Tracking vehicles by tracking lateral objects

Abstract

The invention relates to a method for tracking a neighbouring vehicle (VV1) in an environment of a main vehicle (VP), the method being implemented by a driver assistance device (2) of the main vehicle (VP) capable of acquiring images within a field of view (FOV) at acquisition times (T0,T1), the method comprising: - obtaining (400), for an initial acquisition time (T0), a first set of parameters associated with a virtual object () associated with the neighbouring vehicle (VV1); - detecting (410), from a first image acquired at a first acquisition time (T1) subsequent to the initial acquisition time (T0), a lateral object (OBJ) associated with a second set of parameters; - tracking the neighbouring vehicle (VV1) by updating (430) the first set of parameters with the second set of parameters and, if (440) a second criterion is met, by estimating (450) a length associated with the virtual object (OV).

Description

Description Title: Vehicle tracking by lateral object tracking Technical field

[0001] This disclosure relates to the field of driver assistance systems and more particularly to the automatic detection of vehicles in areas of interest, in particular lateral areas. Prior art

[0002] The rise of Intelligent Transportation Systems (or ITS) has led to the development of numerous embedded systems in vehicles, particularly road transport. Such embedded systems include driver assistance systems and autonomous driving systems. In particular, the detection and / or tracking of objects, including vehicles, plays an important role in issues such as traffic flow, road safety, and road infrastructure management (e.g., variable message signs or speed cameras).

[0003] In the context of systems embedded in autonomous or semi-autonomous vehicles traveling in road traffic (e.g., on a motorway), reliable and long-term tracking of objects surrounding the vehicle in question is particularly important in order to avoid collisions between the vehicle in question and other vehicles, for example. In particular, in lane change assistance or automatic lane change systems, reliable tracking of obstacles such as a central reservation or another vehicle located in a lane adjacent to the vehicle in question over time is required. Indeed, if the autonomous vehicle in question does not detect and, above all, does not track another vehicle in a neighboring lane over time and moves out of the way, a collision may occur between the two vehicles.

[0004] In such a context, most existing vehicle detection and tracking methods are based on classification algorithms for identifying and tracking vehicles from image sequences or video streams acquired by vision sensors (typically, a camera). Such vision sensors are generally arranged at the front and / or rear of the vehicle concerned, so as to acquire images and / or image sequences of a front and / or rear field of view of the vehicle concerned. The classification algorithms can then identify one or more vehicles neighboring the vehicle concerned on the basis of a frontal recognition of the neighboring vehicles (i.e. by identifying and classifying detected vehicles in the field of view from their front or rear face). Moreover, existing classifiers can track detected vehicles by maintaining the detection of such a front face of the vehicles.

[0005] However, such tracking methods using vehicle classifiers prove ineffective when a front or rear face of neighboring vehicles is not yet detectable in the field of view, even if such neighboring vehicles would be partially visible in the field of view, for example via a partial lateral face (or a side). Thus, if the vehicle concerned is located on a motorway lane and a neighboring vehicle is located relatively level with (or slightly behind) the vehicle concerned on an adjacent lane, a camera placed at the rear of the vehicle concerned would capture a rear lateral portion of the neighboring vehicle. In such a situation, the front face of the neighboring vehicle is not yet visible to the camera placed at the rear of the vehicle concerned, for example until the speed differential between the vehicle concerned and the neighboring vehicle is large enough for the vehicle concerned to completely overtake the neighboring vehicle.Until such a complete overtaking occurs, the neighboring vehicle is therefore located substantially at the same level as the vehicle concerned and a lane change of the vehicle concerned into the neighboring lane would result in a collision with the neighboring vehicle. Such a criticality situation is then not detectable by existing classification methods.

[0006] Furthermore, most existing vehicle tracking methods rely on predefined dimensions associated with detected vehicle types. Such dimensions are generally not updated, so that even if vehicle tracking is implemented (e.g., by a classifier), such tracking does not allow reliable knowledge of the vehicle dimensions and therefore accurate tracking of the extent of the space occupied by the detected vehicle in the environment of the main vehicle.

[0007] There is therefore a need to secure the decision-making of driver assistance and / or autonomous driving systems, particularly in such a lane change decision context. In particular, there is a need for early and reliable detection, tracking and knowledge of vehicles surrounding a vehicle in question even before a front or rear face of such surrounding vehicles is visible in a field of view of a vision sensor of the vehicle in question - and therefore that conventional tracking of such surrounding vehicles by existing classifiers is possible. Summary

[0008] This disclosure improves the situation.

[0009] A method is proposed for tracking at least one neighboring vehicle present in the environment of a main vehicle, said main vehicle and said neighboring vehicle being motor vehicles, the method being implemented by a device configured to provide a driving assistance function for the main vehicle, said device being connected to at least one camera embedded in the main vehicle and capable of acquiring images of a scene surrounding the main vehicle according to at least one field of view and at acquisition times, the method comprising the following steps: - obtain, for an initial acquisition time, a first set of parameters associated with a virtual object, said virtual object being associated with a detection of the neighboring vehicle in the environment, - detecting, from at least a first image acquired at a first acquisition time subsequent to the initial acquisition time, an object having a vertical plane extending laterally relative to the main vehicle, said object being associated with a second set of parameters, - if a first criterion is satisfied, ensuring the tracking of the neighboring vehicle between the initial acquisition time and the first acquisition time by updating the first set of parameters from the second set of parameters, said update comprising, if a second criterion is satisfied, an estimation of a length associated with the virtual object.

[0010] Therefore, the proposed method advantageously allows the tracking of a neighboring vehicle over time. In particular, such tracking is possible even when the neighboring vehicle is no longer fully visible in the field of view of the camera and the detected portion of the vehicle corresponds only to a lateral portion of the vehicle (e.g., a front face of the neighboring vehicle is no longer visible in the field of view). Furthermore, the method further allows the updating of parameters associated with the tracked vehicle, in particular a length associated with the tracked vehicle. The method then allows an estimation of the actual dimensions of the neighboring vehicle, which ensures greater tracking accuracy and therefore increased safety and assistance in road traffic.

[0011] An initial acquisition time may refer to a time at which the camera's field of view captures an initial situation of the environment, and in particular of the neighboring vehicle relative to the main vehicle. At such an initial acquisition time, an initial image may be captured in which the neighboring vehicle may be fully included in the field of view (e.g., a front face of the neighboring vehicle may be visible in the field of view). A first acquisition time subsequent to the initial acquisition time may refer to an acquisition time subsequent to the Initial acquisition time, during which the movements of the neighboring vehicle and the main vehicle have changed the situation captured by the field of view. In particular, the neighboring vehicle may have advanced relative to the main vehicle, such that the front face of the neighboring vehicle is no longer visible in the field of view. The field of view may include, for example, only a lateral portion of the neighboring vehicle. At the first acquisition time, the method detects an object (lateral) that may or may not correspond to the neighboring vehicle.

[0012] By first and second parameter sets, reference may be made to one or more parameters for representing the objects detected in the environment. Elements of these parameter sets may, for example, include a position (or a set of positions), a velocity, an acceleration. These parameter sets may be associated with three-dimensional coordinates. In particular, these parameter sets may also be associated with sets of pixels (or two-dimensional coordinates) in the acquired images.

[0013] By a virtual object, reference may be made to a model of the neighboring vehicle as detected in the environment. Such a model may be represented in two dimensions on an image, or may be represented as a set of positions in the environment representing an occupancy of the neighboring vehicle as estimated by the method.

[0014] According to another aspect, a device is proposed configured to provide a driving assistance function for a main vehicle by ensuring tracking of a neighboring vehicle present in an environment of a main vehicle, said device being connected to at least one camera on board the main vehicle and capable of acquiring images of a scene surrounding the main vehicle according to at least one field of view and at acquisition times, said device being configured to ensure tracking of the neighboring vehicle by implementing the proposed method.

[0015] According to another aspect, there is provided a non-transitory recording medium readable by a computer on which is recorded a program for implementing the proposed method, when this program is executed by a processor.

[0016] The features set out in the following paragraphs may, optionally, be implemented, independently of each other or in combination with each other:

[0017] In one embodiment, the first criterion is satisfied if a distance between at least a first element of the first set of parameters and a second element of the second set of parameters does not exceed a first predefined threshold.

[0018] Therefore, the method makes it possible to test whether the object detected at the first acquisition time is likely to belong to (or correspond to) the virtual object, and therefore to the detected neighboring vehicle. If such a first criterion is satisfied, it is then considered that the first object and the virtual object can be associated and correspond to the neighboring vehicle being tracked.

[0019] By a distance, we can refer to a gap, a difference or a mathematical distance (eg, a Mahonois distance) allowing to quantify a difference in magnitude between the two elements.

[0020] By a first element and a second element, reference may be made to two homogeneous, comparable quantities, characterizing respectively the virtual object and the detected object. Such a first / second element may for example be a position or a speed. In other words:

[0021] in one embodiment, the first set of parameters including a first position associated with the virtual object and the second set of parameters including a second position associated with the detected object, and the first criterion is satisfied at least if a difference between the first position and the second position does not exceed a first predefined position threshold; or else:

[0022] in one embodiment, the first set of parameters including a first speed associated with the virtual object and the second set of parameters including a second speed associated with the detected object, and the first criterion is satisfied at least if a difference between the first speed and the second speed does not exceed a first predefined speed threshold.

[0023] In one embodiment, a first set of pixels and a second set of pixels being respectively associated with the virtual object and the object detected on the first acquired image, and the first criterion is satisfied if a distance between said first set of pixels and said second set of pixels does not exceed a first predefined threshold.

[0024] Therefore, the method advantageously makes it possible to test whether the virtual object and the detected object can be associated in the image. In particular, the method takes into account a margin of uncertainty (for example linked to the imprecision of obtaining the virtual object and / or to the detection of the object): the virtual object and the detected object can be associated even if they do not overlap (i.e., the first predefined threshold can be greater than 0).

[0025] By a distance between sets of pixels, it can be referred to a gap or mathematical distance separating two nearest pixels each belonging to a set of pixels. Such a distance is considered zero if the intersection between the two sets of pixels is not zero.

[0026] In one embodiment, the first set of parameters including a first position associated with the virtual object and the second set of parameters including a second position associated with the detected object, and wherein the second criterion is satisfied at least if a difference between the first position and the second position does not exceed a second predefined position threshold.

[0027] Therefore, the method advantageously makes it possible to determine criteria for updating the parameters associated with the neighboring vehicle being tracked. Indeed, the relevance of such an update, in particular of the estimated length of the neighboring vehicle, may depend on a situation of the virtual object in the field of view, such as for example the occupation of such a virtual object in the field of view.

[0028] In one embodiment, a first set of pixels being associated with the virtual object on the first acquired image, the second criterion is satisfied if at least one of the following elements is satisfied:

[0029] - a size of the first set of pixels does not exceed a second predefined threshold,

[0030] - a ratio between the size of the first set of pixels and a number of pixels occupied by the neighboring vehicle when a front face of the neighboring vehicle is visible in the field of view does not exceed a predefined ratio.

[0031] Therefore, the method advantageously makes it possible to determine the relevance of an update of the length associated with the virtual object according to the number of pixels occupied by the virtual object in the acquired image, and / or the percentage of the size of the virtual object included in the image. Indeed, if the virtual object is still “too” included in the image, an estimation of its length is not considered relevant.

[0032] The number of pixels occupied by the neighboring vehicle when a front face of the neighboring vehicle is visible in the field of view can for example be obtained at the initial acquisition time, by quantifying the number of pixels occupied by the virtual object at such an instant.

[0033] In one embodiment, the first set of parameters including a first position associated with the virtual object and the second set of parameters including a second position associated with the detected object, the first set of parameters including first dimensions associated with the virtual object, the estimation of the length includes an extension of a dimension among the first dimensions by an amount relating to a deviation between the first position and the second position.

[0034] Therefore, the method advantageously proposes to update the length of the virtual object using the parameters of the detected object (considered at this stage as belonging to the neighboring vehicle). In particular, an extension of the first dimensions may correspond to the extension of an estimated length associated with the neighboring vehicle being followed. In addition, the method advantageously makes it possible to update the estimated length of the virtual object even when the detected object does not coincide precisely with the virtual object.

[0035] In one embodiment, the first position and the second position correspond, in a predefined coordinate system, to sets of coordinates associated with portions of the environment respectively occupied by the virtual object and the detected object.

[0036] Therefore, the method makes it possible to compare and quantify distances between the virtual object and the detected object in the environment as in the image, when these objects correspond to sets of positions (of optical flows for example). In other words, in the environment, the virtual object and the detected object can each correspond to a set of positions reflecting the space occupied by each object. In the image, these sets of coordinates (or positions) can correspond to sets of pixels occupied on the image by each of the objects as captured in the field of view.

[0037] In one embodiment, the tracking of the neighboring vehicle is ensured for a plurality of acquisition times by repeating the steps of detection, updating and estimation for a plurality of objects detected from a plurality of images acquired at the acquisition times of said plurality of acquisition times.

[0038] Therefore, the proposed method makes it possible to track the neighboring vehicle on a plurality of acquired images. In addition, the method also makes it possible to test the association of the virtual object with several objects detected on the images (such detected objects may correspond to different sections of the neighboring vehicle or to other elements / objects present in the environment and distinct from the tracked vehicle).

[0039] In one embodiment, the tracking of the neighboring vehicle ends at a final acquisition time when a probability of existence in the field of view associated with the neighboring vehicle is less than a predefined threshold, said probability of existence depending at least on a difference between the initial and final acquisition times and on the first and second sets of parameters.

[0040] Therefore, the proposed method advantageously makes it possible to optimize the tracking resources of a vehicle by delimiting the tracking time of each detected vehicle. For example, from an initial acquisition time during which a given vehicle is detected for the first time and an estimate of its speed, the method can estimate an evolution of such a vehicle in the environment (from hypotheses of movement of the main vehicle and the tracked vehicle), so as to estimate a duration beyond which the tracked vehicle will no longer be in the field of view. Thus, the method can advantageously track a plurality of objects in the environment by optimizing the use and storage of the computing resources required during the association and / or creation of detected objects. Such an embodiment proves particularly advantageous when a plurality of vehicles are tracked and a newly detected object can potentially be associated with each of these tracked vehicles. Brief description of the drawings

[0041] Other features, details and advantages will become apparent upon reading the detailed description below, and upon analyzing the attached drawings, in which:

[0042] [Fig. 1] Figure 1 shows a schematic of a main vehicle according to one embodiment.

[0043] [Fig. 2] Figure 2 shows an aerial view of a scene surrounding the main vehicle according to one embodiment.

[0044] [Fig. 3] Figure 3 shows a diagram of a driving assistance device according to one embodiment.

[0045] [Fig. 4] Figure 4 shows steps of a method for tracking a vehicle according to one embodiment.

[0046] [Fig. 5] Figure 5 shows a shot of a scene surrounding the main vehicle according to one embodiment.

[0047] [Fig. 6] Figure 6 shows a view of a scene surrounding the main vehicle according to an embodiment.

[0048] [Fig. 7] Figure 7 shows a view of a scene surrounding the main vehicle according to an embodiment. Description of embodiments

[0049] Reference is made to Figure 1. Figure 1 schematically shows a main vehicle VP. The main vehicle VP can be a motor vehicle. There is no limitation on the type of vehicle to which the main vehicle VP belongs. The main vehicle VP can for example be a special, utility, industrial vehicle, and can for example correspond to a car, a van, a motorcycle, a truck or a bus. The The main vehicle (VP) can also be a towed vehicle, for example a trailer, a semi-trailer or a caravan. The dimensions of the main vehicle (VP) can, for example, be between 2 and 20 meters in length, between 0.5 and 5 meters in width and between 1 and 5 meters in height.

[0050] The main vehicle (VP) is equipped with at least one on-board system providing a plurality of functions or applications of the main vehicle (VP). Such functions may, for example, correspond to cruise control, power steering, automated airbag deployment, automatic headlight adjustment, etc.

[0051] The main vehicle VP is in particular equipped with an on-board system allowing detection of objects surrounding the main vehicle VP, for example as part of a driving assistance function such as obstacle detection, lane change assistance or automatic lane change. For this, the on-board system of the main vehicle VP comprises a driving assistance device 2. The device 2 is capable of providing a driving assistance function on the basis of a plurality of shots (or images) of the environment ENV of the main vehicle VP. For this, the device 2 is in particular connected to one or more vision sensors such as a camera or a camera 1 (the vision sensor will be considered to be a camera 1 in the remainder of the description). The device 2 may also be capable of transmitting or communicating data, for example relating to the driving assistance function provided.For this, the device 2 can be connected to a communication interface 50. In a particular embodiment, the interface 50 can be integrated into the device 2. Such a communication interface 50 can for example be a human-machine interface. The interface 50 can include a display screen, a touch screen, a dashboard, and / or even a speaker. The data transmitted by the device 2 to the interface 50 can for example correspond to assistance information indicating to the driver of the main vehicle VP whether or not he can change lanes.

[0052] Reference is now made to Figure 2. Figure 2 illustrates an ENV environment (or scene), in which the main vehicle VP is located. Figure 2 is an aerial view of such an ENV scene.

[0053] The ENV environment can be defined in a three-dimensional frame (X,Y,Z), called a "world frame" as illustrated in Figures 1 and 2. The origin of such a world frame (X,Y,Z) is predefined and fixed in the ENV environment.

[0054] The main vehicle VP is considered to be moving in the ENV scene. Such an ENV scene corresponds, for example, to a road or a motorway made up of several traffic lanes. These traffic lanes may in particular be parallel to each other. to others, as represented by the vertical dotted lines linked in Figure 2. Figure 2 illustrates for example three traffic lanes, the main vehicle VP being located on the middle traffic lane. The main vehicle VP is considered to be moving in the main direction X of the (X,Y,Z) reference frame, such a main direction being called the longitudinal direction. Such a movement is shown diagrammatically in Figures 1 and 2 by an arrow attached to the main vehicle VP. The main vehicle VP can also have a movement in a Y direction of the (X,Y,Z) reference frame, called the lateral direction. Such a movement in the Y direction, called a lateral movement (or displacement), can for example take place when the main vehicle VP changes traffic lane. A displacement in the Z direction, that is to say in height, of the main vehicle VP is considered absent or negligible.The main vehicle VP is therefore considered to be kept on the ground and therefore has a height in the constant Z direction, corresponding to a predefined dimension of the main vehicle VP. In one embodiment, such a height in the Z direction of the main vehicle VP may vary on the order of a centimeter or a decimeter, such a variation in height in the Z direction being for example linked to shock absorbers of the main vehicle VP and / or to reliefs or roughness present in the ENV environment (in particular on the traffic lane of the main vehicle VP). In the remainder of the description, such a height in the Z direction of the main vehicle VP is considered to be known.

[0055] The scene ENV surrounding the main vehicle VP also includes other OV elements and vehicles VV1, VV2, VV3. The vehicles VV1, VV2, VV3 are neighboring vehicles of the main vehicle VP. Such neighboring vehicles VV1, VV2, VV3 are, like the main vehicle VP, motor vehicles moving in the scene ENV. Such neighboring vehicles VV1, VV2, VV3 may in particular have dimensions and movement characteristics (in terms of speed or acceleration for example) similar to or different from the main vehicle VP. For example, figure 2 may represent a main vehicle VP moving on a motorway lane and neighboring vehicles VV1, VV2, VV3 traveling on the motorway lanes neighboring the lane taken by the main vehicle VP. OV objects are neighboring elements of the main vehicle VP and can refer to any distinct element of a neighboring vehicle VV1, VV2, VV3.A neighboring element OV can, for example, correspond to an obstacle located in the scene such as a central reservation separating two traffic lanes, an indication sign or even a bird flying in the ENV scene.

[0056] As illustrated in Figure 2, the main vehicle VP is considered to be equipped with a camera 1 positioned at the rear of the main vehicle VP. In another embodiment (not shown in Figure 2), the camera 1 may be positioned at the front of the vehicle. main vehicle VP or several cameras 1 can be positioned both at the front and at the rear of the main vehicle VP. In the context of the present description, the camera 1 is considered positioned at the rear of the main vehicle VP and the field of view FOV is a rear field of view of the main vehicle VP. The camera 1 makes it possible to capture images (or shots) of the ENV scene of the main vehicle VP according to a field of view FOV (or in English "Field of View"). Such a field of view FOV depends in particular on the type of camera 1 used and the positioning of the camera 1 in (or on) the main vehicle VP. With reference to FIG. 2, the field of view FOV covers a part of the ENV scene, so that a limited portion of the ENV scene is captured according to the field of view FOV.Thus, in Figure 2, the field of view FOV represented covers a portion of the traffic lane in which the main vehicle VP is located and respective portions of the neighboring traffic lanes. In particular, at a time T1 considered in Figure 2, the field of view FOV covers a portion of the scene ENV in which the neighboring vehicle VV2 is entirely located and a portion of the scene ENV in which the neighboring vehicle VV1 is partially located (grayed area not hatched). For example, as illustrated in Figure 2, a left rear end of the neighboring vehicle VV1 (for example including the left rear wheel of the neighboring vehicle VV2) belongs to the field of view FOV. However, a reference point W, for example located at the middle of the front face of the vehicle, is not in the field of view FOV in Figure 2.For comparison, the same neighboring vehicle VV1 is represented at an initial time T0 preceding the time T1 considered: such a neighboring vehicle VV1 is represented in dotted lines in Figure 2. At such an initial time T0, the neighboring vehicle VV1 is fully visible in the field of view FOV of the camera 1. In particular, at such an initial time T0, the reference point, noted W. n , for example located at the middle of the front face of the neighboring vehicle VV1, is visible in the field of view FOV.

[0057] At the instant T1 considered, subsequent to the initial time T0, a portion of the neighboring vehicle VV1 (including in particular the reference point W) is no longer visible in the field of view FOV of the main vehicle VP. For example, the situation of the neighboring vehicle VV1 represented in Figure 2 between the instants T0 and T1 may correspond to the case of overtaking on the right of the main vehicle VP by the neighboring vehicle VV1.

[0058] In Figure 2, the neighboring vehicle VV3 is not visible in the field of view FOV of the main vehicle VP. The non-visible parts of the neighboring vehicles VV1, VV3 in the field of view FOV are represented in Figure 2 by striped areas. A neighboring object OV, for example a bird, may be visible in the field of view FOV.

[0059] In the context of an ENV scene as represented in figure 2 according to the (X,Y,Z) reference frame, it is for example considered that the main vehicle VP moves at a speed main V vp known. To facilitate the rest of the description, such a main speed can be considered to be of constant VVP standard. The neighboring vehicle VV1 is considered to be moving at a neighboring speed unknown to the driving assistance device 2 of the main vehicle VP. In the embodiment described below, it can be considered that the VP standard of the neighboring speed is higher than the VVP standard of the main speed V^, for example in the case of a neighboring vehicle VV1 overtaking the main vehicle VP on a neighboring lane. Under such an assumption and under the assumption that the neighboring speed of the neighboring vehicle VV1 remains substantially constant over a considered time interval, like the main speed of the main vehicle VP, the distance difference in the longitudinal direction X between the main vehicle VP and the neighboring vehicle VV1 will vary over time, so that the neighboring vehicle VV1 will for example approach (in the longitudinal direction X) the main vehicle VP while overtaking and then will move away (in the longitudinal direction X) from the main vehicle VP once overtaking has been carried out. Thus, at an initial instant T0 (or initial acquisition time T0) considered, the neighboring vehicle VV1 can be fully visible in the rear field of view FOV of the main vehicle VP. As overtaking is implemented by the neighboring vehicle VV1, the neighboring vehicle VV1 can gradually "exit" from the rear field of view FOV, until it is completely outside the field of view FOV (like for example the neighboring vehicle VV3).Such an evolution of the neighboring vehicle VV1 in the field of view FOV is for example represented between figure 5 on the one hand, and figures 6 and 7 on the other hand.

[0060] Figure 5 represents an image acquired by camera 1 at an initial acquisition time T0. At such an initial acquisition time T0, the neighboring vehicle VV1 is entirely in the field of view FOV. In particular, a front face of the neighboring vehicle VV1 is visible in the field of view FOV.

[0061] Figures 6 and 7 represent images acquired by the camera 1 according to two distinct embodiments at a first acquisition time T1, subsequent to the initial acquisition time T0. At such a first acquisition time T1, the neighboring vehicle VV1 has advanced relative to the main vehicle VP in the environment ENV, such that the neighboring vehicle VV1 gradually arrives at the level of the main vehicle VP. In particular, the front face of the neighboring vehicle VV1 is no longer visible in the field of view FOV. Only a lateral portion, shown in Figures 6 and 7, including for example the rear-left wheels of the neighboring vehicle VV1 are visible at the first acquisition time T1. Indeed, the frame shown in each of Figures 6 and 7 delimits the size of the shot and corresponds to the field of view FOV. Thus, the elements represented outside the frame, which will be described later in the description, are illustrated for explanatory purposes and are not visible depending on the point of view of the FOV field of view.

[0062] In the context of a driving assistance function provided by the device 2, for example aiming to assist the main vehicle VP in a lane change maneuver, a process for tracking vehicles detected in the environment ENV of the main vehicle VP is required, so that the main vehicle VP does not move onto a neighboring traffic lane if it risks colliding with one of the surrounding vehicles. Existing techniques for detecting and tracking vehicles surrounding the main vehicle VP are based in particular on the use of classifiers, based for example on convolutional neural network (CNN) methods, K-nearest neighbors (KNN) or support vector machines (SVM).Such classifiers allowing the detection and tracking of vehicles surrounding the main vehicle VP rely in particular on the detection and classification of a frontal (front or rear) view (or face) of the vehicles surrounding the main vehicle VP. The use of such classifiers may in particular comprise a learning phase based on a plurality of images representing frontal views of various types of vehicles. Thus, with reference to FIG. 2, a front frontal view of the neighboring vehicle VV2 belonging to the field of view FOV of the camera 1, the detection and tracking of the neighboring vehicle VV2 can be implemented on the basis of existing classification techniques. The same applies to the neighboring vehicle VV1 in the image of FIG. 5 at the initial acquisition time T0.

[0063] However, the positions of the neighboring vehicle VV1 shown in Figure 2 at time T1 and Figures 6 and 7 correspond to situations in which the existing classifiers are unable to effectively track the neighboring vehicle VV1, or at least not without incurring significant computational costs and / or time, due to the lack of a visible frontal view of the neighboring vehicle VV1 in the viewpoints of Figures 6 and 7 (the reference point W in Figure 2 positioned on the front face of the neighboring vehicle VV1 no longer being visible in the field of view FOV of camera 1 at these stages).These situations, however, correspond to critical situations during which the main vehicle VP could collide with the neighboring vehicle VV1 if the main vehicle VP changes position and moves into the lane of the neighboring vehicle VV1, in the absence of lane change assistance information indicating that the neighboring vehicle VV1 is detected near such a position.

[0064] A method for tracking the neighboring vehicle VV1 is then proposed and described in figure 4 to track the neighboring vehicle VV1, and in particular at the stage of the situations represented in Figures 2, 6 and 7. Such a tracking method can in particular be described as a tracking maintenance method for a vehicle leaving the field of view FOV, in the sense that the vehicle, initially detected, can continue to be tracked even when a front face of the vehicle is outside the (and therefore no longer visible in the) field of view FOV. Such a method for tracking the neighboring vehicle VV1 is implemented by a driving assistance system shown in Figure 3.

[0065] Reference is now made to Figure 3. Figure 3 represents a diagram of an on-board system of a main vehicle VP. In particular, such an on-board system corresponds to a driving assistance system for the main vehicle VP. The driving assistance system of the main vehicle VP makes it possible in particular to provide a lane change assistance or automatic lane change function for the main vehicle VP, when the main vehicle VP is moving on a traffic lane as shown for example in Figure 2.

[0066] The system firstly comprises the driving assistance device 2. The device 2 may itself comprise a unit 20 for detecting objects on acquired images, a unit 30 for comparing the objects detected by the unit 20 with existing virtual objects and a unit 40 for updating parameters associated with the existing virtual objects for tracking elements associated with the existing virtual objects in the ENV environment.

[0067] The driving assistance device 2 is further connected, on the one hand, to a vision sensor of the camera or camera type 1. The device 2 may also comprise an input unit (not shown in FIG. 3) allowing the device 2 to receive in substantially real time a data stream from the camera 1. Such a data stream corresponds to a discrete or continuous succession of images (or shots) of the scene ENV according to the field of view FOV of the camera 1. Each image received is associated with an acquisition time of the image by the camera 1. Each image may be time-stamped.

[0068] On the other hand, the driving assistance device 2 can be connected to a communication interface 70. Such an interface 70 can correspond to a human-machine interface integrated into the on-board system of the main vehicle VP. Such an interface 70 can also be integrated into the device 2. The communication interface 70 can also be a remote interface. The communication interface 70 can comprise a display screen, a touch screen, a dashboard or even a loudspeaker, making it possible to transmit, for example by visual, haptic and / or audible information, indications relating to assistance in driving the main vehicle VP. In particular, the communication interface makes it possible to transmit a detection status of a vehicle neighboring the main vehicle VP or even an indication relating to a possibility of changing lane of the main vehicle VP. Such information transmitted by the interface 70 may for example correspond to an estimated aerial visual representation of the respective positions of the main vehicle VP and the neighboring elements belonging to the field of view FOV in real time (eg, as represented in FIG. 2), an audio stimulus alerting of a risk of collision, or even a superposition of a virtual object highlighted and updated in real time on a stream of images coming from the camera 1, as illustrated for example in FIGS. 5, 6 and 7.

[0069] The units 20, 30 and 40 of the device 2 may each comprise a processing circuit including at least one processor (21, 31, 41) and a memory unit (22, 32, 42) in order to implement one or more steps of the method for tracking a neighboring vehicle VV1, which will be described in FIG. 4. In particular, each processing unit 20, 30, 40 of the device 2 may rely on data processed and / or obtained by other units in order to implement one or more steps of the method for tracking a neighboring vehicle VV1, as illustrated by the arrows in FIG. 3.

[0070] Reference is now made to Figure 4. Figure 4 illustrates a succession of steps for implementing a method for tracking a previously detected neighboring vehicle VV1. Such a tracking method can be implemented by a system including a device 2 for assisting the driving of a main vehicle VP as shown in Figure 3.

[0071] Throughout the method described below, the device 2 receives a plurality of images (or shots) associated with respective acquisition times of the images from the camera 1. Such images may in particular be received continuously, for example via a video stream. In such a case, the device 2 may discretize the received video stream so as to obtain a set of discrete time-stamped images associated with respective acquisition times.

[0072] In a step 400, the neighboring vehicle VV1 is previously detected on an initial image, acquired at an initial acquisition time T0. Such an initial image is for example shown in FIG. 5. At such an initial acquisition time T0, the neighboring vehicle VV1 is fully visible in the field of view FOV. In particular, the front face of the neighboring vehicle VV1 is visible in the field of view FOV. The neighboring vehicle VV1 can then be detected by conventional methods, for example by a vehicle classifier based on recognition of different types of vehicle front faces. In another embodiment, the neighboring vehicle VV1 can be detected via other methods, for example based on optical flow detection. In step 400, the neighboring vehicle VV1 is then detected on the initial image and a first position p nassociated with the neighboring vehicle VV1 can be determined in the image frame (y,z) and / or in the world frame (X,Y,Z). In one embodiment, such a first position p n associated with the neighboring vehicle VV1 may correspond to a set of coordinates associated with a first portion of the environment ENV occupied by the neighboring vehicle VV1. Such a first set of coordinates may be expressed in three dimensions according to the world frame (X,Y,Z), such that the first position reflects an occupation space of the environment ENV estimated to be occupied by the detected neighboring vehicle VV1. Equivalently, such a first set of coordinates may also be expressed in two dimensions according to the image frame (y,z) such that the first position reflects an occupation space of the initial image occupied by the neighboring vehicle VV1 as detected on the initial image. The conversion of the first position p nbetween the image (y,z) and world (X,Y,Z) coordinates can be obtained from the pinhole model and hypotheses (eg, flat world hypothesis). Alternatively, the first position p n associated with the neighboring vehicle VV1 can correspond to the position of a first reference point W n positioned on the detected neighboring vehicle VV1, for example on the front of the neighboring vehicle VV1. Such a landmark W n can be expressed in the world frame (X,Y,Z) or, equivalently, be a (virtual) reference point w n (corresponding to a pixel for example) expressed in the image reference point (y,z). Such a reference point w n is for example represented in Figure 5.

[0073] More generally, in step 400, the detected neighboring vehicle VV1 may be associated with a first set of parameters, such a first set of parameters being able for example to include one or more parameters from among: - the first position p n , - a first speed v n reflecting a speed of the neighboring vehicle VV1 detected in the environment ENV, - first dimensions s n including for example first values ​​of length, width and / or height associated with the neighboring vehicle VV1 detected in the ENV environment, - a first iTTC collision time n associated with the neighboring vehicle VV1 detected, - a first lateral distance d n reflecting for example a lateral deviation (i.e. in the so-called lateral direction Y in the world frame (X,Y,Z) or y in the image frame (y,z)) separating the detected neighboring vehicle VV1 from the main vehicle VP.

[0074] Such parameters can for example be determined or estimated by one or more classifiers from one or more initial acquired images associated with the initial acquisition time T0. Such parameters can also be estimated from measurements made on the initial acquired images.

[0075] From such a first set of parameters, the neighboring vehicle VV1 detected in step 400 can be associated with a virtual object OV n . Such a virtual object OV n can for example correspond to a set of pixels (and therefore of two-dimensional coordinates in the image frame (y,z)) superimposable on the space occupied by the neighboring vehicle VV1 in the initial image. In particular, the virtual object OV n can also correspond to a three-dimensional virtual object in the world frame (X,Y,Z) allowing the space occupied by the neighboring vehicle VV1 to be modeled in the environment ENV. Such a virtual object OVn is characterized by one or more kinematic parameters corresponding to the first set of parameters and making it possible to delimit the neighboring vehicle VV1 as detected in time (here, at the initial acquisition time T0) and in space (here, in the environment ENV). Thus, at the initial acquisition time, the neighboring vehicle VV1 is detected and modeled by the virtual object OV n (or equivalently, by a first set of parameters).

[0076] In a step 410, an object OBJ is detected on at least one of the acquired images, called the first image(s). With reference to FIGS. 6 and 7 representing two possible variants of the detected object OBJ, such a detected object OBJ is represented at a first acquisition time T1. In one embodiment, the detected object OBJ is a so-called lateral object in the acquired image, in that the detected object OBJ1 belongs to a lateral plane of the image, such a lateral plane being for example parallel to the plane (X,Z) in the world reference frame (X,Y,Z). The detection of such a lateral object OBJ can in particular take place without a front face of the neighboring vehicle VV1 being visible in the field of view FOV. In particular, at the stage of step 410, the detected object OBJ corresponds to a non-inert body which is not a priori not yet assimilated to the neighboring vehicle VV1 previously detected.In other words, at the stage of step 410, the detected object OBJ corresponds to a body in motion in the environment ENV (and therefore, in the world frame of reference (X,Y,Z)) and therefore having at least one speed, called second speed v* (for example, a speed called longitudinal corresponding to a movement along the X axis in the world frame of reference (X,Y,Z), a speed called lateral corresponding to a movement along the Y axis and / or a rotation speed).

[0077] The object OBJ can be detected in step 410 on an image (for example, the image of FIG. 6 or the image of FIG. 7) by processing one or more successively acquired images. In one embodiment, the object OBJ can for example be detected by a classifier capable of detecting a lateral object OBJ on the acquired image, for example a classifier detecting a lateral portion of a vehicle, such as a vehicle wheel detector. Alternatively, the object OBJ can be detected by homographies on at least two successively acquired images, so as to identify one or more related sets, or optical flows, of relatively homogeneous speed on the images in a delimited time interval. According to such a variant, the detection of the object OBJ1 then includes determining a displacement of pixels at a substantially common speed from one image to the next, or optical flow. Each optical flow can be determined by matching points between several successive images, for example by the Lucas-Kanade method. The determined optical flows can in particular be segmented into one or more segments having a common point or vanishing line and belonging to lateral planes in the field of view FOV. In other words, the acquired image can be segmented into one or more segments, each segment defining a grouping of optical flows (and therefore a grouping of pixels having substantially the same speed and describing the same movement in a given time interval).

[0078] Thus, at step 410, the detected object OBJ is a set of pixels that can be formed by an instance, a contour, a polygon, or a rectangle (or in English, a "bounding box") detected and classified by an object segmentation classifier in the acquired image, for example. Alternatively, the detected object OBJ is a set of pixels that can be formed by an optical flow or segment including several optical flows. In particular, the set of pixels forming the detected object OBJ can be a set of coplanar pixels parallel to a lateral plane in the world coordinate system (X, Y, Z).

[0079] In particular, the detected object OBJ can be associated with a second set of parameters, such a second set of parameters can for example include one or more parameters among: - a second position p*, - the second velocity v’, - second dimensions s* including for example second values ​​of length, width and / or height associated with the detected object OBJ in the environment ENV, - a second collision time ITTC* associated with the detected object OBJ, - a second lateral distance d n reflecting for example a lateral deviation (i.e. in the so-called lateral direction Y in the world frame (X,Y,Z) or y in the image frame (y,z)) separating the detected object OBJ from the main vehicle VP.

[0080] Like the first position p n associated with the virtual object OV n, the second position p* associated with the detected object OBJ may correspond to a set of coordinates (in the world frame (X,Y,Z) and / or in the image frame (y,z)) representing a portion of the environment ENV occupied by the detected object OBJ or, alternatively, correspond to the position of a reference point belonging to the detected object OBJ. Such a reference point may for example correspond to a pixel located closest to the ground.

[0081] The second set of parameters can for example be determined or estimated by one or more classifiers from the first acquired images associated with the first acquisition time T1. Such parameters can also be estimated from measurements made on the first acquired images.

[0082] At step 420, the device therefore obtains, on the one hand, a virtual object OV nassociated with a first set of parameters representing the neighboring vehicle VV1 initially detected and on the other hand, a detected object OBJ associated with a second set of parameters.

[0083] At a step 420, a consistency (or plausibility) test is implemented, so as to determine whether the detected object OBJ and the virtual object OV n corresponds to the same element in the ENV environment, namely to the neighboring vehicle VV1. For this, the consistency test can include the test of a first criterion. Such a first criterion can for example correspond to a comparison between the first position p n associated with the virtual object OV n and the second position p* associated with the detected object OBJ. The first criterion can then be satisfied if the difference between the first position p n and the second position p* is less than a first predefined position threshold. In particular, if the first position p nand the second position p* correspond to the respective coordinates of a reference point W n and a reference point belonging respectively to the virtual object OV n and to the detected object OBJ, the first position threshold can be set to a lateral coordinate (i.e., along the X axis) of the first position p n and the second position p*, respectively noted p n (X and p*(X). In other words, the consistency test can be satisfied if:

[0084] Math.1 |p„(X) - p*(X)| < Pi where: - p„(X) is the coordinate of the first position p n along the X axis in the world frame (X,Y,Z), - p*(X) is the coordinate of the second position p* along the X axis in the world frame (X,Y,Z), - P ± is the first predefined position threshold.

[0085] Alternatively, if the first position p nand the second position p* correspond to sets of coordinates (i.e., sets of points or pixels reflecting the spaces respectively occupied by the virtual object OV n and to the detected object OBJ), the first criterion can then be satisfied if the difference between two closest points (or pixels), the two points belonging respectively to the two sets of coordinates, is less than the first position threshold.

[0086] Alternatively, the first criterion may also relate to other parameters of the parameter sets than the positions p n , p*. In one embodiment, the first criterion can be satisfied if the difference between the first speed v n and second speed v* does not exceed a first predefined speed threshold.

[0087] Alternatively, the first criterion may be satisfied if a determined Mahalanobis distance between the first set of parameters and the second set of parameters does not exceed a first predefined threshold.

[0088] The first criterion may also relate to a comparison or a combination of the above comparisons.

[0089] If, at the end of step 420, the detected object OBJ is not determined as belonging to the pre-existing virtual object OV n (in other words, if the first criterion is not satisfied), a step 421 may be implemented. Such a step 421 may notably include the creation of a new virtual object distinct from the virtual object OV nassociated with the neighboring vehicle VV1. In other words, if the first criterion is not satisfied, it is determined that the detected object OBJ reflects the movement in the environment of a moving body distinct from the neighboring vehicle VV1, such a moving body may for example correspond to another vehicle than the neighboring vehicle VV1, to an obstacle in the environment ENV or to a bird for example. A new virtual object can then be created on the basis of the second set of parameters. In another embodiment, if the detected object OBJ is not determined to belong to the pre-existing virtual object OV n , the detected object OBJ can be ignored and steps 400 and 410 are repeated until a detected object OBJ is identified as belonging to the pre-existing virtual object OV n (in other words, the process may seek to track a particular VV1 vehicle).

[0090] On the other hand, if at the end of step 420, the detected object OBJ is determined as belonging to the pre-existing virtual object OV n (in other words, if the first criterion is satisfied), a step 430 is implemented.

[0091] At step 430, an update of the pre-existing virtual object OV n is implemented, so as to obtain an up-to-date virtual object Oy n+1. For this, the second set of parameters can be fed to a data estimator, for example to a Kalman filter of the Extended Kalman Filter (EKF) type. Such a data estimator can also obtain the first set of parameters (for example, fed by one or more classifiers of the system). On the basis of the parameters supplied to the data estimator, the device 2 can obtain, in step 430, an updated set of parameters. Such an updated set of parameters can in particular be estimated by the Kalman filter, on the one hand, from the second set of parameters provided and on the other hand, of a prediction estimated by the Kalman filter from at least the first set of parameters provided.

[0092] Thus, in step 430, the device 2 may obtain an updated set of parameters. In particular, such an updated set of parameters may include at least: - an up-to-date position p n+1 , determined from at least the first position p n and the second position p*, and - an up-to-date speed v n+1 , determined from at least the first speed v n and the second speed v*.

[0093] From such up-to-date settings, the virtual object OV n associated with the neighboring vehicle VV1 can be updated to an up-to-date virtual object OV n+1 , thus allowing tracking of the neighboring vehicle VV1 moving in the environment ENV between times T0 and T1. In particular, an updated reference point w n+1 can be associated with the updated virtual object OV n+1 . Such a landmark up to date w n+1 is for example represented in figures 6 and 7.

[0094] In most existing tracking methods, the dimensions associated with the neighboring vehicle VV1 being tracked (i.e. the first dimensions s n) correspond to pre-existing dimensions (e.g., those of a typical truck), so that the space occupied by a detected vehicle can be estimated from a landmark (e.g., W n ) and predefined dimensions. When updating the OV virtual object n , such dimensions are then preserved and the space occupied by the virtual object updated OV n+1 is moved accordingly.

[0095] In the context of the present disclosure, an update of the dimensions associated with the neighboring vehicle VV1 is also considered, so as in particular to allow an estimation of the actual length of the detected neighboring vehicle VV1.

[0096] For this, at a step 440, a length estimation test is implemented, so as to determine whether an update of the length of the pre-existing virtual object OV n will be performed. In step 440, a second criterion is tested.

[0097] Such a second criterion may for example include estimating the size of the portion of the neighboring vehicle VV1 still visible in the field of view FOV. For this, the space occupied by the virtual object OV n in the acquired image can be estimated (in number of pixels occupied in the image). In one embodiment, the lateral space occupied by the virtual object OV n in the acquired image can be estimated (in number of pixels occupied in the image). The second criterion can then be satisfied if the number of pixels occupied by the virtual object OV n in the image does not exceed a predefined pixel threshold, e.g. 100 pixels. Alternatively, the second criterion can be satisfied if the ratio between the number of pixels occupied by the virtual object OV n in the image considered (for example in figure 6 or 7) and the number of pixels occupied by the virtual object OV nin the initial image (in Figure 5) does not exceed a predefined percentage of pixels, for example 10 percent. More generally, other variants may make it possible to determine that the second criterion is satisfied if less than a given percentage of the entire virtual object OV n is visible in the image.

[0098] Indeed, an update of the length of the neighboring vehicle VV1 may be considered too imprecise if the neighboring vehicle VV1 belongs entirely or almost entirely to the field of view FOV. The perspective of the field of view FOV indeed biases the length associated with the virtual object OV n .

[0099] In one embodiment, the second criterion may be satisfied if the distance 5 between the first position p n associated with the virtual object OV nand the second position p* associated with the detected object OBJ does not exceed a second predefined position threshold. Like the first criterion previously described in step 420 according to one embodiment (for example, that of Math. 1), the second criterion can be satisfied if the spaces respectively occupied by the virtual object OV n and the detected object OBJ substantially overlap. The second predefined position threshold may be similar to or different from the first position threshold. In particular, the presence of the second position threshold means that the second criterion can be satisfied even if the virtual object OV n and the detected object OBJ have a zero intersection. Referring to Figure 6, there is shown an embodiment in which the gap between the virtual object OV n and the detected object OBJ is considered null in that the virtual object OV nand the detected object OBJ overlap, as represented by the hatched area in Figure 6. With reference to Figure 7, there is shown an embodiment in which the gap between the virtual object OV n and the detected object OBJ is non-zero in that the virtual object OV n and the detected object OBJ do not overlap: a non-zero gap 5, represented in Figure 7, exists between the virtual object OV n and the detected object OBJ. In particular, the tolerance of a deviation 5 less than the second predefined position threshold means that it is considered that the second criterion is satisfied if the detected object OBJ on the one hand, and the virtual object OV n fictitiously extended according to the lateral component, noted s n(X) , of a value corresponding to the second position threshold, on the other hand, overlap. Such a condition is considered to be satisfied in Figure 7. In one embodiment, the second criterion may be considered satisfied based on one or both of the previously described comparisons.

[0100] If, at the end of step 440, the second criterion is not satisfied, a step 451 is implemented. In step 451, the updated dimensions s n+1 associated with the neighboring vehicle VV1 are preserved and correspond to s n .

[0101] If, at the end of step 440, the second criterion is satisfied, a step 450 of estimating the updated length of the neighboring vehicle VV1 is implemented. In step 450, updated dimensions s n+1 associated with the neighboring vehicle VV1, distinct from the first dimensions s n are determined. In particular, an up-to-date lateral component, denoted s n+1(X), up-to-date dimensions n+1 is determined in step 450. For this, the second dimensions s* associated with the detected object OBJ are considered. In particular, the lateral coordinate associated with the pixel of the detected object OBJ furthest from the virtual object OV n is considered. Such a lateral coordinate then corresponds to the updated lateral coordinate of the end pixel of the updated virtual object OV n+1 . In other words, the lateral object OV n is extended in the lateral direction, until it laterally includes the detected object OBJ. The value of such an extension then corresponds to compensating for the difference (according to the lateral component) between the first position p n and the second position p*, so that the updated virtual object OV n+1is extended until it laterally occupies the space occupied by the detected object OBJ. Such an extension is shown in each of Figures 6 and 7, so that an up-to-date virtual object OV n+1 is obtained. In particular, the up-to-date dimensions n+1 associated with the updated virtual object OV n+1 were determined so as to allow an estimation of the length of the neighboring vehicle VV1.

[0102] Therefore, the neighboring vehicle VV1 can be tracked between times T0 and T1 while allowing an estimation of the length of the neighboring vehicle VV1.

Claims

Claims

1. Method for tracking at least one neighboring vehicle (VV1) present in an environment (ENV) of a main vehicle (VP), said main vehicle (VP) and said neighboring vehicle (VV1) being motor vehicles, the method being implemented by a device (2) configured to provide a driving assistance function for the main vehicle (VP), said device (2) being connected to at least one camera (1) on board the main vehicle (VP) and capable of acquiring images of a scene surrounding the main vehicle (VP) according to at least one field of view (FOV) and at acquisition times (T0, T1), the method comprising the following steps: - obtain (400), for an initial acquisition time (T0), a first set of parameters associated with a virtual object (OV n ), said virtual object (OV n ) being associated with a detection of the neighboring vehicle (VV1) in the environment (ENV), - detecting (410), from at least a first image acquired at a first acquisition time (T1) later than the initial acquisition time (T0), an object (OBJ) having a vertical plane extending laterally relative to the main vehicle (VP), said object (OBJ) being associated with a second set of parameters, - if (420) a first criterion is satisfied, ensuring the tracking of the neighboring vehicle (VV1) between the initial acquisition time (T0) and the first acquisition time (T1) by updating (430) the first set of parameters from the second set of parameters, said update (430) comprising, if (440) a second criterion is satisfied, an estimation (450) of a length associated with the virtual object (OV).

2. The method of claim 1, wherein the first criterion is satisfied if a distance between at least a first element of the first set of parameters and a second element of the second set of parameters does not exceed a first predefined threshold.

3. A method according to any preceding claim, the first set of parameters including a first position (p n ) associated with the virtual object (OV n ) and the second set of parameters including a second position (p*) associated with the detected object (OBJ), and in which the first criterion is satisfied at least if a difference between the first position (p„) and the second position (p*) does not exceed a first predefined position threshold.

4. A method according to any preceding claim, the first set of parameters including a first speed (v„) associated with the virtual object (OV n) and the second set of parameters including a second speed (v*) associated with the detected object (OBJ), and in which the first criterion is satisfied at least if a difference between the first speed (v„) and the second speed (v*) does not exceed a first predefined speed threshold.

5. A method according to any preceding claim, a first set of pixels and a second set of pixels being respectively associated with the virtual object (OV n ) and to the detected object (OBJ) on the first acquired image, and wherein the first criterion is satisfied if a distance between said first set of pixels and said second set of pixels does not exceed a first predefined threshold.

6. A method according to any preceding claim, the first set of parameters including a first position (p n ) associated with the virtual object (OV n) and the second set of parameters including a second position (p*) associated with the detected object (OBJ), and in which the second criterion is satisfied at least if a difference between the first position (p„) and the second position (p*) does not exceed a second predefined position threshold.

7. A method according to any preceding claim, a first set of pixels being associated with the virtual object (OV n ) on the first acquired image, and in which the second criterion is satisfied if at least one of the following is satisfied: - a size of the first set of pixels does not exceed a second predefined threshold, - a ratio between the size of the first set of pixels and a number of pixels occupied by the neighboring vehicle (VV1) when a front face of the neighboring vehicle (VV1) is visible in the field of view (FOV) does not exceed a predefined ratio.

8. A method according to any preceding claim, the first set of parameters including a first position (p n ) associated with the virtual object (OV n ) and the second set of parameters including a second position (p*) associated with the detected object (OBJ), the first set of parameters including first dimensions (s„) associated with the virtual object (OV n ), and wherein the length estimation includes an extension of a dimension (s„(X)) among the first dimensions (s„) by an amount relative to a gap between the first position (p„) and the second position (p*)-

9. Method according to any one of the preceding claims, the first set of parameters including a first position (p„) associated with the virtual object (OV n) and the second set of parameters including a second position (p*) associated with the detected object (OBJ), and in which the first position (p„) and the second position (p*) correspond, in a predefined coordinate system, to sets coordinates associated with portions of the environment (ENV) respectively occupied by the virtual object (OV n ) and the detected object (OBJ).

10. Method according to any one of the preceding claims, in which the tracking of the neighboring vehicle (VV1) is ensured for a plurality of acquisition times by repeating the steps of detection (410), updating (430) and estimation (450) for a plurality of objects detected from a plurality of images acquired at the acquisition times of said plurality of acquisition times.

11. Method according to any one of the preceding claims, in which the tracking of the neighboring vehicle (VV1) ends at a final acquisition time when a probability of existence in the field of view (FOV) associated with the neighboring vehicle VV1 is less than a predefined threshold, said probability of existence depending at least on a difference between the initial (T0) and final acquisition times and on the first and second sets of parameters.

12. Device configured to provide a driving assistance function for a main vehicle (VP) by ensuring tracking of a neighboring vehicle (VV1) present in an environment (ENV) of a main vehicle (VP), said device (2) being connected to at least one camera (1) on board the main vehicle (VP) and capable of acquiring images of a scene surrounding the main vehicle (VP) according to at least one field of view (FOV) and at acquisition times (T0, T1), said device (2) being configured to ensure tracking of the neighboring vehicle (VV1) by implementing the method according to one of claims 1 to 11 by a processor of said device (2).

13. Non-transitory recording medium readable by a computer on which is recorded a program for implementing the method according to one of claims 1 to 11 when this program is executed by a processor.

Images