Methods and a server for identifying locations on roads with an increased risk of collisions

Abstract

The invention relates to a method for identifying locations on roads (100, 101) with an increased risk of collision, comprising an image processing step and a transmission step, wherein in the image processing step an evaluation unit divides an aerial image on which at least one road (100, 101) and adjacent buildings (300) are depicted into geometric patterns, and at least one building (300) is identified in a pattern.For each of the identified buildings (300), a footprint is calculated by the evaluation unit. If a certain number of the buildings (300) fall below a predefined footprint threshold, a number of buildings (300) with a predefined footprint shape are determined by the evaluation unit. Each open space (400) in the aerial image not covered by an identified building (300) is identified, and each of these open spaces (400) is assigned a line (500) by the evaluation unit running along a longitudinal dimension of the respective open space (400). Finally, an intersection (600) of at least one of the lines (500) with a road (100, 101) is identified as a location with an increased risk of collision.In the transmission step, a warning regarding this location is transmitted via a communication unit to a vehicle's navigation system, and the navigation system indicates if the vehicle is moving towards this location.

Description

[0001] The invention relates to a method and a server for identifying locations on roads with an increased risk of collision.

[0002] Densely populated areas exist in many parts of the world. Particularly in developing countries, densely populated areas with inadequate infrastructure and generally haphazard settlement structures are often referred to as "slums." These areas sometimes border highways or other busy roads, as other construction is unattractive due to the proximity of these busy roads. Because of this proximity, people often enter or cross the roads illegally, posing a significant danger to both traffic and pedestrians, especially after dark.

[0003] Since the development of slums typically occurs without planning guidelines and is usually not accurately represented on official maps, and can also change rapidly, no intersection points between paths within the slums and official roads can be derived from such maps.

[0004] The present invention therefore aims to propose a method that avoids these disadvantages, thus enabling the reliable detection of areas on roads with an increased risk of collision.

[0005] This problem is solved according to the invention by a method according to the main claim and a server according to the dependent claim.

[0006] Advantageous embodiments of the invention are the subject of the dependent claims.

[0007] In a method according to the invention for identifying locations on roads with an increased risk of collision, an image processing step and a transmission step are performed. In the image processing step, an evaluation unit divides an aerial image, depicting at least one road and adjacent buildings, into geometric patterns and identifies at least one building within each pattern. For each of the identified buildings, the evaluation unit calculates a footprint and, if a certain number of buildings fall below a predetermined footprint threshold, determines a number of buildings with a predetermined footprint shape. Each open area of ​​the aerial image not covered by an identified building is determined, and the evaluation unit assigns each of these open areas a line running along a longitudinal dimension of the respective open area.Finally, the intersection of at least one of the lines with a road is identified as a location with an increased risk of collision. In the transmission step, a warning regarding this location is sent via a communication unit to a vehicle's navigation system and displayed by the navigation system if the vehicle is moving in that direction. Since an aerial image is used, the actual building situation is also taken into account. The information can be transmitted not only to the vehicle's navigation system but also to the relevant authority, so that warning signs or physical access restrictions can be erected at the identified road location with an increased risk of collision.

[0008] In the image processing step, current information, such as that obtained from an aerial image (typically a satellite image, as opposed to maps specifically for slums), is analyzed using several criteria or thresholds. Buildings and paths leading past them are identified with a sufficiently high probability by the evaluation unit, which is typically electronic. Where one of these paths intersects with a road also present in the image, this indicates a location with an increased risk of collision from pedestrians suddenly appearing on the road.

[0009] This information, determined by the evaluation unit, is transmitted as a warning to the vehicle's navigation system during the transmission step and, if necessary, is issued by the navigation system in a way that is recognizable to the driver, for example as an acoustic and / or visual warning signal, if the vehicle is moving in the direction of this location.

[0010] In one embodiment, the width of a clear area at the point of increased collision risk can be determined during the image processing step. The warning message is only transmitted if the determined width falls below a certain value. This is intended to minimize false alarms or false positives by adding an additional criterion that must be met before the warning message is issued. By focusing on the smallest possible width, it can be concluded that this clear area is a footpath that is particularly dangerous because it is difficult to see and from which pedestrians can enter a roadway.

[0011] It is preferably intended that the determined value for the calculated width be 3 meters. Greater widths are easier to see across, and a width of 3 meters is also suitable for allowing several pedestrians to walk side by side.

[0012] In one embodiment of the method, the geometric patterns can be rectangles or squares, preferably with a width of 250 m and a length of 250 m. This ensures complete coverage of the aerial image, while the dimensions are chosen such that buildings can be reliably detected.

[0013] In another embodiment of the method, the specified number of buildings can correspond to two-thirds of the number of buildings identified on the geometric pattern. A threshold of two-thirds ensures a sufficiently high probability and leaves open the possibility of considering larger, unbuildable open spaces such as rivers.

[0014] The procedure may stipulate that the floor area threshold is 10 m². 2 Since slums predominantly feature smaller buildings, contributing to high building density, a relatively low floor area threshold is justified as a criterion.

[0015] In another embodiment of the method, the specified base shape can be a quadrilateral, typically a rectangle. Since such buildings are usually simple in design, correspondingly simple base shapes are used. Constructing buildings with a quadrilateral base is relatively easy and inexpensive. A rectangular shape, in particular, often allows for optimal use of available space, especially in densely populated areas.

[0016] In another embodiment of the method, the line running along a longitudinal dimension of the respective area can be a center line. Based on the determined built-up area, a line running centrally across an open space, the center line, can be easily identified and designates a path that is likely to be used by a large number of people.

[0017] In another embodiment of the method, the evaluation unit can be a component of a server. By providing a server, the relevant information can be processed in one location and transmitted from there to the various vehicles.

[0018] A server according to the invention comprises an evaluation unit and a communication unit. The evaluation unit is configured to perform an image processing step in which an aerial image depicting at least one street and adjacent buildings is divided into geometric patterns. At least one building is identified in one of the patterns. For each of the identified buildings, the evaluation unit calculates a footprint, and if a certain number of the buildings fall below a predetermined footprint threshold, the evaluation unit determines a number of buildings with a predetermined footprint shape.Each area of ​​the aerial image not covered by an identified building is determined, and each of these areas is assigned a line running along its longitudinal extent by the evaluation unit. Finally, in the image processing step, the intersection of at least one of these lines with a road is identified as a location with an increased risk of collision. The communication unit is configured to perform a transmission step in which a warning regarding this location is sent via a communication unit to a vehicle's navigation system and displayed by the navigation system if the vehicle is moving in the direction of this location.

[0019] The server is typically configured to perform the procedure described above.

[0020] A system consists of a server with the properties described above and a vehicle whose navigation system is set up to receive the warning message.

[0021] A computer program product comprises instructions that, when executed by a computer, cause it to perform the previously described procedure or steps of the procedure. The computer program product (or parts thereof) can accordingly also be executed on the described vehicle and / or server.

[0022] Exemplary embodiments of the invention are explained in more detail below with reference to drawings.

[0023] Corresponding parts are marked with the same reference symbols in the figures.

[0024] They show: Fig. 1. A schematic flowchart of a procedure for identifying locations on roads with an increased risk of collision; Fig. 2 an aerial photograph of a residential development with an adjacent road; Fig. 3 one Fig. 2. corresponding view of the aerial photograph, in which open areas have now been identified and Fig. 4 one Fig. 2. A corresponding view of the aerial photograph, where areas with an increased risk of collision were identified.

[0025] Fig. Figure 1 shows a schematic diagram of a procedure for identifying locations on roads with an increased risk of collision.

[0026] In an initial step S100, an aerial photograph of an area with dense residential development and an adjacent road is used as the basis of the process. An evaluation unit divides this aerial photograph into geometric patterns. In the embodiment discussed here, squares with a length of 250 meters and a width of 250 meters are chosen as the geometric pattern. However, other geometric shapes and / or dimensions can be used in further embodiments. The outlines of buildings are identified within each of these rectangles.

[0027] In a subsequent step S101, one of these squares is considered and the outlines of buildings located in the square are determined.

[0028] After step S101 is performed, in a subsequent step S102, the evaluation unit calculates the floor area of ​​each of the identified buildings for a selected square, i.e., one of the geometric patterns obtained in step S101. It also checks whether a majority of at least two-thirds of the identified buildings have a floor area of ​​less than 20 square meters, as this indicates an area with high building density and inadequate infrastructure, also known as a "slum." In other implementation examples, a different threshold value can, of course, be used.

[0029] If this test criterion is not met, i.e., the selected square has less than two-thirds buildings with a footprint of 20 square meters, the procedure continues with step S104 (the fact that the test criterion is negative is indicated in Fig. 1 (marked by “-”). In further embodiments, a different threshold value can also be used.

[0030] In step S104, the next square is selected, and in step S101, building outlines are determined in this newly selected square, and the procedure is continued as described above.

[0031] If the test criterion is met (which is in Fig. In step S103, buildings with a rectangular footprint are identified (marked by a "+"). The presence of a rectangular footprint, in conjunction with the comparatively small footprint, suggests the presence of a slum, where construction typically occurred and continues without planning regulations, resulting in the absence or unusability of cartographic material. Consequently, even maps, if available, provide no information regarding potential intersections of paths within the slum with roads bordering it. In a further embodiment, a check can also be performed for the presence of buildings with a rectangular footprint.

[0032] If the test in step S103 is negative, the process continues with step S104 and considers the next square. However, if the test is positive, open spaces within the respective square are identified in step S105. An open space is defined here as any area not covered by a building. These open spaces are highly likely to be footpaths.

[0033] In step S106, the evaluation unit calculates a line running along the longitudinal extent of each of the identified open spaces and assigns it to each space. This line is generally a center line running lengthwise through the respective open space, i.e., a line running through the middle of the open space.

[0034] In step S107, at least one intersection of the lines with a road is therefore identified as a location with an increased risk of collision. Such intersections between the identified paths and the roads also mapped represent, with a sufficiently high probability, danger points where people stepping uncontrollably onto the roadway can endanger themselves and vehicles on the road.

[0035] Steps S100 to S107 represent an image processing step in which danger points are identified from a given aerial image (such as a satellite image or a drone image) by the evaluation unit.

[0036] In a subsequent transmission step, a warning regarding this location identified as a hazard is transmitted via a communication unit to a vehicle's navigation system. If the vehicle is moving towards this location, the warning is issued, i.e., the driver is warned visually and / or audibly and urged to exercise increased caution in this area.

[0037] At the in Fig. In the procedure shown in Figure 1, this image processing step is supplemented by step S109, in which, after step S107, it is checked whether at least one of the lines has a narrow entrance at the identified intersection point.

[0038] Optionally, in step S108, which precedes step S109, it is examined whether brightness differences are detected in building facades in images obtained from one or more vehicles. If such a brightness difference is detected, this is interpreted as indicating a recess in the building. This recess is typically one of the previously identified open spaces.

[0039] In step S109, the system checks whether there is a corresponding difference in brightness, i.e., whether there is a gap in the building facades, or whether this open space falls below a certain width (this width could be, for example, 5 meters). If a condition is met, i.e., if the aerial photograph or at least one image taken from a vehicle indicates that the intersection point, as a location with an increased risk of collision, also has a comparatively narrow open space, essentially serving as an entrance to the residential area, the corresponding warning is transmitted.

[0040] This occurs in step S110, which, as already explained, can also directly follow step S107. Step S110 can directly include the transmission of the warning message as a transmission step. Alternatively or additionally, the storage of a corresponding message in a database can also be provided first.

[0041] If the conditions are deemed not to be met in step S109, the procedure is terminated in step S111. If the database already contains an indication of a location with an increased risk of collision, this location will be removed from the database of locations with an increased risk of collision if, for example, not all criteria are met in step S109 due to an image taken by a vehicle.

[0042] The process is typically performed on a server that includes the evaluation unit for performing the image processing step and the communication unit for performing the communication step. The server may also include a database for storing the identified areas with a high risk of collision. A computer program can be executed on the server, containing commands that, when executed by a computer such as the server, cause it to perform the described process or steps of the process.

[0043] Together with a vehicle whose navigation system is set up to receive the warning, the server forms a system for identifying locations on roads with an increased risk of collision.

[0044] Fig. Figure 2 schematically shows an aerial photograph underlying the described method, which is, for example, a drone image or a satellite image. Next to two roads 100 and 101, which in the illustrated embodiment are highways or expressways, there is an area 200 with a high building density. Buildings 300 are recognizable through their roofs and can be identified by image processing as described above.

[0045] Fig. 3 represents in a Fig. 2 corresponding view the edited aerial image from Fig. Step S105 has already been executed, meaning that open areas 400 were determined during image processing and are shown hatched. The white areas indicate the areas covered by buildings 300.

[0046] Fig. 4. Finally, the final result after the image processing step is shown: Center lines 500 were determined for each of the free areas 400 and intersection points 600 (which are in Fig.4 (also marked by a “!”) between center lines 500 and road 101 were determined and drawn. Reference symbol list S100 - S111 Procedure steps 100 Street 101 Street 200 Area with high building density 300 buildings 400 open spaces 500 center lines 600 intersection points

Claims

[1] Method for identifying locations (600) on roads (100,101) with an increased risk of collision, comprising an image processing step and a transmission step, wherein - in the image processing step, an aerial image depicting at least one street (100,101) and adjacent buildings (300) is subdivided into geometric patterns (S100) by an evaluation unit, and - in a pattern at least one building (300) is identified (S101) and - for each of the identified buildings (300) a floor area is calculated by the evaluation unit (S102), and, if a certain number of the buildings (300) fall below a specified floor area threshold, a number of buildings (300) with a specified floor area shape are determined by the evaluation unit (S103), wherein - each open space (400) of the aerial image not covered by an identified building (300) is identified (S105) and each of these open spaces (400) is assigned by the evaluation unit a line (500) running along a longitudinal extent of the respective open space (400) (S106), wherein - finally, an intersection (600) of at least one of the lines (500) with a road (100,101) is identified as a location with an increased risk of collision (S107) and - in the transmission step a warning regarding this location is transmitted via a communication unit to a vehicle's navigation system and is displayed by the navigation system if the vehicle is moving towards this location (S110). [2] Method according to claim 1, wherein during the image processing step the width of a free area (400) at the location with increased risk of collision is determined, wherein the warning message is only transmitted in the transmission step if a determined width falls below a certain value. [3] Method according to claim 2, wherein the determined value for the determined width is 3 meters. [4] Method according to any of the preceding claims, wherein the geometric patterns are rectangles which preferably have a width of 250 m and a length of 250 m. [5] Method according to any of the preceding claims, wherein the specified number of buildings (300) corresponds to two-thirds of the number of buildings (300) identified on the geometric pattern. [6] Method according to one of the preceding claims, wherein the floor area threshold is 10 m² 2 amounts. [7] Method according to any of the preceding claims, wherein the specified base shape is a quadrilateral. [8] Method according to any of the preceding claims, wherein the line (500) running along a longitudinal extent of the respective surface is a center line. [9] Method according to any of the preceding claims, wherein the evaluation unit is a component of a server. [10] Server with an evaluation unit and a communication unit, wherein - the evaluation unit is set up to perform an image processing step in which - an aerial photograph showing at least one street (100,101) and adjacent buildings (300), subdivided into geometric patterns (S100) and - in one of the patterns at least one building (300) is identified (S101), and - for each of the identified buildings (300) a floor area is calculated by the evaluation unit (S102), and, if a certain number of the buildings (300) fall below a specified floor area threshold, a number of buildings (300) with a specified floor area shape are determined by the evaluation unit (S103), wherein - each area of ​​the aerial image not covered by an identified building (300) is determined (S105) and each of these areas is assigned by the evaluation unit a line (500) running along a longitudinal extent of the respective area (S106), wherein - finally, an intersection (600) of at least one of the lines (500) with a road (100,101) is identified as a location with an increased risk of collision (S107), and wherein, - the communication unit is set up to carry out a transmission step in which a warning regarding this location is transmitted via a communication unit to a vehicle's navigation system and is displayed by the navigation system if the vehicle is moving towards this location (S110). [11] System comprising a server according to claim 10 and a vehicle whose navigation system is configured to receive the warning message. [12] Computer program product comprising instructions which, when the program is executed by a computer, cause it to execute the method / steps of the method according to any one of claims 1-9.

Images