Method for detecting marker lights

By detecting active warning lights in the environment surrounding the vehicle and using a camera system and neural network to identify emergency vehicles, the problem of unreliable identification of emergency vehicles in existing technologies has been solved, improving traffic safety and driver reaction speed.

CN112166433BActive Publication Date: 2024-11-05CONTINENTAL AUTONOMOUS MOBILITY GERMANY GMBH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN201980033579.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2018-06-12
Filing Date
2019-05-08
Publication Date
2024-11-05
Estimated Expiration
2039-05-08

AI Technical Summary

Technical Problem

Existing technology cannot reliably identify vehicles used for emergency missions, preventing drivers from taking timely special actions to ensure traffic safety.

Method used

By detecting active marker lights in the environment surrounding the vehicle, the system uses a camera system to capture images and perform bright spot detection, cropping, and classification. A neural network classifier is used to identify light spots of specific colors, and a counter threshold control ensures accurate identification. The information is then provided to the driver or an assistance system.

Benefits of technology

It enables reliable identification of vehicles used for emergency missions, improves driver reaction speed and traffic safety, and allows the assistance system to take autonomous measures.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112166433B_ABST
    Figure CN112166433B_ABST
Patent Text Reader

Abstract

The invention relates to a method for detecting a marker light (4) in the surroundings of a vehicle (5), the method comprising the following steps: photographing (S1) a sequence of camera images (1) of the surroundings with the aid of a camera system (6) of the vehicle (5), detecting (S2) bright light points in the photographed camera image (1), cutting out (S3) a region (3) in the camera image (1) containing the detected bright light points, classifying (S4) the cut-out region (3), and classifying (S5) the bright light points as marker lights (4) if the classified bright light points are repeatedly detected, thereby exceeding a counter threshold.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a method for detecting a marker light. Background Art

[0002] From the prior art, for example, methods for identifying traffic lights are known, wherein it is also possible to determine which color the traffic light shows. In addition, it is also known to identify other road users based on their headlights and / or taillights.

[0003] A disadvantage of the known method is that, for example, only the presence of a vehicle can be determined, but no other information about the vehicle can be obtained. Emergency vehicles such as police cars cannot be identified. However, such emergency vehicles require special actions by the driver under certain conditions. Summary of the invention

[0004] Therefore, the object of the present invention is to provide a method which can ensure reliable identification of vehicles used for emergency tasks.

[0005] The object of the present invention is solved by the technical solution of claim 1. Other technical solutions can be found in the dependent claims.

[0006] The first thing to consider is that emergency vehicles are most reliably identified by active marker lights. For example, blue light is characteristic of police cars, fire trucks, ambulances, etc. If active marker lights are detected, the presence of an emergency vehicle can be directly inferred.

[0007] Therefore, according to the present invention, a method for detecting marker lights in the surrounding environment of a vehicle is proposed, and the method has the following steps:

[0008] - a sequence of camera images of the surrounding area recorded by means of the vehicle's camera system,

[0009] - detect bright spots in the captured camera image,

[0010] - cut out the area of ​​the camera image that contains the detected bright spot,

[0011] - classify the clipping areas,

[0012] If a classified bright light point is detected several times in succession and thus exceeds a threshold value of a counter, the classified bright light point is classified as a marker light.

[0013] The term “marker lights” here refers to so-called rotating marker lights, such as are used by the German police, for example, or front marker lights on the radiator grilles of motorcycles or ambulances, for example, for emergency missions.

[0014] The camera system may be a monocular camera, a stereo camera or a panoramic surround view camera system.

[0015] Bright spots are understood to be, for example, the 10 to 50 brightest areas in a particular color channel. If adjacent areas are pixels of the same brightness, they are considered one area and are only entered into the classifier once, thus only being counted once in the 10 to 50 brightest areas. This prevents all bright areas from being assigned to one lamp.

[0016] The counter threshold used here is the number of detected classified light points that must be present to infer the presence of marker lights. It is important here that only correspondingly classified light points are taken into account, since only one or more specific colors are relevant here. This can prevent, for example, a turn signal from being mistakenly identified as a marker light.

[0017] In a particularly preferred embodiment, a neural network is used for the classification.

[0018] The neural network particularly preferably forms a classifier for light, in particular colored light. The colored light is particularly preferably blue light. In this embodiment, the brightest areas or maxima in a specific channel are particularly taken into account, in the blue channel the blue light is taken into account. For other colors, such as red, the maximum value in the red channel is taken into account. A corresponding classifier can be used for each corresponding color. Different classifiers can also be used in parallel in order to reliably identify a combination marker light consisting of blue and red, such as is used in the United States. It is also conceivable to train the classifier with a plurality of categories so that a single classifier is used for different colors.

[0019] At least one of the classifiers preferably has an output, whereby one color of light is output. This is particularly advantageous when a plurality of classifiers are used for different colors.

[0020] In a further step, information is preferably provided to the driver and / or a driver assistance system by means of a marker light. For example, the information can be provided to the driver optically, for example, by means of a head-up display or a display arranged in or on the instrument panel. As an alternative or in addition, an audio prompt can also be given. Providing information to the driver is beneficial because the driver can, if necessary, perform corresponding driving maneuvers based on this information and / or increase his attention to potentially unusual driving behavior of other road users. Furthermore, providing information to a driver assistance system is beneficial because in this way, the vehicle can, for example, autonomously take measures to form an emergency lane / give-way lane for vehicles in emergency situations.

[0021] In a preferred design, the counter is incremented after each detection of a classified light spot, and the counter is decremented if no classified light spot is detected. Subsequently, the counter continues to increase until a certain number of qualified detections is reached. A qualified detection is a detection that classifies an emergency mission vehicle as an emergency mission marker light. On the other hand, if, for example, the emergency mission marker lights of an emergency mission vehicle are detected only once, the counter first increases, but if the counter does not exceed the threshold value subsequently, it is decremented again. The initial value is, for example, 1, which is exceeded from the first detection and is therefore also increased. The maximum value may be, for example, 3. Therefore, the counter is increased twice from 1 to 3, and the fourth detection is decisive for the classification of the detection as a marker light. This design configuration method is particularly beneficial because in this way the output of erroneous detection results can be suppressed. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Further advantageous embodiments are an integral part of the drawings.

[0023] in:

[0024] Figure 1 is a schematic diagram of an exemplary traffic situation;

[0025] Figure 2 is a schematic flow chart of a preferred embodiment of the method. DETAILED DESCRIPTION

[0026] Figure 1 A schematic diagram of an exemplary traffic situation is shown. In this traffic situation, there is a vehicle 5 on lane FS2, another vehicle 2a next to the vehicle 5 on lane FS3, and another vehicle 2b on lane FS2 behind the vehicle 5 on road F. Here, the three vehicles are traveling in the same direction of movement R1. In addition, the lanes include an emergency lane FS1. Here, the vehicle has a camera 6, which has a detection range E and an aperture angle α for rearward detection, for example, as a fixed component of a panoramic camera system. The other vehicle 2b includes a marker light 4, which is designed as a rotating marker light, for example. The other vehicle 2b is detected by means of the camera 6 of the vehicle 5, and bright light points are detected in the image captured in this way to determine whether the marker light 4 is present. If the active marker light 4 of the other vehicle 2b is detected, the vehicle can, for example, switch to the emergency lane FS1 to make way for the other vehicle 2b that is classified as an emergency vehicle.

[0027] Figure 2A schematic flow chart of a preferred embodiment of the method is shown. In step S1, a camera image sequence 1 of the surrounding area is recorded by means of a camera system 6 of the vehicle 5. In step S2, bright light points or areas 3 are detected in the recorded camera image 1. In a subsequent step S3, the area with the bright light points is cut out in the camera image 1. The cut-out area 3 is then classified in step S4. In step S5, the bright light points are classified as marker lights 4 if the classified bright light points are detected several times in succession in time and exceed a threshold value. The marker lights 4 are located in front of the vehicle 2b in the camera image recorded here.

[0028] Reference numerals list

[0029] 1 Camera image

[0030] 2a, 2b Other vehicles

[0031] 3 Regions

[0032] 4 Marker lights

[0033] 5 This car

[0034] 6 Camera System

[0035] E Detection area

[0036] F Road

[0037] FS1-FS3 lanes

[0038] R1 The direction of movement of the vehicle

[0039] S1-S5 Method Steps

Claims

1. A method for detecting a marker light (4) in the surroundings of a vehicle (5), the method comprising the following steps: A sequence of camera images (1) of a surrounding area is recorded (S1) by means of a camera system (6) of the vehicle (5), and bright light points in the recorded camera images (1) are detected (S2), wherein: Bright spots are the 10 to 50 brightest areas in a particular color channel. Cut out (S3) the area (3) with the detected bright light point in the camera image (1), The clipped region (3) is classified (S4), If the classified bright light point is detected repeatedly and exceeds the counter threshold, the bright light point is classified (S5) as a marker light (4) and the counter is incremented until a predetermined number of qualified detections of the classified bright light point is reached.

2. The method according to claim 1, in, Use neural networks for classification.

3. The method according to claim 2, in, The neural network constitutes at least a classifier for light.

4. The method according to claim 3, in, The neural network forms at least a classifier for colored light.

5. The method according to claim 3, in, At least one classifier has an output whereby a color of the light is output.

6. The method according to claim 1, in, In a further step, information is provided to the driver and / or a driver assistance system via the indicator light (4).

7. The method according to claim 1, in, After each detection of a classified light point, the counter is incremented, and if no classified light point is detected, the counter is decremented.