METHOD FOR DETECTING FALSE POSITIVES OF AN IMAGE PROCESSING DEVICE OF A CAMERA

Description

[0001] The invention relates to the field of vehicles equipped with a driving assistance system.

[0002] A method for detecting false positives of an image processing device of a camera is provided.

[0003] Known from the state of the art are vehicles comprising at least one camera comprising an image processing device capable of providing modeling data of a marking line of a traffic lane of a vehicle.

[0004] State-of-the-art camera image processing devices are not always very reliable, especially when the vehicle is traveling on a poorly maintained roadway or when the detection of markings is ambiguous. False detections on marking lines can lead to untimely corrections of the vehicle trajectory. This reduces the comfort of vehicle passengers and can cause user mistrust of semi-autonomous or autonomous driving systems.

[0005] The purpose of this invention is to provide a method for detecting false positives of a camera image processing device making it possible to avoid false positive detections and the associated discomfort.

[0006] The invention relates to a method, according to claim 1, for detecting false positives of an image processing device of a camera on board a vehicle 1, said image processing device being capable of providing modeling data of a marking line of a traffic lane of the vehicle in a reference frame fixed relative to the vehicle, the false positive detection method comprising the following steps: a first step of determining, from the marking line modeling data, a first distance from the vehicle to the marking line, at a first instant in time, a second step of determining, from the marking line modeling data, a second distance from the vehicle to the marking line, at a second instant in time, an initial discontinuity calculation step in which an initial discontinuity is calculated, the initial discontinuity being equal to the absolute value of the difference between the first distance and the second distance, an initial discontinuity false positive detection step, in which the initial discontinuity is compared to a predetermined initial low threshold, an initial discontinuity false positive being detected if the initial discontinuity is greater than the initial low threshold.

[0007] According to the invention, the initial low threshold is equal to a value between 0.1 meters and 0.3 meters, the time difference between the first instant and the second instant being less than or equal to the quotient of the initial low threshold by a lateral speed substantially equal to two meters per second, the time difference being greater than one millisecond.

[0008] According to the invention, the initial discontinuity is further compared to a predetermined initial high threshold, a false positive of initial discontinuity being detected if the initial discontinuity is greater than the initial low threshold and less than the initial high threshold.

[0009] According to one aspect of the invention, the initial high threshold is equal to a value between 1.7 meters and 2.3 meters.

[0010] According to one aspect of the invention, the reference system comprises a ordinate axis substantially parallel to a lateral direction of the vehicle and a point of origin located at the level of the vehicle camera, the first distance and second distance correspond to the ordinate at the point of origin of a point of the marking line respectively at the first instant and at the second instant in time.

[0011] According to one aspect of the invention, the reference system comprises an abscissa axis substantially parallel to a longitudinal direction of the vehicle, the method for detecting false positives further comprises the following steps: a third step of determining, from the modeling data of the marking line, a first ordinate of a horizon point of the marking line whose first abscissa is a distance corresponding to the distance to be covered by the vehicle in a predefined horizon duration, at the first instant in time, a fourth step of determining, from the modeling data of the marking line, a second ordinate of a horizon point of the marking line whose second abscissa is a distance corresponding to the distance to be covered by the vehicle in said predefined horizon duration, at the second instant in time, a horizon discontinuity calculation step in which a horizon discontinuity is calculated, the horizon discontinuity being equal to the absolute value of the difference between the first ordinate and the second ordinate, from which the initial discontinuity is subtracted,a horizon discontinuity false positive detection step, wherein the horizon discontinuity is compared to a predetermined horizon threshold, a horizon discontinuity false positive being detected if the horizon discontinuity is greater than the horizon threshold.

[0012] According to one aspect of the invention, the predefined horizon time is a value between 0.9 seconds and 1.1 seconds, the horizon threshold is a value between 0.2 meters and 1 meter.

[0013] According to one aspect of the invention, the modeling data comprises an optimal visibility distance, the false positive detection method further comprising a step of detecting a false range positive, in which the optimal visibility distance is compared to a range threshold equal to the distance to be covered by the vehicle in a predefined range duration, a false range positive being detected if the optimal visibility distance is less than the range threshold.

[0014] According to one aspect of the invention, the predefined range duration is a value between 0.8 seconds and 0.99 seconds.

[0015] According to one aspect of the invention, the method for detecting false positives further comprises a step of displaying a visual alert to a user of the vehicle if a false positive among an initial discontinuity false positive, a horizon discontinuity false positive or a range false positive is detected, to warn the user of a temporary loss of detection of the marking line by the camera.

[0016] According to one aspect of the invention, the modeling data of the marking line are of polynomial type of degree greater than or equal to three.

[0017] The invention also relates to a computer program product, according to claim 9, comprising the program instructions implementing at least one step of the method for detecting false positives, when the program instructions are executed by a computer.

[0018] The invention also relates to a readable information medium, according to claim 10, on which the computer program product is stored.

[0019] Other advantages and characteristics of the invention will become apparent from reading the description and the drawings. [ Fig 1 ] is a diagram representing a vehicle equipped with an image processing device of a camera traveling on a traffic lane at a first instant in time. [ Fig 2 ] is a diagram showing a vehicle equipped with an image processing device of a camera traveling on a traffic lane at a second instant in time. [ Fig 3 ] is a diagram showing an alternative embodiment of the figure 1 . [ Fig 4 ] is a diagram showing an alternative embodiment of the figure 2 . [ Fig 5 ] is a diagram representing a variant of the figure 2 . [ Fig 6 ] is a camera image. [ Fig 7 ] is a diagram representing the steps of a method for detecting false positives of an image processing device of a camera, according to the invention. [ Fig 8a ] represents a display screen on which are represented marking lines of a vehicle traffic lane for which no false positives have been detected. [ Fig 8b ] represents a display screen on which are represented marking lines of a vehicle lane for which a false positive has been detected.

[0020] There figure 1 represents a vehicle 1, at a first instant T1 in time, traveling on a traffic lane 10 comprising two marking lines 11, 12 located on either side of said vehicle 1.

[0021] The vehicle 1 comprises a camera 2 comprising an image processing device capable of providing modeling data of a marking line 11, 12.

[0022] The modeling data makes it possible to model the marking line 11, 12 in a reference frame 3 fixed relative to the vehicle 1.

[0023] The modeling data makes it possible to model a marking line 11, 12 as a continuous curve, whether the ground markers, for example reflective strips, are present continuously or discontinuously along the marking line 11, 12.

[0024] Preferably, the modeling data is of polynomial type of degree greater than or equal to three. For example, the image processing device is capable of providing the coefficients of the polynomial.

[0025] The fixed reference frame 3 comprises a Y ordinate axis substantially parallel to a lateral direction of the vehicle 1 and an X abscissa axis substantially parallel to a longitudinal direction of the vehicle, the X abscissa axis and the Y ordinate axis intersecting at a point of origin O.

[0026] The longitudinal direction of vehicle 1 is, for example, a straight line passing through the axle center of vehicle 1, the front of the vehicle and the rear of the vehicle.

[0027] Preferably, the reference frame 3 is an orthogonal reference frame.

[0028] The lateral direction of vehicle 1 is perpendicular to the longitudinal direction of vehicle 1, for example a straight line passing through the left front door and through the right front door.

[0029] Advantageously, the point of origin O is located at the level of the camera 2 of the vehicle 1. The camera 2 is a front camera located at the front of the vehicle 1.

[0030] The marking line 11 comprises an initial point Pi whose coordinates in the reference frame 3 are represented by an abscissa and an ordinate and can be determined from the modeling data provided by the image processing device of the camera 2.

[0031] Advantageously, the abscissa and the ordinate of the origin point O are equal to zero.

[0032] The vehicle 1 comprises a false positive detection device 5 capable of determining the ordinate of a point on the marking line 11, 12 from its abscissa.

[0033] In a preferred embodiment, the abscissa Xi of the initial point Pi is substantially equal to zero. The abscissa Xi is therefore substantially coincident with the point of origin O. The ordinate of the initial point Pi is a first distance Yi from the vehicle 1 to the marking line 11 at a first instant T1 in time. The first distance Yi is the ordinate from the initial point Pi to the point of origin O. The first distance Yi is determinable from the modeling data of the marking line 11. The false positive detection device 5 is capable of determining the first distance Yi from the modeling data of the marking line 11 provided by the image processing device of the camera 2.

[0034] The marking line 11 also includes a horizon point Ph whose abscissa at the first instant T1 in time is a first abscissa Xh corresponding to the distance to be covered by the vehicle 1 in a predefined horizon duration dTh.

[0035] The vehicle 1 comprises a speed sensor 6. The false positive detection device 5 is capable of recovering from said speed sensor 6 the speed of the vehicle at a given time. From the vehicle speed, the false positive detection device is capable of calculating the distance to be covered by the vehicle 1 in a predefined time horizon dTh.

[0036] The predefined horizon duration dTh is preferably between 0.9 seconds and 1.1 seconds, typically equal to 1 second.

[0037] For example, for a horizon duration dTh equal to one second and a vehicle speed equal to 25 meters per second, the first abscissa Xh is equal to 25 meters.

[0038] The false positive detection device 5 is capable of determining the ordinate of the horizon point Ph called the first ordinate Yh at the first instant T1, from the first abscissa Xh previously calculated and the modeling data of the marking line 11 provided by the image processing device of the camera 2.

[0039] There figure 2 differs from the figure 1 in that it represents the vehicle 1 traveling on the traffic lane 10 at a second instant T2 in time. The first instant T1 is prior to the second instant T2. When the vehicle 1 travels on the traffic lane 10, the vehicle travels a non-zero distance on the traffic lane 10 between the first instant T1 and the second instant T2.

[0040] The time difference DT, in absolute value, between the first instant T1 and the second instant T2 is preferably between 1 millisecond and 150 milliseconds.

[0041] The reference frame 3 being fixed relative to the vehicle and the vehicle 1 having changed position on the traffic lane 10, the marking line 11 is distant from the vehicle 1 by a second distance Y'i, at the second instant T2. The second distance Y'i from the vehicle 1 is the ordinate of an initial point P'i whose abscissa Xi is therefore coincident with the point of origin O, as on the figure 1 .

[0042] The marking line 11 includes a new horizon point P'h whose abscissa at the second instant T2 in time is a second abscissa X'h corresponding to the distance to be covered by the vehicle 1 in the time at the horizon dTh predefined at the second instant T2.

[0043] The first abscissa Xh and the second abscissa X'h are not necessarily equal but they are very close. Indeed, the time at the horizon dTh is a fixed value, but the speed of vehicle 1 is not necessarily the same at the first instant T1 and at the second instant T2. However, the time difference DT, in absolute value, between the first instant T1 and the second instant T2 being less than 150 milliseconds, the difference between the first abscissa Xh and the second abscissa X'h is small.

[0044] For example, if the difference in vehicle speed between the first instant T1 and the second instant T2 is equal to 0.05 meters per second, then the difference between the first abscissa Xh and the second abscissa X'h is equal to 0.05 meters, for a duration at the horizon dTh equal to one second.

[0045] From the vehicle speed at the second instant T2, the false positive detection device 5 is able to calculate the distance to be covered by the vehicle 1 in a predefined time horizon dTh, which corresponds to the second abscissa X'h.

[0046] The false positive detection device 5 is capable of determining the ordinate of the horizon point P'h called second ordinate Y'h at the second instant T2, from the second abscissa X'h previously calculated and the modeling data of the marking line 11 provided by the image processing device of the camera 2.

[0047] THE figures 3 et 4 represent an alternative embodiment respectively of the figures 1 et 2 .

[0048] There figure 3 differs from the figure 1 in that the abscissa Xi of the initial point Pi is not confused with the point of origin O. The abscissa Xi is at a predefined distance from the point of origin O.

[0049] As on the figure 1 , the first distance Yi from vehicle 1 to marking line 11 at the first instant T1 in time is defined by the ordinate of point Pi.

[0050] The false positive detection device 5 is capable of determining the first distance Yi from the abscissa Xi which is predetermined and known, and from the modeling data of the marking line 11 provided by the image processing device of the camera 2.

[0051] There figure 4 differs from the figure 2 in that the abscissa Xi of the initial point P'i is not confused with the point of origin O. The abscissa Xi is the same predefined distance from the point of origin O as on the figure 3 .

[0052] As on the figure 2 , the second distance Y'i from vehicle 1 to marking line 11 at the second instant T2 in time is defined by the ordinate of point P'i.

[0053] The false positive detection device 5 is capable of determining the second distance Y'i from the abscissa Xi which is predetermined and known and from the modeling data of the marking line 11 provided by the image processing device of the camera 2.

[0054] As represented in the figures 3 et 4 , it may be interesting not to confuse the point of origin O with the abscissa Xi in the case where camera 2 is not a front camera located at the front of the vehicle.

[0055] For example, camera 2 is located on the roof of vehicle 2. Advantageously, the abscissa Xi is the distance between the origin point O located on camera 2 and the front of vehicle 2.

[0056] There figure 5 is a variant of the figure 2 which can also apply to the figure 4 .

[0057] There figure 5 differs from the figure 2 in that the second ordinate Y'h at the second time T2 is determined by the false positive detection device 5 from the first abscissa Xh calculated at the first time T1. This has the advantage of determining the first ordinate Yh and the second ordinate Y'h at the same first abscissa Xh.

[0058] There figure 6 is an example of image 20 from camera 6. Marker 3 is superimposed on image 20 from camera 6. Marking lines 11, 12 are discontinuous.

[0059] We find the point of origin Pi and the horizon point Ph, located on the marking line 11, as illustrated in the figure 1 .

[0060] There figure 7 illustrates a method for detecting false positives of an image processing device of a camera 2 on board a vehicle 1 as shown in figures 1 à 5.

[0061] The false positive detection method allows to detect false positives among a false positive of initial discontinuity FPi, a false positive of discontinuity at the horizon FPh and a false positive of range FPp.

[0062] To detect a false positive of initial discontinuity FPi, the false positive detection method includes the following steps: a first step 101 of determining, from the modeling data of the marking line 11, the first distance Yi of the vehicle 1 to the marking line 11, at the first instant T1, a second step 102 of determining, from the modeling data of the marking line 11, the second distance Y'i of the vehicle 1 to the marking line 11, at the second instant T2, an initial discontinuity calculation step 103 in which an initial discontinuity Di is calculated, the initial discontinuity Di being equal to the absolute value of the difference between the first distance Yi and the second distance Y'i, an initial discontinuity false positive detection step 104, in which the initial discontinuity Di is compared to a predetermined initial low threshold TBi, an initial discontinuity false positive FPi being detected if the initial discontinuity Di is greater than the initial low threshold TBi.

[0063] According to the invention, the initial discontinuity Di is further compared to a predetermined initial high threshold THi, a false positive of initial discontinuity being detected if the initial discontinuity Di is greater than the low threshold TBi and less than the initial high threshold THi. This makes it possible not to detect false positives in the case of a change of traffic lane by the vehicle 1.

[0064] The initial low threshold TBi is equal to a value between 0.1 meters and 0.3 meters, typically equal to 0.15 meters.

[0065] The initial high threshold THi is equal to a value between 1.7 meters and 2.3 meters.

[0066] The time difference DT, in absolute value, between the first instant T1 and the second instant T2 is a value less than or equal to the quotient of the initial low threshold TBi by a lateral speed substantially equal to two meters per second.

[0067] Below a lateral speed of two meters per second, lateral movement is considered involuntary. Above this speed, lateral movement is considered voluntary. Therefore, it is advantageous to choose a speed of two meters per second as this covers scenarios of involuntary lateral movement.

[0068] The time difference DT, in absolute value, between the first instant T1 and the second instant T2 is greater than one millisecond in order to give the false positive detection device 5 sufficient time to execute the steps of the false positive detection method.

[0069] For an initial low threshold TBi equal to 0.1 meters, the time difference DT, in absolute value, is therefore a value between 1 millisecond and 50 milliseconds, which is the result of the quotient of 0.1 by 2.

[0070] For an initial low threshold TBi equal to 0.3 meters, the time difference DT, in absolute value, is therefore a value between 1 millisecond and 150 milliseconds, which is the result of the quotient of 0.3 by 2.

[0071] To detect a false positive of discontinuity at the FPh horizon, the false positive detection method includes the following steps: a third step 201 of determining, from the modeling data of the marking line 11, the first ordinate Yh of the horizon point Ph of the marking line 11, at the first instant T1, a fourth step 202 of determining, from the modeling data of the marking line 11, the second ordinate Y'h of the horizon point P'h of the marking line 11 at the second instant T2, a step of calculating discontinuity at horizon 203 in which a discontinuity at horizon Dh is calculated, the discontinuity at horizon Dh being equal to the absolute value of the difference between the first ordinate Yh and the second ordinate Yh', from which the initial discontinuity Di is subtracted, a step of detecting a false positive of discontinuity at horizon 204, in which the discontinuity at horizon Dh is compared to a predetermined horizon threshold THh, a false positive of discontinuity at horizon FPh being detected if the discontinuity at horizon Dh is greater than the horizon threshold THh.

[0072] The determination of the first ordinate Yh of the horizon point Ph and the second ordinate Y'h of the horizon point P'h is carried out from the first abscissa Xh or the second abscissa X'h which are calculated as explained in the description of the figures 2 , 4 And 5 , and modeling data of the marking line 11.

[0073] The THh horizon threshold is between 0.2 meters and 1 meter.

[0074] The modeling data includes an optimal visibility distance Xp.

[0075] An example of optimal visibility distance Xp is given on the figure 6 A truck 21 is traveling on the traffic lane 10. For the camera 6, the truck 21 is an obstacle on the traffic lane 10. The truck 21 obstructs part of the marking lines 11, 12. In this example, the optimal visibility distance Xp corresponds substantially to the distance between the truck 21 and the camera 6.

[0076] To detect a false positive of range FPp, the false positive detection method further comprises a false positive detection step of range 304, in which the optimal visibility distance Xp is compared to a range threshold THp equal to the distance to be covered by the vehicle in a predefined range duration dTp, a false positive of range FPp being detected if the optimal visibility distance Xp is less than the range threshold THp.

[0077] Advantageously, the false positive detection step of scope 304 is carried out at the second time T1 and / or at the second time T2.

[0078] The predefined range duration dTp is a value between 0.8 seconds and 0.99 seconds, typically equal to 0.95 seconds.

[0079] For example, for a range duration dTp equal to 0.95 seconds and a vehicle speed equal to 25 meters per second, the range threshold THp is equal to 23.75 meters, which is the result of the product of 0.95 by 25.

[0080] It is of course considered equivalent to convert the optimal visibility distance Xp into a time that vehicle 1 would take to cover said optimal visibility distance Xp and to compare said time to the range threshold THp.

[0081] A false positive is detected if any of an initial discontinuity false positive FPi, a horizon discontinuity false positive FPh, or a range false positive FPp is detected.

[0082] Advantageously, if a false positive is detected, the method for detecting false positives further comprises a step 124 of displaying a visual alert to a user of the vehicle 1, in order to warn the user of a temporary loss of detection of the marking line 11 by the camera 2.

[0083] For example, the vehicle 1 comprises a display screen 100 on which the marking lines 11, 12 of the traffic lane 10 are represented. If a false positive is detected on the marking line 11 by the false positive detection device 5, then the representation 110 of the marking line 11 on the display screen 100 is modified, for example by a change of color. The representation 120 of the marking line 12 is not modified if a false positive has not been detected on the marking line 12.

[0084] On the figure 8a , no false positive is detected, the representation 110, 120 of the two marking lines 11, 12 is black.

[0085] On the figure 8b , a false positive is detected on the marking line 11. The representation 110 of the marking line 11 is gray, the representation 120 of the marking line 12 is unchanged.

[0086] This visual alert embodiment is not limiting. The visual alert can be a pictogram, a light, etc.

[0087] The false positive detection device 5 comprises a program comprising the program instructions implementing the steps of the false positive detection method.

[0088] The false positive detection device 5 is connected to the speed sensor 6 of the vehicle 1 and to the image processing device of the camera 2, preferably by a wired connection. The connection between the speed sensor 6 and the false positive detection device 5 may be direct or indirect. For example, an intermediate computer may be located between the speed sensor 6 and the false positive detection device 5, the intermediate computer making it possible to filter and verify data before their transmission to the false positive detection device 5.

[0089] The communication protocol between the speed sensor 6 and the false positive detection device 5 is for example of the CAN type.

[0090] The communication protocol between the image processing device of the camera 2 and the false positive detection device 5 is for example of the CAN type.

[0091] Advantageously, the program instructions implementing the steps of the false positive detection method are executed in a loop according to a period equal to the time difference DT. During the first execution of the program, the first instant T1 corresponds to an instant of said first period, the second instant T2 corresponds to an instant of said second period. During the second execution of the program, the first instant T1 corresponds to said instant of said second period, the second instant T2 corresponds to an instant of said third period, and so on.

[0092] The above examples and description are given taking marking line 11 as an example. By analogy, the same examples and explanations apply to marking line 12.

Claims

1. Method for detecting false positives of an image-processing device of a camera (2) located on-board a vehicle (1), said image-processing device being able to deliver modelling data of a marking line (11, 12) of a lane (10) of the vehicle (1) in a frame of reference (3) that is tied to the vehicle (1), the method for detecting false positives comprising the following steps: - a first step (101) of determining, from the modelling data of the marking line (11, 12), a first distance (Yi) of the vehicle (1) to the marking line (11, 12), at a first instant (T1) in time, - a second step (102) of determining, from the modelling data of the marking line (11, 12), a second distance (Y'i) of the vehicle (1) to the marking line (11, 12), at a second instant (T2) in time, - an initial-discontinuity-computing step (103) in which an initial discontinuity (Di) is computed, the initial discontinuity (Di) being equal to the absolute value of the difference between the first distance (Yi) and the second distance (Y'i), - a step (104) of detecting a false initial-discontinuity positive, in which step the initial discontinuity (Di) is compared to a preset low initial threshold (TBi), a false initial-discontinuity positive (FPi) being detected if the initial discontinuity (Di) is larger than the low initial threshold (TBi), the low initial threshold (TBi) being equal to a value comprised between 0.1 metres and 0.3 metres, the time difference (DT) between the first instant (T1) and the second instant (T2) being smaller than or equal to the quotient of the low initial threshold (TBi) divided by a lateral speed substantially equal to two metres per second, the time difference (DT) being larger than one millisecond, the initial discontinuity (Di) furthermore being compared to a preset high initial threshold (THi), a false initial-discontinuity positive being detected if the initial discontinuity (Di) is larger than the low initial threshold (TBi) and smaller than the high initial threshold (THi).

2. Method for detecting false positives according to the preceding claim, the high initial threshold (THi) being equal to a value comprised between 1.7 metres and 2.3 metres.

3. Method for detecting false positives according to any one of the preceding claims, the frame of reference (3) comprising an ordinate axis (Y) substantially parallel to a lateral direction of the vehicle (1) and an origin (O) located level with the camera (2) of the vehicle (1), the first distance (Yi) and the second distance (Y'i) corresponding to the ordinate at the origin (O) of a point (Pi, P'i) of the marking line (11, 12) at the first instant (T1) and at the second instant (T2) in time, respectively.

4. Method for detecting false positives according to the preceding claim, the frame of reference (3) comprising an abscissa axis (X) substantially parallel to a longitudinal direction of the vehicle (1), the method for detecting false positives furthermore comprising the following steps: - a third step (201) of determining, from the modelling data of the marking line (11, 12), a first ordinate (Yh) of a horizon point (Ph) of the marking line (11, 12) the first abscissa (Xh) of which is a distance corresponding to the distance to be travelled by the vehicle (1) in a predefined horizon time (dTh), at the first instant (T1) in time, - a fourth step (202) of determining, from the modelling data of the marking line (11, 12), a second ordinate (Y'h) of a horizon point (P'h) of the marking line (11, 12) the second abscissa (X'h) of which is a distance corresponding to the distance to be travelled by the vehicle (1) in said predefined horizon time (dTh), at the second instant (T2) in time, - a horizon-discontinuity-computing step (203) in which a horizon discontinuity (Dh) is computed, the horizon discontinuity (Dh) being equal to the absolute value of the difference between the first ordinate (Yh) and the second ordinate (Yh'), minus the initial discontinuity (Di), - a step (204) of detecting a false horizon-discontinuity positive, in which step the horizon discontinuity (Dh) is compared to a preset horizon threshold (THh), a false horizon-discontinuity positive (FPh) being detected if the horizon discontinuity (Dh) is larger than the horizon threshold (THh).

5. Method for detecting false positives according to the preceding claim, the predefined horizon time (dTh) being a value comprised between 0.9 seconds and 1.1 seconds, the horizon threshold (THh) being a value comprised between 0.2 metres and 1 metre.

6. Method for detecting false positives according to any one of the preceding claims, the modelling data comprising an optimal visibility distance (Xp), the method for detecting false positives furthermore comprising a step (304) of detecting a false range positive, in which step the optimal visibility distance (Xp) is compared to a range threshold (THp) equal to the distance to be travelled by the vehicle in a predefined range time (dTp), a false range positive (FPp) being detected if the optimal visibility distance (Xp) is smaller than the range threshold (THp).

7. Method for detecting false positives according to the preceding claim, the predefined range time (dTp) being a value comprised between 0.8 seconds and 0.99 seconds.

8. Method for detecting false positives according to any one of the preceding claims, furthermore comprising a step (124) of displaying a visual warning to a user of the vehicle (1) if a false positive among a false initial-discontinuity positive (FPi), a false horizon-discontinuity positive (FPh) and a false range positive (FPp) is detected, in order to warn the user of a temporary loss of detection of the marking line (11, 12) by the camera (2).

9. Computer-program product comprising program instructions that implement the steps of the method for detecting false positives according to any one of the preceding claims, when the program instructions are executed by a computer.

10. Readable data medium on which is stored the computer-program product according to the preceding claim.