Tunnel global tracking method based on old camera re-identification technology

Abstract

The invention belongs to the technical field of intelligent traffic, particularly relates to a tunnel global tracking method based on an old camera re-identification technology, and aims to accurately realize global tracking and trajectory generation of a tunnel traffic target in a low-cost manner. The method comprises the steps of receiving an RSTP data stream, processing the RSTP data stream through a detection algorithm to obtain a sensing result, and feeding back the sensing result to the image tracking module. And processing a time sequence sensing result, adding vehicle speed information and a vehicle ID, and generating a tracking result including the vehicle ID, a target frame, a vehicle speed, a vehicle type, a color, a timestamp, a base station number and a camera number. On the basis of vehicle pixel picture data, upstream sample library candidates are matched for a current point position sample library target, if matching succeeds, the target inherits the global ID, if matching fails, the point position ID is reserved, and the target is output to a downstream point position and a simulation platform after the sample library is updated. And the trajectory generation post-processing module constructs a target matching relationship between adjacent camera sections through a frame insertion mode according to a feature re-identification result, and generates a tunnel global complete trajectory chain.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a tunnel full-area tracking method based on reused camera re-identification technology. Background Art

[0002] With the digital transformation of urban transportation and the increasing demand for tunnel safety supervision, real-time monitoring of tunnel traffic conditions and event warnings have become key technology areas. Currently, tunnel safety monitoring typically deploys multiple sensors, such as lidar, millimeter-wave radar, and video surveillance, to achieve vehicle detection, trajectory tracking, and abnormal event identification. However, these multi-sensor solutions are subject to high construction costs and complex deployment. This is especially true for existing tunnels, where adding new sensors requires infrastructure modifications, making implementation difficult and uneconomical. Summary of the Invention

[0003] The purpose of the present invention is to provide a tunnel full-area tracking method based on reused camera re-identification technology, aiming to accurately and cost-effectively achieve full-area tracking and trajectory generation of tunnel traffic targets.

[0004] To achieve the above objectives, the present invention employs the following technical solution: The present invention provides a tunnel global tracking method based on re-identification technology using repurposed cameras, comprising: S1: The image detection module receives the RSTP data stream and processes it through a detection algorithm to obtain a perception result containing the vehicle's target frame, vehicle model, color, timestamp, base station ID, and camera ID, which is then fed back to the image tracking module. The RSTP data stream is the remote switching port data stream that transmits tunnel image data. S2: The image tracking module processes the time-series perception result, adds vehicle speed information and vehicle ID, and generates a tracking result containing the vehicle ID, target frame, speed, vehicle model, color, timestamp, base station ID, and camera ID. S3: The feature re-identification module matches the target in the current point sample library with candidates in the upstream sample library based on the vehicle pixel image data. If the match is successful, the target inherits the global ID. If the match fails, the current point ID is retained. After updating the sample library, the global ID is output to downstream points and the simulation platform. The global ID uniquely identifies the vehicle within the entire tunnel. S4: The trajectory generation post-processing module constructs the target matching relationship between adjacent camera sections through interpolation based on the feature re-identification results, and generates a complete trajectory chain in the entire tunnel.

[0005] The tunnel full-area tracking method based on the reused camera re-identification technology provided in the embodiment of the present application avoids adding expensive sensors such as lidar and millimeter-wave radar by receiving the RSTP data stream (remote switching port data stream) of the reused camera, directly utilizing the original monitoring equipment in the tunnel, and significantly reducing the construction cost. The image detection module is used to collect multi-dimensional data to provide rich visual features for subsequent tracking and re-identification. The image tracking module constructs the target motion trajectory within a single camera by processing the time series perception results and adding the vehicle speed and vehicle ID, solving the trajectory breakage problem caused by camera time asynchrony in the prior art. The global ID allocation mechanism of the feature re-identification module ensures the identity consistency of the vehicle throughout the tunnel, builds a unified identification system for the global twin, and supports upper-layer applications such as event detection and traffic flow analysis.

[0006] In some embodiments, before step S3, it also includes: the section trigger module intercepts the fixed-time detection results based on the timestamp, retains the targets within the section interval threshold, crops the pixel image through the target detection frame, and outputs a sample library containing vehicle ID, target frame, pixel image, vehicle model, speed, color, timestamp, base station number and camera number.

[0007] In some embodiments, the tunnel global tracking method further includes: the sample library screening algorithm module screening upstream sample library candidates by timestamp and vehicle speed information, and outputting a screening result with the current sample library as the query and the upstream candidates as the gallery.

[0008] In some embodiments, the tunnel full-area tracking method further includes: a logic re-identification module classifies samples based on vehicle type information, narrows the matching range, and updates the sample library.

[0009] In some embodiments, the RSTP data stream comes from a reused camera or a newly built camera, where the reused camera is an existing camera deployed in the tunnel and the newly built camera is a newly installed image acquisition device.

[0010] In some embodiments, the image tracking module achieves target matching tracking by analyzing changes in position, shape, and color features of the vehicle in adjacent frame images.

[0011] In some embodiments, the feature re-identification module assigns a global ID to the target at the first cross-section perception point for the first time after detection and tracking, and directly outputs the sample library to downstream points and simulation platforms.

[0012] In some embodiments, the trajectory generation post-processing module inserts virtual frames at missing time points based on the time interval between adjacent frames and the vehicle speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1This is a flowchart of a tunnel full-area tracking method based on reused camera re-identification technology provided in an embodiment of the present application. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described 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 making creative efforts are within the scope of protection of the present invention.

[0015] In the description of the invention, it should be understood that the terms "upper," "lower," "left," "right," "front," "back," "inner," "outer," and the like, indicating directions or positional relationships, are based on the directions or relative positional relationships shown in the accompanying drawings and are intended solely to facilitate the description of the invention and simplify the description. They do not indicate or imply that the devices or components referred to must have a specific direction, be constructed, or operate in a specific direction. Therefore, they should not be construed as limitations on the invention. Unless otherwise specified, the above-mentioned directions may be flexibly set in actual application, provided that the relative positional relationships shown in the accompanying drawings are met.

[0016] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0017] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connected," and "communicated" should be understood broadly. For example, they may refer to fixed connections, detachable connections, or integral connections. They may be directly connected, indirectly connected through an intermediary, or internally connected between two components. Those skilled in the art will understand the specific meanings of these terms in the present invention based on specific circumstances.

[0018] In embodiments of the present invention, the terms "comprises," "comprising," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not preclude the presence of other identical elements in the process, article, or apparatus comprising the element.

[0019] In the embodiments of the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present invention should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0020] As mentioned in the background technology section, leveraging existing cameras deployed in tunnels (i.e., existing surveillance cameras) to achieve low-cost, efficient, and comprehensive traffic awareness has become a pressing technical challenge for the industry. However, current purely visual tunnel tracking technologies face challenges with camera time synchronization, blind spots in detection coverage, and difficulty ensuring 3D positioning accuracy.

[0021] In view of this, the embodiment of the present application provides a tunnel full-area tracking method based on the reuse of old camera re-identification technology. For example, Figure 1 The tunnel global tracking method includes: S1: The image detection module receives the RSTP data stream and processes it through the detection algorithm to obtain the perception results including the vehicle target frame, model, color, timestamp, base station number and camera number, and feeds it back to the image tracking module.

[0022] The RSTP data stream is the remote switching port data stream that transmits image data within the tunnel. The image detection module, as the first execution module, receives the RSTP data stream transmitted by the tunnel camera and extracts multi-dimensional features, including the vehicle target frame, vehicle model, color, timestamp, base station ID, and camera ID, to provide raw data support for subsequent tracking and re-identification.

[0023] It should be understood that RSTP data streams come from either reused cameras or newly installed cameras. Reused cameras are existing cameras deployed in the tunnel, while newly installed cameras are image acquisition devices. Vehicle appearance features provide a matching basis for cross-camera re-identification (such as the feature re-identification module), while timestamps, base station numbers, and camera numbers are used to mark the time and spatial attributes of data acquisition, which helps with time alignment and spatial coordinate mapping to solve the problems of difficult camera time synchronization and difficult to ensure three-dimensional positioning accuracy in traditional technologies.

[0024] S2: The image tracking module processes the temporal perception results, adds vehicle speed information and vehicle ID, and generates tracking results containing vehicle ID, target frame, speed, vehicle model, color, timestamp, base station number, and camera number.

[0025] By processing perception results through a time-matching algorithm, discrete detection results can be concatenated into a continuous trajectory, helping to address the blind spots in detection coverage found in traditional technologies. Vehicle speed calculation combines target displacement with timestamps, eliminating the need for additional sensors and meeting low-cost requirements.

[0026] S3: The feature re-identification module matches the target in the current point sample library with the candidate in the upstream sample library based on the vehicle pixel image data. If the match is successful, the target inherits the global ID. If the match fails, the current point ID is retained and the sample library is updated and output to the downstream points and simulation platform.

[0027] Among them, the global ID is the identity identifier that uniquely identifies the vehicle within the entire tunnel.

[0028] After a successful match, the global ID is inherited; if a match fails, the local ID is retained. This ensures trajectory continuity while avoiding ID confusion. This provides accurate and unified basic data for higher-level applications such as event detection and traffic flow analysis, improving the real-time and reliability of intelligent tunnel monitoring. It also addresses the perception blind spots caused by camera layout, ensuring unique vehicle identification throughout the tunnel, building a continuous trajectory chain, and enhancing global twinning. Furthermore, the system is suitable for mixed scenarios involving both reused and newly built cameras.

[0029] As a possible implementation method, before step S3, it also includes: the section trigger module intercepts the fixed-time detection results based on the timestamp, retains the targets within the section interval threshold, crops the pixel image through the target detection frame, and outputs a sample library containing vehicle ID, target frame, pixel image, vehicle model, speed, color, timestamp, base station number and camera number.

[0030] The cross-section triggering module processes and outputs a sample library based on the output of the image tracking module. The timestamp determines the data's validity range, and the vehicle's speed and camera ID are used to calculate the target's position within the cross-section. The output of the cross-section triggering module serves as the input to the feature re-identification module. By cropping the pixel image using the object detection box, the module preserves vehicle details (such as the license plate and body texture). This provides sufficient detail for the feature re-identification module while filtering out invalid data, optimizing the computational burden and improving efficiency of the entire method.

[0031] In some embodiments, the sample library screening algorithm module uses timestamp and vehicle speed information to filter upstream sample library candidates, outputting screening results with the current sample library as the query set (query) and the upstream candidates as the matching set (gallery). The logical re-identification module classifies samples based on vehicle model information, narrows the matching range, and updates the sample library.

[0032] The upstream point sample library: Output from the cross-section trigger module of the upstream camera, including information such as the vehicle's global ID, timestamp, and speed. The current point sample library: Output from the cross-section trigger module of the current camera, serving as the basis for the query. Through cross-validation of timestamp and speed information, the difference between the timestamp of the upstream sample and the current sample is used to calculate the traveled distance, combined with the vehicle speed. This is then compared with the actual camera distance to screen out targets with logically consistent locations.

[0033] S4: The trajectory generation post-processing module constructs the target matching relationship between adjacent camera sections through interpolation based on the feature re-identification results, and generates a complete trajectory chain in the entire tunnel.

[0034] By constructing target matching relationships between adjacent camera sections through interpolation, the detection blind spots of adjacent camera sections can be effectively filled, thereby further improving the coverage effect and solving the problem of track interruption caused by camera layout.

[0035] As a possible implementation method, the trajectory generation post-processing module inserts virtual frames at the missing time points based on the time interval between adjacent frames and the vehicle speed.

[0036] Dynamic interpolation based on time intervals and vehicle speed ensures smooth and continuous trajectories, eliminates trajectory jumps caused by detection errors, and helps improve positioning accuracy.

[0037] This application builds a stable and smooth global digital twin system by deploying a feature re-identification algorithm and combining existing or newly added video cameras in the tunnel. The tunnel global tracking method based on the reused camera re-identification technology provided in the embodiment of this application has excellent generalization capabilities and can adapt well to both new and old equipment and cameras that are far and near. At the same time, its deployment process is simple, ensuring the smooth implementation of the digital twin effect and paving the way for the implementation of subsequent intelligent applications such as event detection.

[0038] In the description of this specification, specific features, structures, materials or characteristics may be combined in an appropriate manner in any one or more embodiments or examples.

[0039] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A tunnel full-area tracking method based on re-identification technology using a re-used camera, characterized in that: include: S1: The image detection module receives the RSTP data stream, processes it through a detection algorithm, obtains a perception result including the vehicle target frame, vehicle model, color, timestamp, base station number, and camera number, and feeds it back to the image tracking module. The RSTP data stream is the remote switching port data stream that transmits image data in the tunnel. S2: The image tracking module processes the temporal perception results, adds vehicle speed information and vehicle ID, and generates tracking results containing vehicle ID, target frame, speed, vehicle model, color, timestamp, base station number, and camera number; S3: The feature re-identification module matches the target in the current point sample library with the candidate in the upstream sample library based on the vehicle pixel image data. If the match is successful, the target inherits the global ID. If the match fails, the current point ID is retained. After updating the sample library, it is output to the downstream points and the simulation platform. The global ID is the unique identifier that identifies the vehicle within the entire tunnel. S4: The trajectory generation post-processing module constructs the target matching relationship between adjacent camera sections through interpolation based on the feature re-identification results, and generates a complete trajectory chain in the entire tunnel.

2. The tunnel full-area tracking method based on reused camera re-identification technology according to claim 1 is characterized in that: Before step S3, it also includes: a section trigger module intercepts a fixed-time detection result based on a timestamp, retains the target within the section interval threshold, crops the pixel image through the target detection frame, and outputs a sample library containing vehicle ID, target frame, pixel image, vehicle model, speed, color, timestamp, base station number and camera number.

3. The tunnel full-area tracking method based on reused camera re-identification technology according to claim 2 is characterized in that: The tunnel global tracking method further includes: The sample library screening algorithm module screens upstream sample library candidates by timestamp and vehicle speed information, and outputs the screening results with the current sample library as the query and the upstream candidates as the gallery.

4. The tunnel full-area tracking method based on reused camera re-identification technology according to claim 3 is characterized in that: The tunnel full-area tracking method further includes: a logic re-identification module classifies samples based on vehicle type information, narrows the matching range and updates the sample library.

5. The tunnel full-area tracking method based on reused camera re-identification technology according to claim 1 is characterized in that: The RSTP data stream comes from a reused camera or a newly built camera, wherein the reused camera is an existing camera deployed in the tunnel, and the newly built camera is a newly installed image acquisition device.

6. The tunnel full-area tracking method based on reused camera re-identification technology according to claim 1 is characterized in that: The image tracking module realizes target matching and tracking by analyzing the changes in the position, shape and color characteristics of the vehicle in adjacent frame images.

7. The tunnel full-area tracking method based on reused camera re-identification technology according to claim 1 is characterized in that: The feature re-identification module assigns a global ID to the target at the first cross-section perception point for the first time after detection and tracking, and directly outputs the sample library to downstream points and the simulation platform.

8. The tunnel full-area tracking method based on reused camera re-identification technology according to claim 1 is characterized in that: The trajectory generation post-processing module inserts virtual frames at missing time points based on the time interval between adjacent frames and the vehicle speed.

Images