1 to 6 are implemented.

Images