Method and device for recognizing traffic signs for a vehicle

This application is the U.S. National Phase Application of PCT/DE2013/200238, filed Oct. 21, 2013, which claims priority to German Patent Application No. 10 2012 110 595.2, filed Nov. 6, 2012, the contents of such applications being incorporated by reference herein.

Radio-based systems for traffic sign recognition are state of the art. A traffic sign transmits its meaning and, if appropriate, its position by radio to a radio receiver within the vehicle.

An alternative approach to traffic sign recognition has already been implemented in production vehicles. Traffic signs are recognized by means of a camera sensor system within the vehicle. Such systems can be supported by a navigation system including an electronic map. Traffic sign recognition (TSR) by means of a camera system is based on two essential steps. Step 1 is the search for circular structures (e.g. speed limits) or rectangular structures (e.g. town signs). When suitable candidates have been detected, they are then classified in a subsequent step. For example, numbers are identified using pattern recognition.

An aspect of the present invention is to improve the function of a system for traffic sign recognition which is based on data of a camera sensor system.

It is an aspect of the invention that traffic sign recognition based on a camera system be supported by information of a radar or LiDAR sensor system. Sensor systems which can be used for this purpose include, in particular, those used for an ACC function or an emergency braking function. Based on the data of a radar sensor or LiDAR sensor, at least one element of the following information

the presence of a traffic sign or the size of a traffic sign position of a traffic sign distance and/or orientation of a traffic sign bridge-like structure, in particular a gantry position or type of the vehicle ahead, in particular a truck position of lane delimiting structures

is obtained and used for traffic sign recognition.

In a preferred embodiment of the invention, the at least one piece of information obtained from radar or LiDAR data is used to check the plausibility of data provided by the camera sensor.

When a traffic sign or a potential traffic sign is detected by means of radar, the TSR unit can be informed of this fact. The information (presence, distance, size, orientation, etc.) can then be used there to make a plausibility check.

Another positive embodiment of the invention includes image processing for the analysis of camera data in such a manner that an area where information according to a)-e) has been recorded is specifically searched for traffic signs. This can e.g. be done by a more detailed analysis of the image data with an increased (computational) intensity, or by means of a targeted search for combined structures or patterns or colors which have been stored in a memory and are typical of the shape and contents of traffic signs. In particular, the stored templates can be selected according to country. The information relating to the country can be provided by a navigation device or via radio systems (c2x, broadcast, traffic reporting) or by a sensor which records the surrounding environment. The latter may e.g. be a camera sensor which is designed for traffic sign recognition. This approach ensures that no traffic sign is overlooked.

Another positive embodiment of the invention includes image processing for the analysis of camera data in such a manner that only an area where information according to a)-e) has been recorded is searched for traffic signs. This approach saves computation time during the analysis of the image data of a camera system, or, as the computation time which is available for data processing is limited, the camera data relating to traffic signs in the specified areas can be analyzed better by making more intensive use of the computation time.

In another positive embodiment of the invention either of the elements of the information under c) or d) is analyzed to determine the area where a traffic sign is valid. This is, in particular, done when there are several traffic lanes in the surrounding environment of the vehicle. A radar or LiDAR sensor system can be used to detect lane delimiting structures (e.g. guard rails) g). Traffic signs detected by a radar or LiDAR sensor system can be allocated to these structures in terms of space. This can, in particular, serve as a feature for a plausibility check which is aimed at disregarding or confirming potential traffic signs, depending on the position of these potential traffic signs relative to the lane delimiting structures. In particular, distance or position information can be used for such an allocation. In particular in the case of multi-lane carriageways, junctions, parallel roads, etc., the position of the traffic sign decides to which lane, road, etc. it belongs. If the location of the traffic sign is known, one can determine its relevance to one's own traffic lane. For example, distant signs can be disregarded if they do not belong to the own traffic lane.

A preferred embodiment of the invention includes image processing for the analysis of camera data in such a manner that an area where information according to f) has been recorded is not searched for traffic signs.

A radar or LiDAR sensor system is used to detect vehicles ahead, in particular trucks, and this information is provided to the traffic sign recognition unit. This is e.g. useful in the case of vehicles which have information in the form of speed limits on their rear end. Such information is not valid for other road users.

Preferably, the presence of a traffic sign is detected on the basis of data provided by a radar or LiDAR system, depending on at least one of the following parameters:

Position of the object; traffic signs are, in particular, arranged on the roadside or above the carriageway or the traffic lane. In addition or as an alternative, the position information can be correlated with navigation data. It can e.g. be assumed that there are a higher number of traffic signs near crossings. Distance, in particular to determine the relevance of a traffic sign to one's own traffic lane. Intensity of the radiation reflected on an object or parameters which can be deduced therefrom, such as the radar or LiDAR backscatter cross section; due to the material they are made of (usually metal), traffic signs are good reflectors, i.e. they can be detected from the fact that the backscatter cross section or the intensity of reflection must be above a predefined threshold or lie within a predefined value range. Relative speed; traffic signs are stationary objects, so that they can be detected from the fact that their relative speed must be the same as the vehicle speed or their relative speed to other non-moving objects, e.g. guard rails, must be zero. Dimensions of the object; as traffic signs have standard sizes, only an object whose dimensions lie within a predefined value range can be a traffic sign. The predefined value range can, in particular, depend on further information, e.g. the position of the traffic signs. In particular, gantries are stationary objects which have relatively large dimensions and are located above the carriageway.

In a positive embodiment of the invention, objects are recognized as potential traffic signs on the basis of radar or LiDAR data; in addition, these objects are allocated a probability value which indicates the probability that the relevant object could be a traffic sign.

Preferably, information relating to the position of the object is, in addition, correlated with information provided by a navigation system in order to determine the probability. In particular, there is a higher probability of traffic signs near crossings.