Hump humping vehicle positioning method based on data fusion

Through the fusion technology of radar and video data, the inaccurate length measurement problem caused by rail oxidation and poor wheel contact in the hump length measurement equipment is solved, and higher measurement accuracy and automatic control accuracy are achieved, the calibration process is simplified and the marshalling is improved. Station efficiency.

CN118981013APending Publication Date: 2024-11-19SIGNAL & COMM RES INST OF CHINA ACAD OF RAILWAY SCI +3
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411085244.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

The existing hump length measuring equipment is inaccurate due to rust of rail oxidized conjunctiva, poor wheel rail contact and track bed leakage, and the length measuring data cannot be effectively calibrated when occupied by the length measuring section, which affects the automatic control accuracy of the marshalling station.

Method used

The method of positioning hump vehicle based on data fusion is adopted. Through the time calibration and spatial fusion of radar data and video data, combined with feature fusion, the target information of the fused hump vehicle is obtained for precise positioning and replacing traditional length measuring equipment. .

Benefits of technology

It improves measurement accuracy, reduces the risk of inaccurate data, simplifies calibration difficulty, and improves the control accuracy and composition efficiency of the automatic control system of the marshalling station.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118981013A_ABST
    Figure CN118981013A_ABST
Patent Text Reader

Abstract

The invention discloses a hump humping vehicle positioning method based on data fusion, which identifies and positions a hump humping vehicle through a mode of performing data fusion on video data and radar data, the fusion can make the advantages of a video and a radar complementary, the image information in the video is accurate, the object classification is easy, and the accuracy of the hump humping vehicle positioning is improved. However, the method has the advantages of short image detection sight distance, great influence of natural light environment, large radar range, long detection distance, high resolution for small objects, accurate radar detection result orientation, easy target positioning, applicability to extreme environment with poor light, realization of information complementation between sensors through fusion of the two, improvement of the current length measurement precision, and wide application range. The control precision of the hump automatic control system is improved, and the marshalling efficiency of a marshalling station is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of hump-sliding vehicle positioning, and in particular to a hump-sliding vehicle positioning method based on data fusion. Background Art

[0002] The length measuring device is one of the important basic equipment of the railway marshalling yard automatic control system. It is an indispensable control basis for the speed control of the control system; in speed control, the constant speed of the vehicle is directly related to the length of the length measurement. Therefore, the length measuring device is one of the basic equipment of the marshalling yard automatic control system, and is a measuring device for measuring the running position of the shunting vehicle.

[0003] At present, the hump length measurement technology at home and abroad is realized by measuring the short-circuit input impedance of the rail and the principle that it is approximately proportional to the distance from the short-circuit point to the starting end. The length error is large due to the serious oxidation conjunctiva corrosion of the rail, poor contact between the light vehicle wheel and the rail, leakage of the ballast bed, and the status of cables and various connecting lines. In addition, the idle length of the hump length measurement section in the busier marshalling yard is short, making it difficult to adjust and in a state of large errors for a long time.

[0004] Specifically, the measurement of the idle length of the marshaling line is generally based on the principle that the short-circuit input impedance of the track circuit is approximately proportional to the distance from the short-circuit point to the starting end. The uniformly distributed parameter equivalent circuit of the audio track circuit is as follows: Figure 1 shown.

[0005] According to the uniform transmission line theory, when the transmission distance is x, the input voltage and current of the segment can be expressed by the following equations:

[0006]

[0007] Among them, sh is the hyperbolic sine function, ch is the two-zone cosine function, Z0 is the track circuit characteristic impedance, γ is the transmission coefficient, U1 is the output voltage, U2 is the input voltage, I1 is the output current, and I2 is the input current.

[0008] Computer length measurement uses the two rails of the marshaling line and the short-circuit line at the end to form a closed length measurement track circuit as the basic sensor component. When the shunting train occupies or travels on the length measurement section track, its last axle is equivalent to a moving short-circuit line at the end of the track circuit. As the distance between the axle and the beginning of the track changes, the short-circuit input impedance of the track circuit, that is, the primary parameter of the rail, will change, and the track length measurement signal generated in the length measurement system will change accordingly. The relationship between the input impedance of the track circuit and the length of the length measurement track circuit is as follows:

[0009] Z 入短 =Z0th(γx)=Kx

[0010] Among them, th is the hyperbolic tangent function, Z 入短 is the short-circuit impedance, and K is the proportionality coefficient.

[0011] Computer length measurement uses computers to digitize information data, and uses reasonable mathematical calculation models and software to perform artificial intelligence processing on the length measurement system, calculate the nonlinear characteristics of the entire line, and ensure the accuracy of remote measurement.

[0012] The computer length measurement system adopts a microcomputer to independently constitute the system in terms of architecture, directly exchanges data information with the control and management computer, and transmits the length measurement related data of each channel in the entire field.

[0013] As mentioned above, the above solution: (1) cannot solve the problem of inaccurate length measurement data due to severe oxidized conjunctiva corrosion of the rails, poor contact between the light vehicle wheels and the rails, and leakage of the ballast bed; and (2) cannot solve the problem of being unable to calibrate when the length measurement section is occupied.

[0014] In view of this, the present invention is proposed. Summary of the invention

[0015] The purpose of the present invention is to provide a hump-sliding vehicle positioning method based on data fusion, which can replace the existing hump length measuring equipment, improve the measurement accuracy, reduce the risk of inaccurate data and reduce the difficulty of calibration.

[0016] The objective of the present invention is achieved through the following technical solutions:

[0017] A method for positioning a humpback vehicle based on data fusion, comprising:

[0018] Collect radar data and video data, combine the collection period of radar data and video data, perform time calibration on the two types of data, and calculate the time difference;

[0019] For the time-calibrated radar data and video data, the target information of the hump-slipping vehicle is detected and projected into the road plane coordinate system to obtain the target information of the hump-slipping vehicle in the road plane coordinate system of the radar data; for the video data, the target information is detected and projected into the road plane coordinate system to obtain the target information of the video data in the road plane coordinate system;

[0020] After correcting the hump-slipping vehicle target information of the radar data in the road plane coordinate system in combination with the time difference, completing spatial fusion with the corresponding target information of the video data in the road plane coordinate system, and then performing consistent feature fusion to obtain fused hump-slipping vehicle target information;

[0021] The fused hump-sliding vehicle target information is used to locate the hump-sliding vehicle target.

[0022] It can be seen from the technical solution provided by the present invention that the hump shunting vehicle can be identified and located by applying video data and radar data for data fusion. The fusion can make the advantages of video and radar complement each other. The image in the video obtains accurate information and is easy to classify objects, but the image detection line of sight is short and is greatly affected by the natural light environment. The radar has a large range, a long detection distance, and a high resolution for small objects. The radar detection results are accurate in direction and easy to locate the target, and can be applied to extreme environments with poor light. The fusion of the two can achieve information complementarity between sensors, which can improve the current length measurement accuracy, improve the control accuracy of the hump automatic control system, and improve the marshalling efficiency of the marshalling yard. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use 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 ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0024] Figure 1 An equivalent circuit diagram of a length measuring track circuit provided as the background technology of the present invention;

[0025] Figure 2 A flow chart of a method for positioning a hump-type vehicle based on data fusion provided by an embodiment of the present invention;

[0026] Figure 3 A schematic diagram of an implementation of an idle track length measurement technology provided by an embodiment of the present invention;

[0027] Figure 4 A schematic flow chart of a method for locating a humpback vehicle based on data fusion provided by an embodiment of the present invention;

[0028] Figure 5 A schematic diagram of a radar coordinate system and a road plane coordinate system provided by an embodiment of the present invention;

[0029] Figure 6 A schematic diagram of a radar coordinate system and a road plane coordinate system provided by an embodiment of the present invention;

[0030] Figure 7 A schematic diagram of the overall data fusion process provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.

[0032] First, the terms that may be used in this article are explained as follows:

[0033] The terms "include", "comprises", "contains", "has" or other descriptions with similar semantics should be interpreted as non-exclusive inclusion. For example, including certain technical feature elements (such as raw materials, components, ingredients, carriers, dosage forms, materials, dimensions, parts, components, mechanisms, devices, steps, procedures, methods, reaction conditions, processing conditions, parameters, algorithms, signals, data, products or products, etc.) should be interpreted as including not only certain technical feature elements explicitly listed, but also other technical feature elements known in the art that are not explicitly listed.

[0034] The term "consisting of..." means excluding any technical feature elements not explicitly listed. If this term is used in a claim, it will make the claim closed, so that it does not contain technical feature elements other than the technical feature elements explicitly listed, except for the conventional impurities related to them. If this term only appears in a clause of a claim, it only limits the elements explicitly listed in the clause, and the elements recorded in other clauses are not excluded from the overall claim.

[0035] The following is a detailed description of a method for locating a humpback vehicle based on data fusion provided by the present invention. The contents not described in detail in the embodiments of the present invention belong to the prior art known to professionals in the field. If no specific conditions are specified in the embodiments of the present invention, the conventional conditions in the field or the conditions recommended by the manufacturer are followed. If the manufacturer of the instruments used in the embodiments of the present invention is not specified, they are all conventional products that can be purchased commercially.

[0036] The embodiment of the present invention provides a method for positioning a humpback vehicle based on data fusion, which is used for detecting and positioning humpback vehicle targets. First, before fusing the collected data of sensors (video and radar sensors) of different detection types, time alignment and spatial alignment are required to ensure that the result data detected by the two are from the same time and place, and then consistency detection is performed to improve the reliability of detection. When video and radar are used to detect targets on the track, different detection sources have their own advantages and disadvantages. The use of camera video can obtain richer scene information, but this target detection method is easily limited by the lighting environment. If radar is used to detect targets, although it has strong anti-interference ability and long detection distance, its disadvantage is low resolution of the target and poor effect on small target detection. Fusion of the two realizes information complementarity between sensors, improves the detection accuracy and application scope of the entire system. The main function of data fusion is to take video data and radar data as input, spatially fuse the two data, and finally output the fused data. The video data input by data fusion includes: the category of the video detection target, pixel coordinate position, pixel area size, and target confidence; the input radar data includes: the target coordinate position, target type, target speed, and target size in the radar detection plane area. The fused data output by data fusion includes: the category of the target, pixel coordinate position, pixel area size, target coordinate space position coordinate, and target confidence.

[0037] Figure 2 The overall process of the above method of the present invention is shown, which mainly includes:

[0038] (1) Collect radar data and video data, combine the acquisition cycle of radar data and video data, perform time calibration on the two types of data, and calculate the time difference.

[0039] (2) For the time-calibrated radar data, the target information of the humpback vehicle is detected and projected into the road plane coordinate system to obtain the target information of the humpback vehicle in the radar data in the road plane coordinate system; for the video data, the target information is detected and projected into the road plane coordinate system to obtain the target information of the video data in the road plane coordinate system.

[0040] (3) After correcting the hump-slipping vehicle target information of the radar data in the road plane coordinate system in combination with the time difference, complete spatial fusion with the corresponding target information of the video data in the road plane coordinate system, and then perform consistent feature fusion to obtain the fused hump-slipping vehicle target information.

[0041] In this part, the main consideration is that radar measurement of moving targets is relatively accurate, but the detection effect of stationary targets and target features is poor. Therefore, it is necessary to combine image recognition for feature recognition, filter out the influence of non-shunted vehicles on the detection results, and obtain the hump-shunted vehicle target information.

[0042] (4) Using the fused hump-sliding vehicle target information, the hump-sliding vehicle target is located.

[0043] The hump-shunting vehicle positioning method based on data fusion provided by the embodiment of the present invention can make the advantages of video images and radar complement each other. The image obtains accurate information and is easy to classify objects. The image detection line of sight is short and is greatly affected by the natural light environment. The radar has a large range, a long detection distance, and a high resolution for small objects. The radar detection results are accurate in direction and easy to locate the target, and can be applied to extreme environments with poor light. When both radar and video can detect the target, the accurate judgment of the object type and direction information can be achieved through the "and" judgment. However, the resolution of small objects is poor, and the obstacle detection under all time and space conditions can be achieved through the "or" relationship.

[0044] Based on the above scheme, the fused data can also be superimposed on the video image frame, then displayed on the video image frame and output to the preview interface. The main data output to the video image includes the target type, target pixel position, target pixel area size, etc. Other additional information of the target can be selectively superimposed.

[0045] To facilitate understanding, some principle introduction is provided below.

[0046] The sensing solution using radar and camera fusion is the mainstream trend in the current field of obstacle detection for autonomous vehicles, mainly because radar and camera have complementary characteristics. Radar is the best sensor for detecting distance and radial velocity. It has all-weather capabilities, especially considering that it can still work normally at night. However, radar cannot distinguish colors and has poor target classification capabilities. Camera has good color perception and classification capabilities, and its angular resolution capabilities are not inferior to lidar. However, they are limited in estimating speed and distance. In addition, image processing relies on chip computing power, and does not require information processing from millimeter-wave radar. Making full use of radar sensing information can greatly save computing resources. By comparing the characteristics of radar and camera, it can be found that there are many complementary characteristics between them. Therefore, applying radar and visual fusion perception technology to the field of target detection can effectively improve perception accuracy and enhance target detection capabilities.

[0047] The hump length measurement technology uses image recognition and processing technology to track the train sets in real time on the track through high-definition image acquisition equipment installed on the hump field light bridge. Taking into account the impact of meteorological and time factors on the image acquisition process, the millimeter-wave multi-target tracking radar can be used to assist the image acquisition process, and the millimeter-wave radar, high-definition camera, and embedded high-performance processor can be integrated into a radar-vision integrated device. At the same time, the video acquisition algorithm, radar measurement algorithm, and deep learning algorithm are embedded in the high-performance processor, and the high-precision detection capability of the multi-target radar is combined with the video feature analysis capability to realize the dynamic tracking and static detection of the train sets in the hump track. The idle track length measurement technology based on image recognition and processing can be realized as follows: Figure 3 shown.

[0048] In practical applications, according to the tracking capability of the multi-target tracking radar and the tracking requirements for the number of trains in the track, a radar-visual integrated device can be installed every 2-4 tracks on the hump light bridge, and transmitted to the hump automatic control system room via optical fiber, which can not only ensure the real-time data transmission, but also avoid lightning interference. This optimization solution can improve the current length measurement accuracy, improve the control accuracy of the hump automatic control system, and improve the marshaling efficiency of the marshaling yard.

[0049] For ease of understanding, the overall process of the present invention is described in detail below. Figure 4 The relevant processes are shown.

[0050] 1. In order to facilitate on-site installation and debugging, the camera and radar can be designed as an integrated structure, and ensure that the line drawn in the camera video coincides with the line drawn in the radar detection. At the same time, the camera should have a fixed lens focal length. During the initial installation and debugging, the camera with a fixed lens focal length should be calibrated on-site so that distance measurement can be performed through video.

[0051] 2. To facilitate the effective fusion of radar data and video data, radar data and video data are stored in a loop. Since the radar data cycle is inconsistent with the video data cycle, video data and radar data need to be time-fused (time calibration).

[0052] (1) Let the video data period be T V , radar data cycle is T R , T V With T R The lowest common multiple of c ; Usually T V <T R .

[0053] (2) If the video equipment and radar equipment are powered at the same time, then each time T c At the integer multiple time point, the video data and radar data are completely synchronized and can be projected directly.c The video data and radar data at the integer multiple time points are time-calibrated and then projected.

[0054] (3) Video data and radar data have their own ID numbers. The ID number of the video data closest to the time of the radar data is derived from the ID number of the radar data. Suppose the ID number of the radar data at the current moment is f r , the corresponding running time is T: T = f r ×T R ; Through T and T V The relationship between the current video data ID number f closest to the radar data time is obtained v , (Rounding operation), ID number is f r The radar data and ID number f v The video data is the time-calibrated radar data and video data.

[0055] At the same time, the ID number f can be calculated r The radar data and ID number f v The time difference ΔT between the video data is: ΔT = f v ×T V -f r ×T R The spatial fusion of video data and radar data can be corrected by ΔT.

[0056] 3. For ID number f r The radar data (time-calibrated radar data) is used to calculate the humpback vehicle target information (vehicle speed, distance, direction, length, etc.) in the radar coordinate system through the target distance and time difference ΔT, and then projected into the road plane coordinate system to obtain the humpback vehicle target information of the radar data in the road plane coordinate system.

[0057] (1) Target clustering mainly considers that the same target may have multiple reflection points, and the feature information between multiple reflection points is similar. It is necessary to cluster multiple points of the same target to obtain the target's center of mass position information and target length information.

[0058] In order to facilitate the spatial fusion of radar data and video data, it is necessary to map the target information of the radar coordinate system and the video coordinate system to the road plane coordinate system. The mapping process of the radar coordinate system and the road plane coordinate system is as follows:

[0059] The radar plane coordinate system is a coordinate system based on the center line of the radar detection direction, such as Figure 5 As shown: the horizontal line in the middle is the radar center line, and the radar horizontal angle is ±Q1 (taking the radar center line as the reference, the left angle of the radar detection direction coordinate is a negative angle, and the right angle is a positive angle).

[0060] The target information provided by the radar is based on the radar's own coordinate system, generally including distance (R), speed (V), azimuth (θ) and target length (L). f d is the Doppler frequency, V is the target speed, λ is the wavelength, and θ is the angle between the target speed vector and the extraction line in the radar transmission beam. It can be seen that its speed is the radial speed of the target in the radar plane coordinate system. Its speed Vy along the radar detection direction is related to the angle Q2, which can be obtained by the formula To obtain.

[0061] like Figure 6 As shown, usually, the radar detection direction is installed along the track direction, and the projection of the radar on the track plane coordinate is the coordinate origin. The road plane coordinate system is set as the x, y coordinate system, where the x coordinate represents the lateral distance of the road, and the y coordinate represents the longitudinal distance. Assuming that the radar installation height is H, the angle between the line drawn in the radar detection direction and the road dividing line in the x, y coordinate system is Q1, then when a slipping vehicle target passes, the radar outputs the target information as (R, V, Q3), where R is the distance of the target in the radar coordinate system, V is the speed of the target in the radar coordinate system, and Q3 is the angle between the target in the radar coordinate system and the line drawn in the radar detection, and the projection angles in the x, y coordinate system are consistent.

[0062] Then the radar coordinate system and the (x, y) plane coordinate system can be mapped by the following method:

[0063] The projection of the distance R in the radar coordinate system in the x, y coordinate system is R1, which is expressed as: H is the radar installation height.

[0064] The orientation after projection is recorded as (R x ,R y ), R is calculated by angle Q1 and angle Q3 x , R y :

[0065] R x =R1×sin(Q1+Q3)

[0066] R y =R1×cos(Q1+Q3)

[0067] Among them, the angle Q1 is the angle between the line drawn in the radar detection direction and the road dividing line in the x, y coordinate system, and the angle Q3 is the angle between the target and the line drawn in the radar detection in the radar coordinate system, and their projection angles in the x, y coordinate system are consistent. In addition, Q1 and Q3 both have positive and negative values, with negative values ​​indicating the left side of the origin along the direction of the oncoming vehicle, and positive values ​​indicating the right side of the origin along the direction of the oncoming vehicle.

[0068] The velocity after projection is V x ' y , through the angle Q3 and the radial velocity V xy The relationship calculation is:

[0069]

[0070] The radial velocity The speed V has a sign, a positive sign is the outgoing speed, and a negative sign is the incoming speed; the angle Q2 is the angle between the target radial speed and the projection of the radial speed in the x, y coordinate system. The angle Q2 between the target radial speed and the projection of the radial speed in the x, y coordinate system can be calculated through the distance R and the installation height H, that is:

[0071]

[0072] Among them, π is the ratio of circumference to circumference.

[0073] Since the radar is installed almost horizontally, the length of the radar coordinate system is not much different from the projection on the (x, y) plane coordinate system, and both can be recorded as L.

[0074] At this point, the humpback vehicle target information (R x ,R y , V x ' y , L).

[0075] 4. For the time-calibrated video data, the target information in the image coordinates is calculated through target detection and feature recognition, and then projected into the road plane coordinate system to obtain the target information of the video data in the road plane coordinate system (R' x ,R' y ,L'),R' x ,R' y is the target orientation information projected to the road plane coordinate system, and L' is the target length projected to the road plane coordinate system.

[0076] Specifically, after target detection and feature recognition, the position information of the target in the image coordinates can be obtained, and the lens imaging formula (F is the focal length of the lens, V is the phase distance, and U is the object distance) The distance information of the target. Since the video sensor and the radar sensor are designed as an integrated whole, and the lines drawn in the two are approximately coincident, the target information (R') projected from the video coordinate system target information on the road plane coordinate system can be obtained by using a method similar to that of mapping radar data to the track plane coordinate system. x ,R' y ,L').

[0077] In the embodiment of the present invention, target confidence may be introduced into the above video data to indicate the confidence level of the corresponding target information. During the subsequent feature fusion, target category and other information may also be determined based on the target confidence.

[0078] 5. Obtain (R x ,R y , V x ' y , L) and (R' x ,R' y ,L'), after which the radar data needs to be corrected by the time difference ΔT before the video data and radar data can be spatially fused. Since the radar-visual integrated device is installed along the track, the detection direction is parallel to the track. It is only necessary to correct the longitudinal component R of the azimuth information of the hump-slip vehicle target information in the radar data in the road plane coordinate system. y Correction is performed to obtain the longitudinal component R' of the corrected azimuth information yj , expressed as: R' yj =R y -V x ' y ×ΔT. R' after correction yj R' of the corresponding target y Basically the same, completing spatial fusion.

[0079] 6. After the spatial fusion of radar data and video data, feature fusion is also required to distinguish between vehicle types, locomotives, pedestrians, etc., to avoid abnormal measurement problems caused by inaccurate radar detection target features.

[0080] In the embodiment of the present invention, spatial fusion mainly refers to the fusion of target information obtained by projecting radar target information in the radar coordinate system onto the X, Y coordinate system and target information obtained by projecting target information in the video coordinate system onto the X, Y coordinate system. Feature fusion is mainly based on target features: sled vehicle models, locomotives, pedestrians, etc., because hump length measurement only focuses on sled vehicles, not locomotives and pedestrians, etc., therefore, the influence of locomotives and pedestrians on the length measurement results can be eliminated through feature fusion judgment, and the fused hump sled vehicle target information is obtained, that is, the hump sled vehicle target information of the corrected radar data in the road plane coordinate system, and the hump sled vehicle target information of the video data in the road plane coordinate system (hereinafter referred to as video detection results). 7. After the target features are fused, it is also necessary to complement the advantages and disadvantages of the video sensor and the radar sensor, so as to accurately track and locate the sled vehicle and measure the accurate idle length of the track. Its implementation process is as follows:

[0081] According to the motion state and relevant time period of the hump shunting vehicle target, the relevant information in the fused hump shunting vehicle target information is selected to calculate the track idle length, so as to realize the positioning of the hump shunting vehicle target:

[0082] (1) When the humpback vehicle target is stationary, the video detection result shall prevail, that is, the R' in the humpback vehicle target information of the video data in the road plane coordinate system y As the idle length of the track, it is judged whether it changes from moving to stationary based on the previous and next frames of radar data to avoid abnormal measurement caused by poor measurement of stationary targets by radar sensors.

[0083] (2) When the sled vehicle is in motion, the confidence of the radar data and the confidence of the video data are generally determined based on the weather conditions and the distance of the sled vehicle.

[0084] (2.1) If the current time period is daytime and the target distance is not greater than the set value (for example, 200 meters), the humpback vehicle target information in the road plane coordinate system is calculated in combination with the video data, expressed as: Among them, R L is the idle length of the track, R' y is the longitudinal component of the position information of the humpback vehicle target information in the road plane coordinate system of the video data, L' is the corresponding target length; the target distance uses the longitudinal component R of the position information of the humpback vehicle target information in the road plane coordinate system of the radar data. y .

[0085] (2.2) If it is daytime and the target distance is greater than the set value, the humpback vehicle target information in the road plane coordinate system is calculated in combination with the corrected radar data, expressed as: R' yj is the longitudinal component of the azimuth information of the hump-slip vehicle target information in the road plane coordinate system of the corrected radar data, and L is the corresponding target length.

[0086] (2.3) If the current time period is night time, the hump-slipping vehicle target information is calculated in the road plane coordinate system in combination with the corrected radar data, expressed as:

[0087] Figure 7 The overall process of the above data fusion is presented. By continuously repeating steps 2 to 7, the positioning and identification of the vehicles slid in the hump track can be achieved, thereby realizing the accurate measurement of the idle length of the hump track.

[0088] The above solution provided by the embodiment of the present invention uses video images and radar data to perform data fusion to identify and locate the shunted vehicles in the hump track, which can replace the existing hump length measurement equipment and achieve the following beneficial effects:

[0089] 1. Improve measurement accuracy.

[0090] 2. It can solve the problem of inaccurate length measurement data caused by severe oxidized conjunctiva corrosion of rails, poor contact between light vehicle wheels and rails, and leakage of the ballast bed.

[0091] 3. The parked vehicle can be identified based on the video data, so as to solve the problem of being unable to calibrate when the measuring section is occupied according to the standard target.

[0092] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above embodiments can be implemented by software, or by means of software plus necessary general hardware platforms. Based on such understanding, the technical solutions of the above embodiments can be embodied in the form of software products, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0093] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for positioning a humpback vehicle based on data fusion, characterized in that: include: Collect radar data and video data, combine the collection period of radar data and video data, perform time calibration on the two types of data, and calculate the time difference; For the time-calibrated radar data and video data, the hump-slipping vehicle target information is detected and projected into the road plane coordinate system to obtain the hump-slipping vehicle target information of the radar data in the road plane coordinate system; For the video data, target information is detected and projected into the road plane coordinate system to obtain the target information of the video data in the road plane coordinate system; After correcting the hump-slipping vehicle target information of the radar data in the road plane coordinate system in combination with the time difference, completing spatial fusion with the corresponding target information of the video data in the road plane coordinate system, and then performing consistent feature fusion to obtain fused hump-slipping vehicle target information; The fused hump-sliding vehicle target information is used to locate the hump-sliding vehicle target.

2. A method for positioning a humpback vehicle based on data fusion according to claim 1, characterized in that: Combining the acquisition cycles of radar data and video data, time calibration of the two types of data includes: The video data cycle is T V , radar data cycle is T R , T V With T R The lowest common multiple of is T c ; Every time T c At the time point that is an integer multiple of T, the video data and radar data are completely synchronized and projected directly. c Time calibration is performed on the video data and radar data at the integer multiple time points; The video data and radar data have their own ID numbers. The ID number of the video data closest to the radar data time is derived from the ID number of the radar data. Suppose the ID number of the radar data at the current moment is f r , the corresponding running time is T: T = f r ×T R ; Through T and T V The relationship between the current video data ID number f closest to the radar data time is obtained v , The above formula is a rounding operation, and the ID number is f r The radar data and ID number f v The video data is the time-calibrated radar data and video data.

3. A method for positioning a humpback vehicle based on data fusion according to claim 2, characterized in that: The formula for calculating the time difference is: ΔT=f v ×T V -f r ×T R Wherein, ΔT is the time difference.

4. The method for positioning a humpback vehicle based on data fusion according to claim 1, characterized in that: For the time-calibrated radar data, the hump-out vehicle target information in the radar coordinate system is calculated by target distance and trajectory tracking, and then projected into the road plane coordinate system to obtain the hump-out vehicle target information of the radar data in the road plane coordinate system; Among them, the hump-release vehicle target information in the radar coordinate system calculated by target distance and trajectory tracking includes: distance R, speed V, azimuth θ and target length L; The projection process is as follows: Assume that the road plane coordinate system is the x, y coordinate system, and the distance after projection is R1, which can be expressed as: H is the radar installation height; The orientation after projection is recorded as (R x ,R y ), R is calculated by angle Q1 and angle Q3 x , R y : R x =R1×sin(Q1+Q3) R y =R1×cos(Q1+Q3) Among them, the angle Q1 is the angle between the line drawn in the radar detection direction and the road dividing line in the x, y coordinate system, and the angle Q3 is the angle between the target and the line drawn in the radar detection in the radar coordinate system, and their projection angles in the x, y coordinate system are consistent; The velocity after projection is V x ' y , through the angle Q3 and the radial velocity V xy The relationship calculation is: The radial velocity Angle Q2 is the angle between the target radial velocity and the projection of the radial velocity in the x, y coordinate system; The target length L after projection remains unchanged; At this point, the humpback vehicle target information (R x ,R y , V x ' y , L).

5. A method for positioning a hump-type vehicle based on data fusion according to claim 1 or 4, characterized in that: For the time-calibrated video data, the target information in the image coordinates is calculated through target detection and feature recognition, and then projected into the road plane coordinate system to obtain the target information of the radar data in the road plane coordinate system (R' x ,R' y ,L'),R' x ,R' y is the target orientation information projected to the road plane coordinate system, and L' is the target length projected to the road plane coordinate system.

6. A method for positioning a hump-type vehicle based on data fusion according to claim 1 or 4, characterized in that: The correction of the hump-out vehicle target information of the radar data in the road plane coordinate system in combination with the time difference comprises: The time difference is recorded as ΔT, and the longitudinal component R of the azimuth information of the hump-slip vehicle target information in the radar data in the road plane coordinate system is calculated. y Correction is performed to obtain the longitudinal component R' of the corrected azimuth information yj , expressed as: R' yj =R y -V x ' y ×ΔT.

7. The method for positioning a humpback vehicle based on data fusion according to claim 1, characterized in that: The method of positioning the hump shunting vehicle target by using the fused hump shunting vehicle target information includes: According to the movement state and relevant time period of the hump shunting vehicle target, the relevant information in the fused hump shunting vehicle target information is selected to calculate the idle length of the track, so as to realize the positioning of the hump shunting vehicle target; wherein, the fused hump shunting vehicle target information includes: the hump shunting vehicle target information of the corrected radar data in the road plane coordinate system, and the hump shunting vehicle target information of the video data in the road plane coordinate system.

8. The method for positioning a humpback vehicle based on data fusion according to claim 7, characterized in that: Also includes: When the hump-sliding vehicle target is stationary, the hump-sliding vehicle target information in the road plane coordinate system of the video data is used to determine the idle length of the track, and whether it changes from moving to stationary is judged based on the previous and next frames of radar data.

9. The method for positioning a humpback vehicle based on data fusion according to claim 7, characterized in that: Also includes: When the hump release vehicle target is in motion; If the current time period is daytime and the target distance is not greater than the set value, the humpback vehicle target information in the road plane coordinate system is calculated in combination with the video data, expressed as: The target distance uses the longitudinal component R of the orientation information of the humpback vehicle target information in the road plane coordinate system of the radar data. y ; If it is daytime and the target distance is greater than the set value, the humpback vehicle target information in the road plane coordinate system is calculated in combination with the corrected radar data, expressed as: If the current time period is night time, the hump-slipping vehicle target information is calculated in the road plane coordinate system in combination with the corrected radar data, expressed as: Among them, R L is the idle length of the track, R' yj is the longitudinal component of the azimuth information of the hump-slipping vehicle target information in the road plane coordinate system of the corrected radar data, and L is the corresponding target length; R' y is the longitudinal axis component of the orientation information of the humpback vehicle target information in the video data in the road plane coordinate system, and L' is the corresponding target length.