Classifications

• • G—PHYSICS
  • G08—SIGNALLING
  • G08G—TRAFFIC CONTROL SYSTEMS
  • G08G5/00—Traffic control systems for aircraft
  • G08G5/20—Arrangements for acquiring, generating, sharing or displaying traffic information
  • G08G5/21—Arrangements for acquiring, generating, sharing or displaying traffic information located onboard the aircraft
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T3/00—Geometric image transformations in the plane of the image
  • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
  • G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
  • G06T3/4069—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution by subpixel displacements
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T7/00—Image analysis
  • G06T7/20—Analysis of motion
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
  • G06V20/00—Scenes; Scene-specific elements
  • G06V20/50—Context or environment of the image
  • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
  • G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
• • G—PHYSICS
  • G08—SIGNALLING
  • G08G—TRAFFIC CONTROL SYSTEMS
  • G08G5/00—Traffic control systems for aircraft
  • G08G5/80—Anti-collision systems
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/10—Image acquisition modality
  • G06T2207/10016—Video; Image sequence
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/30—Subject of image; Context of image processing
  • G06T2207/30248—Vehicle exterior or interior
  • G06T2207/30252—Vehicle exterior; Vicinity of vehicle
  • G06T2207/30261—Obstacle

Definitions

• the present invention relates to the field of sense and avoid for a vehicle, and more particularly to a system and a method for calculating time to go, between a vehicle and an intruding object, and the size of the intruding object.
• a UAV typically comprises an air collision avoidance system, sometimes also referred to as a Sense & Avoid system.
• the Sense & Avoid system includes one or several sensors for sensing intruding aircrafts or objects, and collision avoidance functionality that uses the sensed data to perform a safe escape maneuver. Since the collision avoidance system is a safety enhancing system it is crucial that the data supplied to the collision avoidance functionality are of high quality in order to avoid nuisance and unsafe maneuvers.
• TTG Time To Go
• the TTG can be calculated based on data regarding the own aircraft's position and motion and data on surrounding objects, collected by the sensors of the collision avoidance system.
• this object is achieved by a method for calculating a Time To Go, TTG, value between a vehicle and an intruding object, said method comprising the steps of: - retrieving a first image of said intruding object at a first point of time, T 0 , and a second image of said intruding object at a second point of time, Ti;
• a computer program product for use in a vehicle for calculating a Time To Go, TTG, between said vehicle and an intruding object, comprising a computer readable medium, having thereon: computer readable code means which, when run in a processing means of the vehicle causes the processing means to perform;
• An advantage with the method and the system according to embodiments of the present invention is that a very accurate value of the scale factor is achieved that is used to calculate time to go.
• Another advantage with embodiments of the present invention is that the size of an intruding aerial vehicle or object in an image can be estimated with a high degree of certainty.
• Fig. 1 illustrates a top view of the front half of an Unmanned Aerial Vehicle 10 comprising electro optical sensors used in the present invention.
• Fig. 2 is a schematic illustration of a system according to embodiments of the present invention for calculating time to go and the size in an image of the intruding aerial vehicle or object
• Fig. 3 illustrates a principle used in the present invention for calculating time to go.
• Fig. 4 is a flowchart illustrating embodiments of the method according to the present invention. Detailed description
• Fig. 1 illustrates a top view of the front half of an Unmanned Aerial Vehicle (UAV) 10.
• UAV 10 comprises one or several electro-optical (EO) sensors 201 for monitoring surrounding air traffic.
• EO electro-optical
• the UAV 10 is seen to comprise seven electro-optical (EO) sensors 201 which are arranged in a semi-circular pattern on or close to the nose of the UAV 10.
• the EO sensors 201 may be any devices which are able to capture consecutive images of an intruding aerial vehicle or objects in the surrounding airspace.
• the EO sensors 201 are 9 Hz video cameras 201 capturing images having a 2048x2048 pixel resolution. That is, each camera 201 captures nine high-resolution images of the surrounding airspace every second.
• Each camera 201 has a field of view of 35 degrees in azimuth and 30 degrees in elevation.
• the fields of view of two adjacent cameras 201 are overlapping slightly in azimuth, resulting in a total field of view of 220 degrees in azimuth for the entire EO sensor arrangement.
• the EO sensor arrangement thus has a field of view of 220 degrees in azimuth and 30 degrees in elevation, which substantially corresponds to the field of view of the human eyes.
• Fig. 2 is a schematic illustration of a system 200 in an Unmanned Aerial Vehicle (UAV) (not shown) for estimating time to go and the size in an image 206, 21 1 , 213 of an intruding aerial vehicle or object 210 according to embodiments of the present invention.
• the system 200 comprises an electro-optical (EO) sensor 201 as the one described in relation to Fig. 1.
• EO electro-optical
• the electro-optical sensor 201 produces a first sequence of images 206.
• An intruding object or aerial vehicle 210 may be present in some or in all images in the first sequence of images 206 depending on among others a position of the intruding aerial vehicle or object 210 in relation to the electro-optical sensor 201.
• the first sequence of images 206 is provided to a detector 202 via a connection 212 from the electro- optical-sensor 201 to the detector 202.
• the detector 202 detects intruding aerial vehicles or objects 210 in the first sequence of images 206 taken by the electro-optical sensor 201.
• the detector 202 thereby creates a second sequence of images 211 in which the intruding aerial vehicle or object 210 is detected in images 21 1 in the second sequence of images.
• the intruding aerial vehicle or object has 210 has been detected in the second sequence of images 211 as a circle 216 in each image.
• the intruding aerial vehicle 210 is shown at different positions in the second sequence of images 211 , which means that the aerial vehicle or object 210 has moved in relation to the Unmanned Aerial Vehicle (UAV).
• UAV Unmanned Aerial Vehicle
• the second sequence of images 211 is delivered to a tracker 201 via a connection 207.
• the tracker tracks the intruding aerial vehicle or object 210 in the second sequence of images.
• the tracker thereby creates a third sequence of images 213 in which the intruding aerial vehicle or object 210 is tracked in each image.
• the tracker centralize the intruding aerial vehicle or object 210 in each image in the third sequence of images 213.
• the third sequence of images 213 is delivered to a time to go calculator 204 which calculates the time to go according to embodiments of the invention.
• the method according to the present invention for calculating time to go will be described further down in relation to figure 4.
• the detector and the tracker have been illustrated as two separate units.
• the detector and/or the tracker may according to embodiments of the invention also be a part of the time to go calculator 204.
• the second sequence of images 211 is delivered to the time to go calculator via the connection 214.
• the first sequence of images is delivered to the time to go calculator 204 via the connection 215.
• the time to go calculator 204 may also according to embodiments of the invention calculates a size in an image of the intruding aerial vehicle or object.
• the second sequence of images comprises coordinates (not shown) of the intruding aerial vehicle or object in each image in the first sequence of images.
• the third sequence of images comprises coordinates (not shown) of the intruding aerial vehicle or object in each image in the first sequence of images.
• a size A of the target image 320 at the time point Ti is bigger than a size B of the target image 310 at the time point To.
• UAV Unmanned Aerial Vehicle
• a problem with the principle according to figure 3 is that a very exact value of the scale change is required in order to achieve a god estimate of the time to go.
• An advantage with the method and the system according to embodiments of the present invention is that a very accurate value of the scale change is achieved that is used to calculate time to go according to the above principle.
• Figure 4 shows a flow chart of the method in a vehicle (not shown) for calculating time to go, according to embodiments of the present invention.
• a first step S41 the detector 202 detects the intruding aerial vehicle or object in the first sequence of images 212 produced by the EO sensor 201. Detection of the intruding aerial vehicle or object in step S41 results in that the detector creates a second sequence of images 211 in which the intruding aerial vehicle or object is detected in each image.
• An intruding aerial vehicle or object being detected in an image (not shown) in the second sequence of images could for instance mean that the images are associated with coordinates in the images where the intruding aerial vehicle or object is located. Since there are several methods known in the art for detecting objects of certain shapes in images these methods are not further described herein.
• step S42 the tracker tracks the intruding aerial vehicle or object in the second sequence 211 of images from the detector 202.
• a third sequence of images 213 is thereby created in step S42 by the tracker in which the intruding aerial vehicle or object is tracked in the images.
• tracking could for instance mean that the intruding aerial vehicle or object is centralized the images in the third sequence of images. Since tracking an object in consecutive images is well known in the art tracking is not further described herein.
• a first image and a second image of the intruding aerial vehicle of object are retrieved by the time to go calculator 204.
• the first image of the intruding aerial vehicle or object is an image of the intruding aerial vehicle or object at a first point of time
• To The second image of the intruding aerial vehicle or object is an image of the intruding aerial vehicle or object at a second point of time
• TV The first and the second images are, according to embodiments of the present invention, retrieved from the first sequence of images 206.
• the first and the second images are retrieved from the second sequence of images 211.
• the first and the second images are retrieved from the third sequence of images 213.
• first and the second images are retrieved from the first sequence of images 206
• detection and tracking of intruding aerial vehicles or objects are performed in the time to go calculator 204.
• An advantage with placing the detector and tracker in the time to go calculator is that the system 200 then requires less separate units.
• the first and the second images are retrieved from the second sequence of images 211 tracking of intruding aerial vehicles or objects is performed in the time to go calculator 204.
• An advantage with placing the tracker in the time to go calculator is that the system 200 then requires less separate units.
• step S44 in which the first and the second images are filtered by the time to go calculator 204. Filtering in step
• step S44 results in that the first and second images become independent of absolute signal energy.
• the filtering in step S44 also results in that edges of the first and second images become enhanced. Filtering of the first and the second images are necessary operations to achieve accurate values in the correlation operations, which will be described further down.
• step S44 an X fractional pixel position, X FRAC , is set to zero and a Y fractional pixel position, YFRAC, is set to zero in step S45.
• the XFRAC denotes a horizontal displacement at sub pixel level and Y FRAC a vertical displacement at sub pixel level.
• a scale factor, Si is also selected in step S45.
• step S46 is the second image scaled with the scale factor, Si, and resampled to position XFRAC and YFRAC in step S46. This results in a resampled scaled image, RSjI,.
• Correlation values, C XPIX, YPIX, i, between the first image and the resampled scaled image, RSjI, are next calculated in step S47.
• the correlation values in step S46 are calculated for different horizontal displacements at pixel level, Xpix, and different vertical displacements at pixel level, Y PIX , for the resampled scaled image RSjI.
• the correlation values between the first image and the resampled scaled image, RSjI, calculated in step S47 describes how similar the first and the resampled scaled image RSjI are for different displacements of the resampled scaled image RSjI.
• the maximum correlation value between the first image and the resampled scaled second image RSjI can be found among the correlation values calculated in step S47.
• the maximum correlation value between the first image and the resampled scaled image might be in a position where the resampled scaled image RSjI has been displaced a fraction of a pixel in vertical or horizontal direction.
• the maximum correlation value is therefore found at subpixel level which is necessary in order to find the scale factor that gives the best correlation between the first image and the resampled scaled image.
• the maximum correlation value at subpixel level in step S48 might be found with several methods, for instance with interpolation. Since interpolation is well known in the art it will not be further described herein.
• the X fractional pixel position, XFRAC, and the Y fractional pixel position, YFRAC, are also updated in step S48.
• the values are updated to the fractional pixel position XFRAC, YFRAC where the correlation has its maximum.
• a counter j is set to j + 1 .
• the steps S46 to S49 are then repeated a predetermined number of times. For each time steps S46 to S49 are preformed a higher correlation value Q is found and the fractional pixel positions associated with that correlation value.
• the predetermined number of times for repeating steps S46 to S49 can be set to many different values. If the steps S46 to S49 are repeated many times a more accurate value of the correlation C 1 may be achieved. On the other hand is a longer calculation time required if steps S46 to S49 are repeated many times.
• a next step S50 is a counter i set to i+1 .
• the steps S45 to step S50 are then repeated a predetermined number of times.
• a new scale factor is used each time the method executes steps S45 to S50.
• the new scale factor can be selected in many different ways. One way is to select the new scale using the half interval method. The half interval method is well known in the art and will therefore not be further described herein.
• Each time the method performs steps S45 to S50 is thus a maximum correlation value Q for a scale factor Si calculated.
• a step S51 is a largest maximum correlation value C MAX found among the maximum correlation values Q that was calculated each time the method performed steps S45 to S50.
• the step S51 is also the scale factor Si, MAX associated with the largest maximum correlation C MAX found.
• step S52 the time to go is calculated in step S52 using the scale factor Si, MAX that was found in step S51.
• the Time To Go, TTG may be calculated as inversely proportional to the scale factor Si, MAX -
• the time to go is calculated using formula (1 ) below:
• TTG Si, MAX*(TI - To)/(1 - Si, MAX);
• step S51 is the large maximum correlation value C MAX found among the correlation values C 1 and the scale factor Si, MAX associated with C MAX The resulting scale factor Si, MAX that best matches is then used in step S51 to calculate the time-to-go.
• the size in an image of the intruding aerial vehicle or object can also the size in an image of the intruding aerial vehicle or object be calculated.
• the size of the intruding aerial vehicle or object may according to embodiments of the invention be expressed as the number of pixels in the image.
• the size of the intruding aerial vehicle or object is then first estimated by using a gauss fitting method as will be described further down.
• the gauss fitting method it is assumed that the intruding aerial vehicle or object is gauss alike. The assumption works for intruding aerial vehicle or objects that are far away form the vehicle. Therefore the gauss fitting method is used as an initial estimate of the size of the intruding aerial vehicle or object.
• the initial object size is then updated based on the Scale facor Si, MAX that was found in step S51.
• the gauss fitting method is a non-linear problem which is solved by using the Levenberg-Marquardt method. Since the Levenberg-Marquardt method is well known in the art it will not be further described herein.
• the Levenberg-Marquardt method is well known in the art it will not be further described herein.
• the Levenberg-Marquardt method is well known in the art it will not be further described herein.
• the Levenberg-Marquardt method is well known in the art it will not be further described herein.
• the first parameter A is the amplitude of the gauss.
• the parameters SX and SY are the variances of the gauss along the X- and Y-dimension. .
• the parameters X and Y are the mean value of each dimension of the gauss.
• a start value of the size of the intruding aerial vehicle or object is needed.
• the initial size, ⁇ j n , CT ⁇ i n is estimated in an image n between To and Ti in the second sequence of images 211 from the detector.
• the initial size, ⁇ xin , ⁇ j n in embodiments of the invention is estimated by the detector 202.
• the size estimate from the detector may origin from a scale pyramid which results in one of the following values 2, 4, 8, 16 or 32 pixels (width and height estimated separately). The size estimate from the detector is good enough as start parameter.
• amplitude A another parameter in the Gaussian function is the amplitude A.
• a initial value for the amplitude A is needed.
• at least one initial amplitude, Aj n are therefore estimated by calculating a difference between a mean of a background and a mean of the intruding aerial vehicle or object in the image n in the third sequence of images between T 0 and TV
• the estimation of the initial amplitude, Ain is calculated in a step S51 b after the step of estimating an initial size.
• the intruding aerial vehicle object is in this embodiment then assumed to be located in the centre of the image n, i.e. in this embodiment the tracker always tries to centralize the intruding aerial vehicle object in the image n when tracking.
• At least one size, ⁇ n , ⁇ n , in an image n of the intruding aerial vehicle or object is calculated in a step S51c by adapting a Gaussian G n function to the intruding aerial vehicle or object in the image n.
• the adaption of the Gaussian function G n is done so that an error between the Gaussian function, G n , and the object in the image n in the third sequence of images between To and Ti is minimized.
• the initial amplitude Aj n and the initial size, 0 " ⁇ in, ⁇ in are used as start parameters for the Gaussian function. Start values Xin and Yj n for the position of the Gaussian function are also needed.
• the location of the Gaussian function i.e. the mean values X and Y assumed to me located in centre of the image n since the tracker in this embodiment centralizes intruding aerial vehicle of object in the image n.
• the start values Xj n and Yj n are set to the centre of the image n.
• the invention are several sizes, ⁇ n , ⁇ n , from several images ....n-2, n-1 , n filtered to achieve a filtered size, GXF, O ⁇ f .
• An advantage achieved by filtering several sizes ⁇ n , ⁇ n , from several images ....n-2, n-1 , n is that a better estimate of the size of the aerial vehicle or object, the filtered size GXF, ⁇ f , may be achieved.
• the gauss fitting method gives a good estimate of the size in an image of the aerial vehicles or objects if the aerial vehicle of object is far away form the vehicle.
• the aerial vehicle of object is more gauss like than when the aerial vehicle or object is closer to the vehicle.
• the aerial vehicle or object becomes more detailed and less gauss alike. The size of the aerial vehicle or object therefore needs to be estimated in another way at closer range.
• the idea according to the present invention is therefore to use the time-to-go value calculated in step S52 for updating the size ⁇ n+ i , CT ⁇ n+ i for consecutive images n+1. Since the time to go value calculated at a point of time Ti is calculated using the scale factor Si, MAX, the time to go value contains information about the scale change of the intruding aerial vehicle or object at the point of time Ti. If an image n+1 corresponds to the point of time Ti then the time to go value for Ti contains information about the scale change that can be used for updating the calculated size ⁇ n, ⁇ n size in image n of the intruding aerial vehicle of object.
• step S51c If the size of the intruding aerial vehicle or object in an image n is calculated in step S51c.
• the time to go value TTG n +i calculated for the image n+1 can be used for updating the size of the intruding aerial vehicle of object.
• the following formulas may be used in a step S53 (not shown), after the step of calculating time to go.
• ⁇ n , ⁇ n is the size of the intruding aerial vehicle of object in the image n and f is the image frequency.
• ⁇ xF, Oy? used instead of the size ⁇ n , ⁇ n when updating the size in step S53.

Landscapes

• Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Theoretical Computer Science (AREA)
• Aviation & Aerospace Engineering (AREA)
• Multimedia (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Image Analysis (AREA)
• Image Processing (AREA)
• Radar, Positioning & Navigation (AREA)
• Remote Sensing (AREA)

Priority Applications (7)

Applications Claiming Priority (1)

Related Child Applications (2)

Publications (1)

Family

ID=42739842

Family Applications (1)

Country Status (6)

Families Citing this family (3)

Citations (2)

Family Cites Families (16)

• 2009
  • 2009-03-18 EP EP09841974.0A patent/EP2409274B1/en active Active
  • 2009-03-18 CA CA2755737A patent/CA2755737A1/en not_active Abandoned
  • 2009-03-18 ES ES09841974.0T patent/ES2675516T3/es active Active
  • 2009-03-18 BR BRPI0924384-4A patent/BRPI0924384B1/pt active IP Right Grant
  • 2009-03-18 WO PCT/SE2009/050279 patent/WO2010107347A1/en not_active Ceased
  • 2009-03-18 US US13/257,416 patent/US8774457B2/en active Active
• 2014
  • 2014-05-28 US US14/289,201 patent/US9208690B2/en active Active

Patent Citations (2)

Non-Patent Citations (5)

Also Published As

Similar Documents

Legal Events