FR2883655A1 - Reliable anticollision alert system for ships, comprises optical receiver(s) covering the horizon, image processor and anticollision analyzer for calculating progression of visible object position information - Google Patents

Abstract

An anticollision alert system for ships comprises optical receiver(s) (10) at least partially covering the horizon from the ship; an image processor (IMP) for extracting information regarding visible surface object(s) in real time from the received images; and an anticollision analyzer (ACOL) for periodically calculating the progression of visible object position information and evaluating the risk of collision of the ship with the object(s) based on the information. An independent claim is included for a corresponding collision alert procedure, in which an object is classified as dangerous if there is a risk of collision of the ship with the object.

Description

ANTICOLLISION ALERT SYSTEM FOR A MARINE VEHICLE AND

  TECHNICAL FIELD The invention relates to an anti-collision warning system for a marine vehicle and to a method of anti-collision analysis by image processing from an omnidirectional optical sensor installed on board a marine vehicle.

  The invention applies in particular, but not exclusively, to ships as well as to naval drones. State of the Prior Art In the rest of the description, the marine vehicle considered as an example is a ship.

  The risk of collision, which is poorly controlled at sea, comes largely from a lack of optical monitoring means. In the area of maritime transport, an average of 600 collisions occur per year. The consequences of these collisions are often serious for the environment when they involve tankers or chemical tankers.

  In the field of maritime passenger transport, ships (ferries, ro-ro ...) have a high vulnerability.

  In the field of fishing, there are approximately 25,000 collisions per year. The number of victims is high because of the small size of the ships.

  In the field of pleasure, collisions are also quite common.

  The collisions at sea result mainly from a lack of watch. 70% to 90% of accidents are caused by human failure, either due to lack of vigilance or negligence on the part of watch staff, or by routine and false risk assessment. Lack of vigilance is often noted when the ship is in a "priority" situation, under conditions of good visibility. Accidents can also result from a lack of qualification, a lack of knowledge of the rules, or even total incompetence of the shift staff. It is also common that the number of shift staff is insufficient due to a reduction of crews (maneuvers, maintenance, commercial activities, ...), resulting in an overload of work and therefore increased fatigue of the staff shift. These accidents can also result from improper use of the radar system.

  At present, most commercial vessels have a radar system as collision avoidance equipment. This system provides accurate and reliable information, but requires rigorous application of a tuning procedure to be effectively used as an anti-collision system. This procedure is often poorly applied.

  There are also ARPA (Automatic Radar Plotting Aid) type systems that analyze the signals provided by a radar. These systems have a high false alarm rate. As a result, the alarm device associated with the system is frequently shut down.

  An automatic identification system (AIS) has also been developed which will equip all high-risk vessels (passenger transport, transport of dangerous goods) as of 2010. This system is very useful. effective when combined with a precise positioning system such as GPS ("Global Positioning System"). However, the AIS system does not take into account the circulation of the many more ships that are not equipped with this system.

  Today, only the radar is really taken into account to treat the risk of collision. This equipment has a lack of redundancy.

  The object of the invention is to overcome these disadvantages by proposing a fully automated anti-collision warning system and an anti-collision analysis method.

  The invention relates to an anti-collision warning system installed on board a marine vehicle comprising: at least one optical sensor at least partially covering the horizon of the marine vehicle, image processing means for extracting in real-time images provided by the optical sensor position information of at least one visible object on the sea surface, collision avoidance means, which periodically calculate the evolution of position information of the object visible, and which assess a risk of collision of the marine vehicle with the visible object according to the evolution of the position information of the visible object.

  According to one embodiment of the invention, the optical sensor comprises at least one fixed camera with respect to the marine vehicle and covers at least a substantial part of the horizon permanently.

  According to one embodiment of the invention, the optical sensor comprises at least one camera whose optical field is orientable to cover the horizon by rotation.

  According to one embodiment of the invention, the optical sensor comprises: a camera comprising an objective having an optical axis oriented substantially vertically, a rotating assembly driven by a motor and carrying a mirror disposed in the optical field of the objective and oriented substantially at 45 relative to the optical axis of the lens, and a device for measuring the angular position of the rotating assembly.

  According to one embodiment of the invention, the system comprises: a housing constituted by a base serving as a support, and a stabilized tank mounted on a stabilization device, and serving as a support for the optical sensor.

  According to one embodiment of the invention, the stabilizing device comprises a cardan suspension and a gyroscopic steering wheel.

  According to one embodiment of the invention, the sensor comprises a single-line camera driven in rotation about a vertical axis.

  According to one embodiment of the invention, the system comprises: a housing, a rotating assembly held in the housing and driven by a vertical axis motor, a protective cover of the rotating assembly provided with an opening, - A digital camera attached to the rotating assembly, and comprising a single-line sensor, - a lens whose optical axis is able to scan the horizon by rotating the rotating assembly.

  According to one embodiment of the invention, the sensor comprises a protective glass ventilated with the aid of an exhaust fan to protect the objective.

  According to one embodiment of the invention, electronic image processing cards are integral with the rotating assembly.

  According to one embodiment of the invention, the system comprises a device for measuring the angular position of the rotating assembly.

  According to one embodiment of the invention, the digital output data of the optical sensor are transmitted via rotating contacts which also ensure the contact of the power supply of the rotating assembly members.

  According to one embodiment of the invention, the system comprises interface means with a compass.

  According to one embodiment of the invention, the system comprises means for interfacing with other sensors and analysis means performing a coherence analysis with information provided by the other sensors.

  According to one embodiment of the invention, the system comprises signaling means for signaling that a dangerous object is detected.

  According to one embodiment of the invention, the position information of each visible object comprises an azimuth and a dimension of the object.

  The invention also relates to an anti-collision warning method in a marine vehicle anti-collision warning system comprising at least one optical sensor at least partially covering the horizon of the marine vehicle. According to the invention, the method comprises the steps of: extracting in real time images provided by the optical sensor from the position information of at least one visible object on the surface of the sea, periodically calculating the evolution of the information position of the visible object, and evaluate a risk of collision of the marine vehicle with the visible object according to the evolution of the position information of the visible object, a visible object being considered dangerous if there is a risk of collision of the marine vehicle with the visible object.

  According to one embodiment of the invention, the method comprises steps of displaying information relating to each dangerous object, and of transmitting an alarm signal as soon as a new dangerous object is detected.

  According to one embodiment of the invention, the method comprises the steps of: a) framing the useful part of the image provided by the optical sensor, b) measuring the average light intensity on at least a part of the image, c) comparing each pixel with the average light intensity, and assigning a binary value to each pixel according to the result of the comparison, d) searching for pixels of a given value forming groups of adjacent pixels, each group of pixels adjacent pixels constituting a visible object, and measuring a position and dimensions of visible objects; e) determining dangerous objects as a function of variations in the position and / or dimensions of the objects visible in successive images.

  According to one embodiment of the invention, the method comprises steps of evaluating the persistence of each group of pixels in successive images, and of determining the groups constituting visible objects according to their persistence.

  According to one embodiment of the invention, the framing of the useful part comprises an automatic detection of the horizon line.

  According to one embodiment of the invention, steps d) and e) are performed for each image.

Brief description of the figures

  These and other objects, features and advantages of the present invention will be set forth in greater detail in the following description of embodiments of the invention, given in a nonlimiting manner in relation to the attached figures among which: FIG. 1 illustrates a system of the invention installed on a ship; FIG. 2 illustrates in block form the functions of the system of the invention; FIGS. 3A and 3B illustrate a first embodiment of the system of the invention; FIGS. 4A and 4B illustrate a second embodiment of the system of the invention; FIGS. 5A and 5B illustrate a third embodiment of the system of the invention; FIGS. 6A and 6B illustrate a principle of collision risk analysis applied by the method according to the invention.

  DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS As illustrated in FIG. 1, the system of the invention comprises: a set of optical sensors, 10, disposed here in the superstructures of a ship 12, and a computer 13 implanted in the bridge of the ship.

  As illustrated in FIG. 2, the system of the invention comprises, in functional terms, two segments, namely an optical segment SO and a tactical segment ST.

  The optical segment comprises: - one or more optical sensors 10 at least partially covering the horizon of the marine vehicle, - IFC interface means of the sensors with the computer, which ensure the acquisition of the images provided by the sensor and the control sensors, and image processing means IMP which extract information from the position of objects.

  The tactical segment ST comprises: ACOL anti-collision analysis means that perform a periodic calculation of the evolution of the position information of each visible object, and evaluate a risk of collision of the marine vehicle with each visible object, and a human interface / HMI machine, which can be integrated in the control station of the marine vehicle, or remotely deported, and which includes signaling means for signaling that a dangerous object is detected.

  The set of optical sensors 10 advantageously covers the entire horizon, 360, or a significant portion thereof. Depending on the configuration of the ship, and in particular the dimensions of its superstructures, it may be preferable to have a single sensor or several sensors. Ideally, on a large cargo ship, three sensors, one at the front of the ship and one on each side of the superstructures, cover the whole horizon, including nearby areas that are hidden by the superstructures for an observer. located on the ship's navigating bridge.

  In a first embodiment of the invention, each sensor comprises a camera whose optical field is orientable to cover the horizon by rotation. In a second embodiment of the invention, each sensor comprises a steerable single-line camera mounted on a fixed support. In a third embodiment of the invention, each sensor comprises one or more cameras fixed with respect to the marine vehicle and covering at least a substantial part of the horizon permanently.

  Given the movements of the carrier ship, it is a question of finding the best compromise between the resolution of the camera and the requirement of stability. Two solutions are conceivable: - either mount the camera on a stabilized platform to center the image on the useful area around the horizon, - or use information on the attitude of the ship (angles of roll and pitch) to process only the useful part of the image (points located at a certain distance in pixels above or below the horizon line). This information can come from either an external sensor (center of attitude) or an image processing (detection of the horizon line) when the horizon is visible.

  In the first embodiment illustrated in FIGS. 3A and 3B, the optical sensor 10 comprises a camera whose optical field is orientable to cover the horizon by rotation. The camera is mounted on a platform stabilized by a gyroscope. The sensor 10 comprises: - a housing 21 consisting of a base 22 serving as a support and a transparent protective dome 23, - a stabilized tank 24 mounted on a stabilization device consisting of a cardan suspension 25 and a gyro wheel 26 driven by an electric motor 27, a camera 28 attached to the stabilized tank and carrying a goal 29 oriented at the zenith, a rotating assembly 30 mounted in the stabilized vessel, driven by a rotation motor 31 with a vertical axis placed in the bottom of the stabilized tank, and carrying in its upper part a mirror 32 and a sun visor 33, the arrangement of the mirror, oriented at 45, allowing the optical axis of the mirror 32 to scan a horizontal plane by the rotation of the rotating assembly 30, the rays being reflected by the mirror 32 so as to reach the objective of the camera, the motor 31 driving the rotating assembly via a gear reducer, and a measuring device e of the angular position of the rotating assembly, or resolver, 34, possibly integrated in the engine 30 if the technology of the latter allows it (stepper motor).

  Depending on the materials used (alloys and / or composites), the entire housing 21 may have a total mass of less than 2 kg and a volume of less than 2 liters.

  Prototype tests with a rotational speed of the gyroscopic wheel 26 of 9900 rpm ensured a stability characterized by an angular velocity of the tank less than 60 mrad / s (milli-radians per second) in 99% of case. The stabilization device gives, with an integration duration of 1 / 1000th of a second, a "shake" less than 1 / 8th of a pixel, for a roll / pitch tolerance of the ship of +/- 25. Such performance can be further improved with an industrial invoice product (machining, adjustment, and bearing tolerance) and a gyro rotation speed of 15,000 rpm.

  The camera 28 has for example the following characteristics: a 1/3 "CCD cell for real-time operation in a standard format (JPEG), - a high resolution XGA video format (1024 x 768 pixels) offering the best compromise between angular aperture of the lens 29 (36 diagonally) and the desired angular resolution, and a focal lens 9mm 9mm.

  In a second embodiment illustrated in FIGS. 4A and 4B, the optical sensor 10 comprises a steerable single-line camera mounted on a fixed support.

  More specifically, the optical sensor comprises: a cylindrical housing 41 fixed to the mast of the ship, - a rotating assembly 42 held in the housing 41 by ball bearings 43 and driven by a rotation motor 44 with a vertical axis placed in the bottom of the housing, the upper part of the rotating assembly comprising a cylindrical protective cover 45 provided with a rectangular window 49 which allows the optical rays to pass through the hood to reach the objective, a digital camera 46 fixed to the rotating assembly 42 and whose mono line sensor 47 is offset above the body of the camera; the sensor typically comprises from 512 to 8192 sensitive photo elementary cells arranged in a vertical line, an objective 48, with an angular aperture typically of 30 to 55, arranged in such a way that its optical axis sweeps the horizon with the rotation of the rotating assembly 42, and a flap 50 arranged to close the rectangular opening 49 when the system is stopped to protect the inside of the sensor spray and rain; the movement of the flap is carried out using a magnetic control 51.

  When the system is in operation, the protection of the objective 48 against rain splashing through the rectangular opening 49 is provided by a protective glass 52, ventilated with the aid of an exhaust fan 53 blowing into a dewatering nozzle 54 of the air heated by the different heat sources of the rotating assembly and the housing (rotation motor, electronics and camera).

  The electronic cards 56 integrating the computer and the associated circuits of the collision warning system are integral with the rotating assembly 42.

  Rotating assembly 42 also carries a deposit encoder 57 which measures the angular position of the rotating assembly.

  The digital output data of the anti-collision warning system are transmitted via a rotary switch 58 which also transfers data to the computer from the ship's compass, as well as the contact of the power supply of the control devices. rotating assembly 42.

  In the second embodiment, the panoramic image of the sea 360 around the sensor 10 is achieved by scanning the optical brush determined by the rotation of the single-line sensor. The vertical angular aperture of this brush is chosen so as to cover the useful range of orientation in site even with roll / pitch movements.

  The angular resolution of the image depends on the horizontal and vertical angular resolutions.

  The horizontal angular resolution of the image is a function of the rotation speed and the line reading frequency. Existing linear cameras have line read rates typically ranging from 6 to 87 kHz. In practice, the limit of the reading frequency depends on the illumination conditions and the light sensitivity. For a rotation speed typically ranging from 0.12 to 2 revolutions per second, the horizontal resolution is between 0.01 and 2 mrad / pixel (milliradians per pixel).

  The vertical angular resolution is determined by the number of pixels of the sensor and the angular aperture of the lens. For a number of pixels ranging from 512 to 8192 and a vertical angular aperture of between 4 and 45, the vertical resolution is between 0.01 and 1.5 mrad / pixel.

  In a third embodiment illustrated in FIGS. 5A and 5B, the optical sensor 10 comprises a set of fixed cameras. More specifically, the optical sensor comprises: a housing 61 fixed to the superstructure of the ship and having a glazed opening 62, one or more digital cameras 63 (typically four) fixed to the housing 61, an objective 64 for each camera 63 , the set of objectives being arranged in such a way that the sectors they scan cover a substantial portion of the horizon (typically 187), and - a connection and multiplexing device 65 making it possible to simultaneously process the images coming from all cameras. This device can integrate an electronic card performing part of the image processing.

  In this third embodiment, the resolution of the images is related to the platform movements of the carrier ship. Typically, on a large ship with a roll limit of +/- 10, the use of cameras with a definition of 1280 x 1024 pixels and lenses of 8 to 12 mm allows to obtain the necessary resolution, in all conditions illuminance, for a detection range of up to 10,000 meters.

  The interface with the optical sensors (card and electronic circuit, and / or software, and / or wire connections, and / or wireless connections) between the optical sensors and the associated computer, is adapted to the selected architecture. This interface can therefore include: an interface integrated into the computer, an interface integrated with the sensors, an interface contained in an intermediate box. The interface with the optical sensors provides two functions.

  the acquisition of the images and, in the case where the sensors comprise a rotating assembly, the acquisition of the orientation measurement, and the control of the sensors, that is to say the control of the cameras and, in the case of the case where the sensors comprise a rotating assembly, the control of the orientation motors.

  In the first embodiment described above and illustrated in FIGS. 3A and 3B, where the optical sensor 10 comprises a camera whose optical field is orientable to cover the horizon by rotation, the IFC interface is an electronic circuit integrated in the computer.

  In the second embodiment described above and illustrated in FIGS. 4A and 4B, where the optical sensor 10 uses a rotating single-line camera, the IFC interface is an integrated circuit in the electronic cards 56 mounted on the rotating assembly. 42.

  In the third embodiment described above and illustrated in FIGS. 5A and 5B, where the optical sensor 10 comprises a set of fixed cameras, the IFC interface is an integrated circuit in the electronic cards of the connection and multiplexing device. .

  The image processing method implemented by the image processing means IMP makes it possible to extract in real time, from each image supplied by the optical sensor (10), information of position of visible objects on the surface of the image. the sea, eliminating waves and scum.

  This image processing comprises the steps of: framing the useful part of the image provided by the optical sensor (10), - measuring the average light intensity on at least a portion of the image, comparing each pixel with the average luminous intensity, and assigning a binary value to each pixel according to whether the difference is above or below a light intensity threshold, - searching for pixels of a given value forming groups of adjacent pixels, each group of adjacent pixels constituting a visible object, and - measuring a position and dimensions of visible objects.

  The framing of the useful part may include an automatic detection of the horizon line. It may also include a straightening of the image so that the horizon line remains parallel to the lower edge of the image.

  The useful part of the image can itself be divided into several parts, for example according to the distance between each pixel and the horizon line, which roughly determines the distance of the corresponding point relative to the carrier ship.

  Depending on the type of camera used, the light intensity can be measured only on the luminance, and / or on one or more of the three color components (red, blue, yellow).

  The comparison of each pixel with the average light intensity can be performed simultaneously with several parameters (threshold, color).

  The detection of visible objects may comprise a step of evaluating the persistence of each group of pixels in successive images. For this purpose, the position information of all groups of pixels in each image is recorded in a database.

  The comparison of the successive databases corresponding to the succession of images of the same sector makes it possible to determine groups constituting visible objects according to their persistence, that is to say according to the percentage of appearance of these groups in a number of images. This process eliminates "non-persistent" objects: waves, reflections, foam, and to keep only the objects corresponding to ships or objects floating on the surface.

  In the first embodiment where the optical sensor 10 comprises a camera whose optical field is orientable to cover the horizon by rotation, the image processing is performed on each image, typically every 1 / 15th of a second. This treatment comprises the following steps: - automatic detection of the horizon line, straightening of the image so that the horizon line remains parallel to the lower edge of the image, measurement of the average light intensity on at least one part of the image centered on the horizon, comparing each pixel with a light intensity threshold according to the average light intensity, and assigning a binary value to each pixel according to whether the difference is above or below the light intensity threshold, search for pixels of a given value forming groups of adjacent pixels, record position information of all groups of pixels in a database of detected groups BDG, comparison of group databases detected successively to each image provided by the optical sensor, and determination of groups constituting visible objects according to their persistence in more their databases of detected groups.

  In the second embodiment, the image processing by the image processing means IMP is performed first on each elementary image corresponding to a vertical line, typically every 1 / 3000th of a second. This processing includes the following steps: automatic detection of the horizon line, image registration so that the horizon line remains at a constant distance, in number of pixels, from the lower edge of the image, measurement of the average light intensity over at least a portion of the image centered on the horizon, comparing each pixel with the average light intensity, and assigning a binary value to each pixel according to whether the difference is above or below a threshold of luminous intensity, and reconstitution of an image formed by the succession of vertical lines.

  On each image thus reconstructed, typically every second, the image processing comprises the following steps: searching for pixels of a given value forming groups of adjacent pixels, recording position information of all groups of pixels in a base of group data detected BDG, comparison of the databases of detected groups constituted to each successive image reconstructed (comparison typically carried out on 3 to 10 images), and determination of the groups constituting visible objects as a function of their persistence in several databases of group data detected.

  In the third embodiment, the image processing is performed by the image processing means IMP on each image, typically every second. This processing comprises the following steps: automatic detection of the horizon line, straightening of the image so that the horizon line remains parallel to the lower edge of the image, measurement of the average light intensity over at least 25 minutes. part of the image centered on the horizon, comparison of each pixel with the average luminous intensity, and assignment of a binary value to each pixel according to whether the difference is above or below a luminous intensity threshold , searching for pixels of a given value forming groups of adjacent pixels, recording position information of all groups of pixels in a database of detected groups BDG, comparing databases of detected groups made up of each successive image (typically 3 to 10 images), and - determining the groups constituting visible objects according to their persistence in several databases of groups of groups ected.

  The features of the visible objects are stored in a BDOV visible object database.

  In general, the optical sensor associated with the image processing has the following characteristics: - it provides detection information, namely the presence of an object, its azimuth, and in the case of a system with several sensors, the distance from the object. it allows detection in the visible, under the same conditions of visibility as the human eye, day and night; with appropriate sensors (eg infrared sensitive), it can operate under more severe conditions, - at night, the system may include an infrared illumination device, knowing that the distance of detection of obstacles not illuminated can go up to 'at 2000 meters, - it has a detection distance which, depending on the configuration of the system and the kinematics of the carrier marine vehicle, can reach up to 10 000 meters, while detecting the nearest objects, - it makes it possible to identify objects not detected by the radar, such as small boats and floating wrecks, - the image processing can be performed in both color and black and white, it can operate with the following platform movements: roll (+/- 35), period of 5 to 15 seconds, pitch (+/- 15), angular velocity less than 10 / sec.

  - It is conditioned in a manner adapted to the marine environment in terms of sealing, mechanical and electrical protection, and resistance to sun, spray and vibration.

  The anti-collision analysis method implemented by the ACOL anti-collision analysis means comprises the following steps.

    determining position information of each visible object in the BDOV database, as a function of their position in the images provided by the optical sensor, periodic calculation of the evolution of the position information of each visible object, risk assessment collision of the marine vehicle with each visible object according to the evolution of the position information of the visible object.

  The position information of a visible object includes its azimuth, its site (angular deviation from the horizon line), and possibly its distance when this can be calculated (in the case of a system comprising several sensors). This information is supplemented by an indication of the apparent dimensions of the object in the image: height and / or width in number of pixels.

  The principle of analysis of the risk of collision applied by the invention is based on the evolution of the azimuth of a visible object, and on the evolution of its dimensions and possibly its distance when it is calculated.

  More precisely, the anti-collision analysis method comprises the following steps: periodic calculation of the evolution of the position and dimension information of each visible object by analysis of the database of the visible objects every 30 seconds, on a background of up to 20 minutes, extraction of objects whose azimuth variation is less than 1.5 / min (degrees per minute), extraction of objects whose dimension grows and / or whose distance (when calculated) decreases, and constitution of a database BDOD hazardous objects gathering the characteristics of visible objects extracted in the previous two steps.

  The analysis of the evolution of the azimuth of a visible object can be used as a basis for estimating a risk of collision, as illustrated in FIGS. 6A and 6B. These figures represent the trajectory 1 of the ship and the estimated trajectory 2 of a visible object.

  The points formed on the trajectories 1, 2 show the respective positions of the ship and the object at times t1 to t9.

  Figure 6A illustrates the case of a visible object deemed dangerous. It can be seen in this figure that the azimuth of the visible object is constant with respect to the azimuth of the ship.

  Figure 6B illustrates the case of a visible object deemed non-dangerous. In this case, the azimuth of the visible object with respect to the ship is not constant.

  The variation in the size of each visible object is also an information to consider in estimating a risk of collision. Indeed, if the ship and the visible object follow parallel paths and at the same speed, the azimuth of the object is constant with respect to that of the ship although the object does not pose a threat to the ship. In addition, if the apparent size of a visible object increases, it means that it is getting closer to the ship.

  If the ship is equipped with an ARPA PLC coupled to a radar system, it can also be planned to acquire azimuths and radar echo distances taken into account by the "ARPA" PLC and "AIS" tracks. The positions of radar echoes and AIS tracks can then be compared with the BDOD hazardous object database, to relate each external track to a dangerous object and vice versa. The database of dangerous objects can thus gather all the dangerous objects detected by the system according to the invention and by the other equipment of the ship, each dangerous object being associated in the database with information indicating by which means (s) each object has been detected.

  The BDOV Visible Object and BDOD Dangerous Object Databases are encoded in ANSI in a frame format compatible with software commonly used in on-board data processing systems, typically NMEA (National Maritime Electronics) format. Association) or the XML (eXtended Markup Language) format. Attributes associated with each object in these databases include: time of first detection, azimuth, angular width, distance to ship when calculated, accuracy of distance to ship, persistence , and the contrast.

  The HMI man / machine interface provides crew information and system control functions. The information function of the crew includes: the emission of an alert signal, which may be audible and / or visual and / or adapted to a specific need (for example synthetic voice, or vibration on a mounted case ), as soon as the database of dangerous objects contains a new element, the display of the position information of a dangerous object, and in particular of its azimuth, and an indication of the urgency of the situation (by for example, a distance or delay before collision), and the recording of a database history of visible objects and dangerous objects.

  It can also be provided that the human-machine interface HMI offers the possibility of displaying the real-time image of an object in view, at the request of the operator.

  When the ship has several detection means, the HMI man / machine interface may also display an indication of the detection means (s) of each object. When the ship is equipped with an ARPA device for collision avoidance assistance associated with a radar, the man / machine interface advantageously makes it possible to graphically superpose information from the databases of visible objects and objects. dangerous with the ARPA display. If the vessel is equipped with a digital mapping device, for example according to the Electronic Chart Display and Information System (ECDIS) standard, the man / machine interface can graphically overlay the information from the visible object databases. and dangerous objects with the ECDIS display.

  The control function of the system includes commands for routine management of system operation, and controls for maintaining the collision avoidance system.

  The current management commands include a command to turn the system on and off, and if necessary, commands to select other detection means to take into account.

  The system maintenance commands include commands for turning on, stopping, and adjusting each camera, controls for making internal computer settings (thresholds, parameters), and commands for initiating automatic procedures. calibration and internal testing of the system.

  The HMI man / machine interface is connected to the BDOV visible object database as well as to the DBOD hazardous object database. It can be integrated in the control station of the marine vehicle, or remotely.

  In one embodiment of the system according to the invention, the human / machine interface HMI is integrated into an external navigation system. The alert, display and control functions enhance the existing interfaces of the external navigation system for the operation of ARPA or ECDIS visualization means.

  ARPA display means display on a viewing screen a circular area around the ship. The ARPA view is permanently centered on the ship. Marking is facilitated by a radial line set to the north or relative to the axis of the ship, and by a series of distance rings.

  The ECDIS display means display on a display screen a rectangular geographical map respecting an international standard. This map is characterized by the geographical position of its center, its scale, the type of projection (usually Mercator) and its orientation (most often to the north or along the axis of the ship). The location on this map is facilitated by a grid representing the constant latitudes and longitudes.

  These two types of visualization means display visible objects and dangerous objects in the form of specific symbols. In this way, the risk of collision can be assessed visually.

  Upon intervention by the operator, an image of a visible object can be presented in a window displayed on the viewing screen. This image is refreshed typically every second to allow a visual assessment in real time of the evolution of a detected ship.

  When the system according to the invention receives information from several sensors, it can also perform for each object a measurement of the angular difference of the azimuths of the object provided by the different sensors, as well as a calculation of the distance of the 'object.

  Thanks to these provisions, the anti-collision system according to the invention can offer some redundancy with the existing means. Thus, it offers reliable signaling with a relatively low false alarm rate.

  Thanks to these characteristics, the system of the invention has applications in many fields, and in particular: in the field of armed vessels by professionals (commerce, fishing, recreational boating, public domain vessels), in the field of pleasure craft, by its contribution to the safety and comfort of the crew, especially in the absence of radar, in the field of coastal navigation channels (buoys, lighthouses, beacons, jetties, ...), by a monitoring and control of traffic and redundancy of collision avoidance in the most dangerous areas, in the field of naval drones, by the possibility of ensuring permanent control of collision avoidance, even when the deployment area of a drone is not under the visual control of the control station.

  In addition, on any marine vehicle the system of the invention makes it possible to automatically record all detections of visible objects, which can enrich the data automatically recorded in a "black box" or VDR (Travel Data Recorder) .

  It will be apparent to those skilled in the art that the system according to the present invention is susceptible of various other embodiments and applications.

  Thus, the invention is not limited to a system in which the optical sensor covers the entire horizon of the marine vehicle. The zone monitored by the optical sensor may in fact be limited to a front sector of the marine vehicle. Nor is it necessary for the system to have access to position information such as the azimuth of the marine vehicle or to information provided by other systems on the marine vehicle. Indeed, the position of the detected objects can be determined with respect to the marine vehicle and its heading. Furthermore, the system according to the invention may not comprise a man / machine interface, and simply send the information relating to the detection of dangerous objects to another system of the marine vehicle, comprising a man / machine interface.

Claims (22)

  1. Anti-collision warning system for marine vehicle (12), comprising: at least one optical sensor (10) at least partially covering the horizon of the marine vehicle, image processing means (IMP) for extracting in real time of images provided by the optical sensor of the position information of at least one visible object on the sea surface, anti-collision analysis means (ACOL), which periodically calculate the evolution of the position information of at least a visible object, and which evaluate a risk of collision of the marine vehicle with the visible object as a function of the evolution of the position information of the visible object.   2. System according to claim 1, wherein the optical sensor (10) comprises at least one fixed camera 20 with respect to the marine vehicle and covers at least a substantial part of the horizon permanently.   3. System according to claim 1, wherein the optical sensor (10) comprises at least one camera whose optical field is orientable to cover the horizon by rotation.   4. System according to claim 3, wherein the optical sensor (10) comprises: a camera (28) having a lens (29) having an optical axis oriented substantially vertically, a rotating assembly (30) driven by a motor (31) and carrying a mirror (32) disposed in the optical field of the objective and oriented substantially at 45 with respect to the optical axis of the objective, and a device (14) for measuring the angular position of the rotating assembly (30).   5. System according to one of claims 1 to 4, comprising: - a housing (21) constituted by a base (22) serving as a support, and - a stabilized tank (24) mounted on a stabilizing device (25, 26). ), and serving as a support for the optical sensor (10).   The system of claim 5, wherein the stabilizer comprises a gimbal (25) and a gyro (26).   The system of claim 3, wherein the sensor (10) comprises a single-line camera rotated about a vertical axis.   8. System according to claim 7, comprising: - a housing (41), - a rotating assembly (42) held in the housing (41) and driven by a motor (44) with a vertical axis, a protective cover (45) of the rotating assembly provided with an opening (49), - a digital camera (46) attached to the rotating assembly, and comprising a single-line sensor (47), - a lens (48) whose axis optical is able to scan the horizon by rotating the rotating assembly (42).   12. System according to claim 8, wherein the sensor (10) comprises a protective window (52) ventilated by means of an exhaust fan (53) to protect the objective (48).   The system of claim 8 or 9, wherein electronic image processing boards (56) are integral with the rotating assembly (42).   11. System according to one of claims 8 to   10, comprising a device (57) for measuring the angular position of the rotating assembly (42).   12. System according to one of claims 8 to   11, wherein the digital output data of the optical sensor (10) is transmitted through rotating contacts (58) which also provide contact of the power supply of the rotating assembly members (42).   13. System according to one of claims 1 to   12, comprising interface means with a compass.   14. System according to one of claims 1 to   13, comprising interface means with other sensors and analysis means performing a consistency analysis with information provided by the other sensors.   15. System according to one of claims 1 to   14, comprising signaling means (HMI) for signaling that a dangerous object is detected.   16. System according to one of claims 1 to   15, wherein the position information of each visible object comprises an azimuth and a dimension of the object.   17. Anti-collision warning method in an anti-collision warning system for a marine vehicle (12) comprising at least one optical sensor (10) at least partially covering the horizon of the marine vehicle, characterized in that it comprises steps consisting of at . extracting in real time images provided by the optical sensor (10) position information of at least one visible object on the sea surface, periodically calculating the evolution of the position information of the visible object, and evaluating a risk of collision of the marine vehicle with the visible object as a function of the evolution of the position information of the visible object, a visible object being considered as dangerous if there is a risk of collision of the marine vehicle with the visible object .   18. The method of claim 17, comprising steps of displaying information relating to each dangerous object, and transmitting an alarm signal as soon as a new dangerous object is detected.   19. The method of claim 17 or 18, comprising the steps of: a) framing the useful part of the image provided by the optical sensor (10), b) measuring the average light intensity on at least a portion of the (c) compare each pixel with the average light intensity, and assign a binary value to each pixel according to the result of the comparison, d) search for pixels of a given value forming groups of adjacent pixels, each group of adjacent pixels constituting a visible object, and measuring a position and dimensions of the visible objects, e) determining dangerous objects as a function of variations in the position and / or dimensions of the objects visible in successive images.   20. The method of claim 19, comprising steps of evaluating the persistence of each group of pixels in successive images, and determining groups constituting visible objects according to their persistence.   21. The method of claim 19 or 20, wherein the framing of the useful portion comprises automatic detection of the horizon line.   The method of claim 17, wherein steps d) and e) are performed for each image.