EP1710767B1 - Procedure and system for detecting the presence of an obstacle and activation module for said system - Google Patents

Abstract

The method involves processing images acquired from images of a road segment to detect the presence of an impeding object e.g. damaged vehicles. Vehicles simultaneously present on the segment are counted from data collected by two detectors of the vehicles respectively placed at an entrance and an exit of the segment. An automatic activation of the acquisition of images is controlled based on the counting and an automatic activation of the image processing is controlled based on the counting. An independent claim is also included for an impeding object presence detecting system.

Description

The present invention relates to a method and a system for detecting the presence of an annoying object, and an activation module for this system.

EP-A-1,414,000 describes a method for detecting bottling on a road section.

There are methods for detecting the presence of an annoying object on a road section comprising a step (a) of image processing for detecting the presence of the annoying object from images of said section.

The image processing step is carried out continuously in order to quickly detect the presence of this possible annoying object.

The annoying object may be a stopped vehicle or a damaged vehicle or any other object present on the pavement of the section of road.

Image processing requires a lot of computing power and consumes a lot of energy. This significant energy consumption is particularly problematic when the image processing step is performed by a stand-alone road traffic beacon placed at the edge of the road section.

The invention aims to remedy this drawback by proposing a method for detecting the presence of an annoying object but consuming less energy or requiring a reduced computing power.

The subject of the invention is therefore a detection method according to claim 1.

The presence of an annoying object on a section of road results in a variation of the road traffic and, conventionally, by an increase in the number of motor vehicles simultaneously present on this section. In the above method, the image processing is not activated any more permanently, but only when it seems necessary, the moment when the image processing seems necessary being determined from the counting of the vehicles simultaneously present on the section. Thus, thanks to the above method, it limits the power consumption due to the image processing.

Embodiments of this detection method may include one or more of the features of the dependent claims.

• l'activation automatique de l'étape d'acquisition d'images en fonction du nombre de véhicules automobiles simultanément présents sur le tronçon de route permet également de limiter la consommation d'énergie et de décroître la quantité d'informations transmises entre un module d'acquisition d'images et un module de traitement d'images ;
• l'activation du traitement d'images en réponse à l'accroissement du nombre de véhicules dénombrés permet de détecter rapidement un objet gênant sans que le traitement d'images soit activé en permanence ;
• la fixation du résultat et/ou du seuil d'activation prédéterminé en fonction du nombre moyen de véhicules automobiles simultanément présents sur le tronçon de route accroît la robustesse du procédé de détection vis-à-vis de variations de l'intensité du trafic routier ;
• l'activation automatique de l'étape de traitement d'images, lorsqu'au moins l'un des détecteurs de véhicules est inopérant permet de détecter un objet gênant bien qu'un des détecteurs de véhicules ne soit pas utilisable ; et
• la combinaison de résultats obtenus à partir du traitement d'images et du comptage des véhicules permet d'estimer de façon plus fiable la probabilité qu'un objet gênant soit présent sur le tronçon.

These embodiments of the detection method also have the following advantages:

• the automatic activation of the image acquisition step as a function of the number of motor vehicles simultaneously present on the road section also makes it possible to limit the energy consumption and to decrease the amount of information transmitted between a module of image acquisition and an image processing module;
• the activation of the image processing in response to the increase in the number of vehicles counted makes it possible to quickly detect an annoying object without the image processing being permanently activated;
• setting the result and / or the predetermined activation threshold as a function of the average number of motor vehicles simultaneously present on the section of road increases the robustness of the detection method vis-à-vis variations in the intensity of road traffic;
• the automatic activation of the image processing step, when at least one of the vehicle detectors is inoperative can detect an object annoying although one of the vehicle detectors is not usable; and
• the combination of results obtained from image processing and vehicle counting makes it possible to estimate more reliably the probability that an annoying object will be present on the section.

The invention also relates to a detection system according to claim 8.

Embodiments of this detection system may include one or more of the features of the dependent claims.

The invention also relates to an activation unit that can be implemented in the method or the detection system above.

• la figure 1 est une illustration schématique d'un système de détection d'un objet gênant sur un tronçon de route ;
• la figure 2 est un organigramme d'un procédé de détection d'un objet gênant sur un tronçon de route ; et
• la figure 3 est un autre mode de réalisation du système de la figure 1.

The invention will be better understood on reading the description which follows, given solely by way of example and with reference to the drawings in which:

• Figure 1 is a schematic illustration of a system for detecting an annoying object on a road section;
• Figure 2 is a flow diagram of a method of detecting an annoying object on a road section; and
• Figure 3 is another embodiment of the system of Figure 1.

FIG. 1 represents a system 2 for detecting the presence of an annoying object on a road 4.

To simplify FIG. 1, only two successive sections 6 and 8 of the road 4 are illustrated. The following description of system 2 will be made only in relation to these two sections. Here, sections 6 and 8 are each divided into two portions 6 'and 6 "respectively, and 8' and 8".

The system 2 comprises a road traffic beacon placed at the entrance and exit of each section of road. Here, the system 2 comprises a beacon 10 at the entrance of the section 6, a beacon 11 common at the exit of the section 6 and the entrance of the section 8 and a beacon 12 at the exit of the section 8. The beacons 10 to 12 are, for example, all identical and only the tag 12 will be described here in detail.

The tag 12 comprises a vertical mas 14 at the upper end of which are fixed two cameras 16 and 18 for taking pictures. The apparatus 16 is turned towards the section 8 to take images of the portion 8 "of the section 8 while the apparatus 18 is turned towards the next section of the road 4.

The beacon 12 also comprises a detector 20 of vehicles capable of detecting the passage of a vehicle nearby on the road 4 to count the number of vehicles leaving the section 8. This detector is, for example, made using An acoustic sensor array 22. In FIG. 1, only three acoustic sensors 22 are shown for each detector.

The devices 16 and 18 as well as the various acoustic sensors 22 are connected to a local data processing circuit 24. For example, the circuit 24 is housed in an electrical cabinet placed at the foot of the house 14.

• un module 26 d'acquisition des images prises par les appareils 16 et 18,
• un module 28 de dénombrement propre à dénombrer les véhicules détectés par le détecteur 20 pendant un intervalle de temps δT donné,
• un module conventionnel 30 de traitement des images acquises par le module 26, propre à détecter automatiquement la présence d'un objet gênant sur le tronçon 8 à partir de l'analyse de ces images,
• un module 32 d'établissement du bruit ambiant à partir des mesures réalisées par les capteurs 22, lorsque aucun véhicule n'est présent sur le tronçon 8,
• un module 34 d'activation du module de traitement 30 en fonction du dénombrement établi par le module 28 ou d'informations sur le bruit ambiant établies par le module 32.

The circuit 24 comprises:

• a module 26 for acquiring the images taken by the apparatuses 16 and 18,
• a counting module 28 capable of counting the vehicles detected by the detector 20 during a given time interval δT,
• a conventional module 30 for processing the images acquired by the module 26, capable of automatically detecting the presence of an annoying object on the section 8 from the analysis of these images,
• a module 32 for establishing the ambient noise from the measurements made by the sensors 22, when no vehicle is present on the section 8,
• a module 34 for activating the processing module 30 as a function of the count established by the module 28 or ambient noise information from module 32.

The circuit 24 also comprises a memory 36 in which a predetermined operating range 38 for the acoustic sensors 22 is recorded. The range 38 defines in particular a maximum threshold of ambient noise power beyond which the detector 20 is unusable for detecting the passage of noise. 'a vehicle.

The circuit 24 also comprises a radio module 40 able to exchange information via a radio link with the road traffic beacons placed upstream and downstream along the road 4. Here, only the radio links 41 and 42 between, respectively, beacons 10 and 11, and 11 and 12 are shown.

In the particular case of the beacon 12, the radio module 40 is also able to establish a radio link 44 with an information transmission network 46, so as to be able to communicate with a platform 48 for monitoring road traffic on the road 4 .

The network 46 is, for example, a telephone network.

The platform 48 is a computer server or a set of computer servers adapted to manage the traffic on a road network including route 4.

In FIG. 1, the vehicle detectors of the beacons 10 and 11 respectively carry the references 50 and 52.

The operation of the system 2 will now be described with reference to the method of FIG. 2 in the particular case of the beacons 11 and 12 and of the section 8.

During operation of the system 2, the vehicle detectors operate continuously, during a step 60, to detect the passage of a vehicle near one of the beacons 10 to 12. Typically, the passage of a vehicle near one of the detectors is detected by measuring with the sensors 22 the power of the sound wave generated by this vehicle. For example, the power of the measured sound wave is compared to a threshold and if this threshold is exceeded then a vehicle is detected. In addition, from the direction of movement of the sound wave, each detector determines the direction of movement of the detected vehicle so as to distinguish incoming vehicles or vehicles leaving the section, if the route 4 is two-way.

During a predetermined time interval δT, the module 28 of each beacon counts, during a step 62, the number of vehicles that passed during this interval δT near this beacon from the data recorded by the detector 20.

At the end of the time interval δT, the number of incoming vehicles on the section 8 counted by the beacon 11 is transmitted, during a step 64, to the beacon 12 via the radio link 42.

où :

• S(t) est le nombre de véhicules simultanément présents sur le tronçon 8 pendant l'intervalle de temps δT courant,
• S(t-1) représente le nombre de véhicules simultanément présents sur le tronçon 8 pendant l'intervalle de temps δT précédent,
• N₁₁(t) est le nombre de véhicules entrant sur le tronçon 8 comptés par la balise 11 pendant l'intervalle de temps δT courant, et
• N₁₂(t) est le nombre de véhicules sortant du tronçon 8 comptés par la balise 12 pendant l'intervalle de temps δT courant.

The module 28 of the beacon 12 then counts, during a step 66, the vehicles simultaneously present on the section 8. For this purpose, the beacon 12 uses, for example, the following relation: $$\mathrm{S}\left(\mathrm{t}\right)\mathrm{=}\mathrm{S}\left(\mathrm{t}\mathrm{-}\mathrm{1}\right)\mathrm{+}{\mathrm{NOT}}_{\mathrm{11}}\left(\mathrm{t}\right)\mathrm{-}{\mathrm{NOT}}_{\mathrm{12}}\left(\mathrm{t}\right)$$

or :

• S (t) is the number of vehicles simultaneously present on section 8 during the current time interval δT,
• S (t-1) represents the number of vehicles simultaneously present on the section 8 during the preceding time interval δT,
• N ₁₁ (t) is the number of vehicles entering section 8 counted by beacon 11 during the current δT time interval, and
• N ₁₂ (t) is the number of vehicles leaving section 8 counted by beacon 12 during the current δT time interval.

Then, the beacon 12 calculates, during a step 68, an incident threshold Si based on an average S _m of the number of vehicles simultaneously present on the section 8 calculated over several previous δT time intervals. For example, the threshold S _i is equal to at least twice the average S _m and at least equal to 1.

$$\mathit{Si}{\mathit{S}}_{\mathit{i}}\mathit{>}\mathit{S}\left(\mathit{t}\right)\mathit{\ge }{\mathit{S}}_{\mathit{m}}\mathit{alors}{\mathit{P}}_{\mathit{i}}\mathit{=}\left(\mathit{S}\left(\mathit{t}\right)\mathit{-}{\mathit{S}}_{\mathit{m}}\right)\mathit{/}\left({\mathit{S}}_{\mathit{i}}\mathit{-}{\mathit{S}}_{\mathit{m}}\right)$$

$$\mathit{Si\; S}\left(\mathit{t}\right)\ge S_i\mathit{alors}{\mathit{P}}_{\mathit{i}}\mathit{=}\mathit{1}$$

Then, during a step 70, the beacon 12 calculates a result representative of the increase in the number of vehicles on the section 8. This result is here a probability P _i that an annoying object is actually present on the section 8. The probability P _i is established according to the data recorded by the detectors 20 and 52 and more precisely according to the number of vehicles enumerated during the step 66. For example, this probability P _i is calculated using the relation next : $$\mathit{If\; S}\left(\mathit{t}\right)\mathit{<}{\mathit{S}}_{\mathit{m}}\mathit{so}{\mathit{P}}_{\mathit{i}}\mathit{=}\mathit{0}$$

$$\mathit{Yes}{\mathit{S}}_{\mathit{i}}\mathit{>}\mathit{S}\left(\mathit{t}\right)\mathit{\ge }{\mathit{S}}_{\mathit{m}}\mathit{so}{\mathit{P}}_{\mathit{i}}\mathit{=}\left(\mathit{S}\left(\mathit{t}\right)\mathit{-}{\mathit{S}}_{\mathit{m}}\right)\mathit{/}\left({\mathit{S}}_{\mathit{i}}\mathit{-}{\mathit{S}}_{\mathit{m}}\right)$$

$$\mathit{If\; S}\left(\mathit{t}\right)\ge S_i\mathit{so}{\mathit{P}}_{\mathit{i}}\mathit{=}\mathit{1}$$

In a step 72, this probability P _i is compared with a threshold S _a of activation of the image processing. For example, S _a is equal to 0, 5. If the probability P _i is lower than the threshold S _a , then the process returns to step 60 and the image processing module 30 is not activated or deactivated. .

In the opposite case, the module 34 controls the activation, during a step 74, of the apparatus 16 and the modules 26 and 30 of the beacon 12.

In parallel, during a step 76, the activation unit 34 also controls the activation of the modules 26 and 30 of the beacon 11 as well as the setting device view of the marker 11 facing the portion 8 'of section 8.

In a step 78, the activated cameras take images at regular intervals of the section 8. These images are acquired by the image acquisition modules 26 and transmitted to the respective processing modules of the beacons 11 and 12. During a step 80, the processing modules 30 of the beacons 11 and 12 determine from the analysis of the acquired images, a probability P _v that an annoying object is present on the section 8.

Once this probability P _{v is} determined, the beacon 11 transmits, during a step 82, the probability P _v that it has determined at the beacon 12 via the radio link 42.

où :

• P_v11 est la probabilité P_v de présence d'un objet gênant déterminée par la balise 11,
• P_v12 est la probabilité P_v de présence d'un objet gênant déterminée par la balise 12, et
• P_i est la probabilité d'incident établie par la balise 12 à partir des données relevées par les détecteurs 20 et 52.

In a step 84, the beacon 12 combines the probabilities P _v determined by the beacons 11 and 12 and the probability P _i established by the beacon 12, so as to establish an estimator E _i of incidents. For example, the estimator E _i is calculated using the following relation: $${\mathrm{E}}_{\mathrm{i}}\mathrm{=}{\mathrm{P}}_{\mathrm{v}\mathrm{11}}\mathrm{+}{\mathrm{P}}_{\mathrm{i}}$$

or :

• P _v11 is the probability P _v of presence of an annoying object determined by the beacon 11,
• P _v12 is the probability P _v of presence of an annoying object determined by the tag 12, and
• P _i is the probability of incident established by the beacon 12 from the data recorded by the detectors 20 and 52.

The estimator E _i is compared, during a step 86, with a predetermined threshold S _b of alarm. If the estimator E _i is below the threshold S _b , then the process returns to step 62.

In the opposite case, that is to say if there is a high probability that an annoying object is present on the section 8, then the beacon 12 transmits, during a step 90, an alarm to the platform 48 via link 44 and network 46 and then returns to step 62.

The platform 48 receives this alarm and acts accordingly during a step 92. For example, the platform 48 automatically controls the display on a luminous sign of a message indicating that an annoying object is on the section 8 .

In parallel with steps 60 to 90, the sensors 22 of the beacon 12 are also used, during a step 100, to measure the power of the ambient noise when no motor vehicle is present near the beacon. The power thus measured is then compared, in a step 102, with the operating range 38. If the measured ambient noise power is within the operating range, then the process returns to step 100.

Otherwise, system 2 switches to a degraded operating mode. For example, if the ambient noise near the beacon 12 is too high, the unit 34 automatically and automatically activates the activation, during a step 104, of the apparatus 16, the module 26 and the module 30. the beacon 12 and the apparatus and corresponding modules in the beacon 11, so as to be able to quickly detect the presence of an annoying object on the section 8, by image processing. Thus, in this degraded mode of operation, the image processing is used to compensate for the fact that the detector 20 is unusable or inoperative.

What has been described above in the particular case of section 8 and beacons 11 and 12 applies to all pairs of beacons placed at the entrance and exit of a road section.

Thus, in the system 2, since the image processing is activated only when the probability that there is an annoying object on a section is high, it limits the energy consumption of each of the tags, which increases their autonomy .

FIG. 3 represents a system 110 for detecting an annoying object on the road section 4. In FIG. 3, the elements already described with reference to FIG. 1 bear the same numerical references. Here, only three tags 112, 113 and 114 placed respectively at the location of the tags 10, 11 and 12 of Figure 1 are shown. The tags 112 to 114 are identical and only the tag 114 will be described in detail. The tag 114 is identical to the tag 12 except that it has no image processing module 30.

In the system 110, image processing is performed in the platform 48. For this purpose, the platform 48 includes an image processing module 118 common to all the traffic beacons of the system 110. The module 118 as the module 30 is able to establish a probability P _v that an annoying object is present on a section from images acquired by the cameras for shooting beacons 112 to 114. The platform 48 is here able to trigger an alarm if the probability P _v combined or not with the probability P _i exceeds a predetermined threshold and to act accordingly.

The operation of the system 110 is deduced from the operation of the system 2. The main difference lies in the fact that the images acquired by the module 26 are only transmitted to the module 118 when the probability P _i established by a beacon is greater than the threshold S _a . Thus, in this embodiment, the activation unit 34 makes it possible to limit the bandwidth required to transmit images of a tag to the platform 48. The presence of the activation module 34 also makes it possible to limit the computation power necessary to execute the image processing, since it is very unlikely that the module 118 will have to process in parallel the images acquired by the assembly. 110 system traffic beacons.

Many other embodiments of system 2 and 110 are possible. For example, acoustic sensors can be replaced by microwave radars, ultrasound, magnetic sensors or other sensors able to detect the passage of a vehicle at a given point of a road.

Each beacon may have a single camera or more than two cameras.

Here, the calculation of the probability P _i is carried out locally by the tags. Alternatively, this calculation can be deported in the platform 48, which requires that the numbers S (t) established by each of the beacons are transmitted in real time to the platform 48.

In a variant, the acquisition of the images is permanently activated and only the processing module is activated when necessary by the module 34.

Preferably, the counting module establishes from the data recorded by the detector 20 an average number of vehicles counted accompanied by a standard deviation for this average.

In a variant, the activation threshold S _a is a function of the mean S _m .

Claims (10)

1. Method of detecting the presence of a disruptive object on a stretch of road, this method comprising:

   a) a step of processing images to automatically detect the presence of the disruptive object on the basis of images of the said stretch,
   characterized in that the method comprises:

   b) a step (66) of enumerating the vehicles simultaneously present on the stretch of road on the basis of data gleaned by a first and a second vehicle detector placed respectively at an entrance and at an exit of this stretch, and

   c) a step (74) of instructing the activation of the image processing step, triggered automatically as a function of the enumeration of step b).
2. Method according to Claim 1, characterized in that it comprises:

   - a step of acquiring the images intended to be processed during the image processing step, and

   - a step (76) of instructing the activation of the image acquisition step, triggered automatically as a function of the enumeration of step b).
3. Method according to either of the preceding claims, characterized in that the method comprises:

   - a step (70) of calculating a result representative of an increase in the number of vehicles enumerated, this result being dependent on the number of vehicles enumerated during step b), and

   - a step (72) of comparing the result with a predetermined activation threshold so as to automatically trigger at least one of the instruction steps if this threshold is crossed.
4. Method according to Claim 3, characterized in that it comprises a step (68) of calculating the result and/or the predetermined activation threshold as a function of the mean number of motor vehicles simultaneously present on this stretch observed over previous time intervals.
5. Method according to any one of the preceding claims, characterized in that it comprises a step (104) of instructing the activation of the image processing step if one of the vehicle detectors placed at the entrance or at the exit of the stretch becomes inoperative.
6. Method according to Claim 5 for vehicle detectors each comprising at least one acoustic sensor for detecting the passage of a motor vehicle on the basis of the sound wave generated by this motor vehicle, this or each sensor being intended to work in a predetermined span of ambient noise powers, characterized in that the method comprises:

   - a step (100) of measuring the ambient noise with the aid of these acoustic sensors, and

   - a step (102) of comparing the ambient noise measured with the predetermined span of ambient noise powers to determine whether the detector is inoperative.
7. Method according to any one of the preceding claims, characterized in that the method comprises a step (90) of generating an alarm signal indicating the presence of a disruptive object on the stretch of road, and in that this generating step is triggered automatically as a function at one and the same time of results obtained during the image processing step and during the enumerating step.
8. System for detecting the presence of a disruptive object on a stretch of road, in which the system comprises:

   - an image processing module (30) able to automatically detect the presence of the disruptive object on the basis of images of the said stretch,

   characterized in that the system comprises:

   - at least one first and one second motor vehicle detector (20, 50, 52) placed respectively at an entrance and at an exit of this stretch,

   - a module (28) for enumerating motor vehicles simultaneously present on this stretch on the basis of data gleaned by the first and second detectors, and

   - a module (34) for activating the processing module suitable for automatically triggering the activation of the processing module as a function of the enumeration established by the enumeration module.
9. System according to Claim 8, characterized in that the system comprises a module (26) for acquiring the images intended to be processed by the image processing module, and in that the activation module (34) is also able to automatically trigger the activation of the acquisition module as a function of the enumeration established by the enumeration module.
10. System according to Claim 8 or 9, characterized in that:

    - the motor vehicle detectors each comprise at least one acoustic sensor (22) for detecting the passage of a motor vehicle on the basis of the sound wave generated by this vehicle, this or each sensor being intended to work in a predetermined span of ambient noise powers,

    - the system comprises a module (32) for establishing the power of the ambient noise, and

    - the activation module is able to automatically activate the image processing module if the ambient noise power established is incompatible with the predetermined span of ambient noise powers.

Images