DE102009009099A1 - Potential parking space detecting method for vehicle, involves carrying out search about coordinates of parking space in regional map, where coordinates of parking space is represented between coordinates of parking barriers in map - Google Patents

Abstract

The method involves providing a vehicle with a mono camera (1), which is arranged at a rear side of the vehicle and detects surroundings of the vehicle at discrete time points. Coordinates (7a, 7b) of the images (2) taken by the mono camera are inserted into a regional map (8), where the coordinates represent potential parking barriers. A search about coordinates (9) of a potential parking space is carried out in the regional map, where the coordinates of the parking space is represented between the coordinates of the parking barriers in the map.

Description

Technical area

The The invention relates to a method for detecting potential Parking areas according to the preamble of the claim 1.

State of the art

• a. Aufnehmen einer Abbildung mittels der Monokamera;
• b. Bewegen der Monokamera entlang eines Fahrwegs;
• c. Aufnehmen einer weiteren Abbildung mittels der Monokamera;
• d. Auswerten der Abbildung und der weiteren Abbildung mittels einer Auswerteeinheit, wobei beim Auswerten der Abbildung und der weiteren Abbildung jeweils eine invers perspektivische Projektion ausgeführt wird und entsprechend eine invers perspektivische Projektions-Abbildung und eine weitere invers perspektivische Projektions-Abbildung erzeugt werden, und wobei der zurückgelegte Fahrweg berücksichtigt wird;
• e. Bilden einer Differenzabbildung zwischen der invers perspektivischen Projektions-Abbildung und der entsprechend des Fahrwegs „verschobenen” weiteren invers perspektivischen Projektions-Abbildung; und
• f. Ermitteln von Abmessungen eines Parkraums anhand der Differenzabbildung.

From the DE 10 2005 022 882 A1 a method for parking space measurement by means of a monocamera is known. The method comprises the following steps:

• a. Taking a picture by means of the monocamera;
• b. Moving the monocamera along a driveway;
• c. Taking a further picture by means of the monocamera;
• d. Evaluating the image and the further image by means of an evaluation unit, wherein in evaluating the image and the further image in each case an inverse perspective projection is performed and correspondingly generated an inverse perspective projection image and another inverse perspective projection image, and wherein the traversed Track is taken into account;
• e. Forming a difference map between the inverse perspective projection map and the further inverse perspective projection map corresponding to the travel path; and
• f. Determining dimensions of a parking space based on the difference image.

adversely This affects the loss of information through the perspective Projection off. In addition, a surveillance of the parking space condition not possible. Furthermore the process is computationally expensive.

Object of the invention

Of the Invention is the object of providing a method underlying, which is a more precise (and with less computational effort connected) detection of a potential parking area guaranteed.

solution the task

The Task is achieved by a method according to claim 1 solved.

Advantages of the invention

through of the method according to the invention become the coordinates of a potential parking area based on an area map. The environment map contains a global coordinate system, For example, a world coordinate system, in which coordinates of the rear of the vehicle arranged monocamera taken picture be registered.

in this connection only those coordinates are entered in the map, which represent a potential parking obstacle.

The Search for a potential parking area and identify the corresponding coordinates are based on the area map.

In an advantageous embodiment of the invention will the image taken by the monocamera both an image processing unit as well as an equalization image processing unit.

In the image processing unit is rasterized the image, i. H. it will in lines that run transversely to the direction of travel of the vehicle, and divided into columns that run in the direction of travel. From the rastering resulting cells are subsequently binary coded. The image processing unit transmits to the map of the surroundings the coordinates of the cells to which a certain binary value representing a potential parking obstacle was assigned.

In of the equalization image processing unit, the image is acquired by means of a equalized equalization algorithm, so that straight lines in the Real environment in the rectified image also form straight lines. Then the rectified image is rasterized and binary coded.

The Equalization image processing unit transmitted to the Environment map the coordinates of the cells to which a particular binary value representing a potential parking obstacle was assigned.

to Distinction between the coordinates of the image processing unit and the coordinates of the equalization image processing unit, the Coordinates are marked, for example, the coordinates become of the equalization image processing unit. Further differentiation are also possible.

In the environment map, the coordinates of the potential parking area are adjacent to the coordinates transmitted from the image processing unit and the equalization image processing unit, and within the potential parking area only one unobserved number of coordinates of the image processing unit and the Entzer image processing unit is included.

In a further advantageous embodiment of the invention the search for a potential parking area in the area map takes place two stages.

In a first stage is determined by means of the coordinates of the equalization image processing unit a starting edge, which runs transversely to the direction of travel of the vehicle and contains a minimum number of adjacent coordinates, searched. The search for a starting edge can either already in the equalization image processing unit or first in the environment map respectively.

In a second stage is determined by means of the coordinates of the image processing unit a extending from the starting edge in the direction of travel of the vehicle Area examined. If in this area the coordinates of Image processing unit are such that an arrangement of the potential parking area outside or at a slight overlap of these coordinates - within of the range is possible, the coordinates of the potential Parking area determined.

In a further advantageous embodiment of the invention The update of the map corresponds to the discrete ones Recording times of the pictures by the monocamera. By means of this Updating can be changes of potential Parking obstacles, such as a pedestrian or another in the parking or Ausparkvorgang befindliches Vehicle, to be isolated.

In a further advantageous embodiment of the invention the recorded image is interpreted as a gray image, so that a Reduction of file size allows becomes.

In a further advantageous embodiment of the invention becomes the recorded image of a noise reduction, for example by means of a Gaussian folding operator, so that a Smoothing the image, minimizing the effects of noise and a reduction in vulnerability Sun effects / glare effects are enabled.

In a further advantageous embodiment of the invention the binary coding of the cells is done by means of static Procedures, such as averaging and subsequent determination the standard deviation. This is an adaptive behavior regarding guaranteed the changing environment.

In a further advantageous embodiment of the invention become only the coordinates of the centers the cells are transmitted to the area map, so that the Computing effort is reduced.

In a further advantageous embodiment of the invention the search for the coordinates of the map takes place in the area map potential parking area exclusively in a definable relevant area. By definition of the relevant area For example, the so-called. Driving tube, the area the opposite lane or areas with low pixel resolution per meter are excluded, so that on the one hand the computing time and on the other hand the computational effort is reduced and still the necessary accuracy is achieved.

In a further advantageous embodiment of the invention are in the image processing unit and in the equalization image processing unit the coordinates only in the definable relevant Area determined.

In a further advantageous embodiment of the invention The area map includes both a list of coordinates as also an image of the coordinates in the form of a virtual image.

In a further advantageous embodiment of the invention the search is done in the area map depending on driving data of the vehicle, for example from a navigation system, CAN or individual inputs of the driver.

The The method according to the invention can be or three-dimensional image data. In three-dimensional Image data can be either a two-dimensional or a three-dimensional Environment map to be generated. There is the possibility to map this with a grid-based map in all dimensions or to conserve resources with a so-called 2½ D card implement, where in addition to the occupancy state of the cell height information is carried. The 2½ D card contains thus in each case one dimension for the x- and y-axis and a half dimension for additional attribute information, such as altitude information (alternative to z-axis).

advantageously, In addition, the parking space can also after a pass and monitored during the parking process.

drawings

It demonstrate:

1 a flowchart of a method according to the invention;

2 : a section of an environment map.

The 1 shows a flowchart of a method according to the invention. A rear side of a vehicle arranged monocamera 1 , for example, with an opening angle of at least 130 °, takes pictures at discrete times 2 from the rear environment of the vehicle.

zur Glättung des Graubildes 2a erzielt werden. Hierdurch werden Rauscheffekte minimiert und Anfälligkeiten gegenüber Sonneneinflüssen (Blendeffekten) reduziert.The picture taken 2 is called gray image 2a interpreted, and then the gray image 2a a noise reduction 3 subjected. The noise reduction 3 for example, by means of a Gaussian folding operator with a 3x3 mask

for smoothing the gray picture 2a be achieved. As a result, noise effects are minimized and vulnerabilities to sun effects (glare effects) are reduced.

The noise reduction can alternatively (in 1 not shown) also on the picture 2 be applied.

The picture 2 , the gray picture 2a or the image subjected to the noise suppression 2 B is both an image processing unit 4 as well as an equalization image processing unit 5 fed.

In the image processing unit 4 becomes the picture 2 B a screening 4a subjected. By means of screening 4a becomes the picture 2 B in lines that extend transversely to the direction of travel of the vehicle, and in columns that run in the direction of travel of the vehicle divided. The resulting cells 6 , which for example have a size of 5 × 5 pixels, are then binary-coded.

By means of binary coding 4b will the resolution of the picture 2 B reduced. In binary coding 4b Statistical methods are used. For example, for each cell 6 a mean and then a standard deviation calculated. The result of the static method, for example the value of the standard deviation, is compared with a definable threshold value. The threshold may be set to correspond to a potential parking obstacle. If the threshold value is exceeded, the cell becomes 6 assigning a particular binary value (eg, the value "1") representing a potential parking obstacle; otherwise the cell will 6 the other binary value (eg, the value "0") that does not represent a potential parking obstacle.

Subsequently, a search is carried out 4c by coordinates 7a attached to an environment map 8th be transmitted. For the search 4c become the cells 6 go through systematically. Unless a cell 6 which was assigned the particular binary value (for example, the value "1") becomes the center of the cell 6 certainly. If the coordinate system of the image does not match the coordinate system of the map, the coordinates become 7a the center on the basis of the homography, ie projection matrix resulting from the camera calibration, in the coordinate system, for example, world coordinate system, the map of the surroundings 8th umgewandet. These coordinates 7a will be sent directly to the area map 8th transmitted. Alternatively, the coordinates 7a before submitting an examination (in 1 not shown). For this, it is checked if the coordinates 7a within a definable relevant area. The relevant area may, for example, be outside the driving route and below a distance threshold (distance limit).

In the equalization image processing unit 5 becomes the picture 2 B an equalization 5a subjected. For this purpose, a known equalization algorithm is used. By means of equalization 5a Also, lines in the real environment become straight lines in the rectified image 2c shown.

The rectified picture 2c becomes a rasterization 5b that the rasterization 4a corresponds, subjected. The resulting cells 6 become a binary coding 5c , that of binary coding 4b corresponds, subjected.

Subsequently, a search is carried out 5d after a starting edge. This is what the cells are for 6 go through line by line and in the direction of travel.

The search 5d Alternatively, it can be restricted to one or the definable relevant area.

Unless in the search 5d a number of contiguous cells 6 to which the specific binary value (for example, the value "1") has been determined, which exceeds a definable minimum number, the coordinates become 7b analogous to the image processing unit 4 determined, converted if necessary and then to the area map 8th transmitted. Alternatively, the search 5d be restricted to a relevant area.

The environment map 8th can be designed in two parts.

A first part forms a list containing all coordinates 7a . 7b the image processing unit 4 and the equalization image processing unit 5 contains.

A second part forms a virtual image in which the coordinates 7a . 7b in the form of references, for example points.

The management of the area map 8th may take into account the data information age (for example, recording time) and / or the distance of the coordinates 7a . 7b to the vehicle.

The environment map contains as input variables 8th thus on the one hand by the image processing unit 4 the coordinates 7a of cells 6 or cell centers associated with a particular binary value (eg, the value "1") representing a potential parking obstacle and, on the other hand, the equalization image processing unit 5 the coordinates 7b of cells 6 or cell centers associated with a particular binary value (eg, "1") representing a potential parking obstacle, and having a minimum number of cells contiguous in a row 6 was exceeded with this assigned specific binary value.

For the distinction of origin, the coordinates 7a . 7b be marked accordingly.

Based on the in the area map 8th contained coordinates 7a . 7b becomes a search 8a by coordinates 9 a potential parking area carried out. The coordinates 9 of the potential parking area do not overlap or only with a small number with the coordinates 7a that represent a potential parking obstacle.

The search 8a is then in the direction of travel to the coordinates 7b the equalization image processing unit 5 and line by line (ie, transverse to the direction of travel).

The search 8a can for example also be done by driving data of the vehicle. These include, for example, the vehicle speed, data from a navigation system (for example, highway driving, one-way street or absolute stop) and manual input from the driver (for example, "parking space search").

The column-wise boundary, d. H. the limit across the direction of travel of the vehicle, by means of a definable parking width and / or a or the definable relevant area.

Of the relevant area comprises at least the minimum dimension (transverse to the direction of travel and in the direction of travel) of a potential parking area.

In the search 8a becomes the number of coordinates in a line 7b determined. If the number does not exceed a maximum number, the search becomes 8a continued in the next line. Otherwise, the search will be 8a aborted or the current line forms a new starting edge and the search 8a will continue from this new starting edge. The search 8a is continued line by line until a predetermined parking length (in the direction of travel of the vehicle) is reached. Subsequently, the coordinates describing the potential parking area become 9 determined.

In 2 is a part of an environment map 8th shown schematically.

For distinguishing the coordinates 7a . 7b are the coordinates 7a the image processing unit 4 as circles and the coordinates 7b the equalization image processing unit 5 shown as squares.

The environment map 8th includes an x-axis, which runs in the direction of travel of the vehicle, and a y-axis, which runs transversely to the direction of travel of the vehicle.

In the area map 8th are those recorded at a discrete time and within a relevant range 10 located coordinates 7a . 7b shown.

The search 8a begins by determining the from the equalization image processing unit 5 transmitted coordinates 7b , in the 2 are shown as squares and a starting edge 11 form. Subsequently, the relevant area 10 line by line (along the y-axis y), starting with the line above (with respect to the 2 ) of the starting edge 11 and the number of the image processing unit 4 transmitted coordinates 7a , in the 2 are shown as circles. The number is compared with a threshold, which in the example shown here has the value "2". In the second line above the starting edge 11 There are three coordinates 11a , The threshold is thus exceeded. The search 8a can either be canceled or continued such that this line has a new starting edge 11a forms. The search 8a is continued until the dimensions (in x-axis x, ie in the direction of travel of the vehicle) of a potential parking area 12 are reached.

In the in 2 example shown is a coordinate 7a within the potential parking area 12 , Because the number of coordinates 7a in this line (here: one) the threshold value (here: two), this was used to position the potential parking area 12 not considered.

Provided the minimum dimension in the x-axis x of the potential parking area 12 not within the relevant range 10 would lie (this case is in 2 not shown), so the search 8a be paused and cached. After updating the area map 8th , ie after the processing of the image taken at the subsequent discrete time, the search 8a to be continued.

1

monocamera

2

image

2a

gray image

2 B

image

2c

equalized image

3

noise reduction

4

Image processing unit

4a

scanning

4b

binary coding

4c

search

5

Equalization image processing unit

5a

equalization

5b

scanning

5c

binary coding

5d

search

6

cell

7a

coordinates

7b

coordinates

8th

map

8a

search

9

coordinates

10

relevant Area

11

start edge

11a

new start edge

12

potential parking area

x

X axis

y

y-axis

QUOTES INCLUDE IN THE DESCRIPTION

This list The documents listed by the applicant have been automated generated and is solely for better information recorded by the reader. The list is not part of the German Patent or utility model application. The DPMA takes over no liability for any errors or omissions.

Cited patent literature

• DE 102005022882 A1 [0002]

Claims (12)

Method for determining a potential parking area ( 12 ) for a vehicle which has a rear-mounted monocamera ( 1 ), which at discrete times images ( 2 ) of an environment, characterized in that coordinates ( 7a . 7b ) of the monocamera ( 1 ) recorded image ( 2 ) in an environment map ( 8th ), the coordinates ( 7a . 7b ) represent a potential parking obstacle, and in the environment map ( 8th ) a search ( 8a ) by coordinates ( 9 ) of a potential parking area ( 12 ) he follows. Method according to claim 1, characterized in that the recorded image ( 2 ) both an image processing unit ( 4 ) as well as an equalization image processing unit ( 5 ), and where the map of the surroundings ( 8th ) both coordinates ( 7a ) of cells ( 6 ) of a particular binary value representing a potential parking obstacle in the environment map ( 8th ), the rasterized and binary coded image ( 2 ) of the image processing unit ( 4 ) as well as coordinates ( 7b ) of cells ( 6 ) of the determined binary value representing a potential parking obstacle in the environment map ( 8th ), the equalized, rasterized and binary coded picture ( 2c ) of the equalization image processing unit ( 5 ), and in the environment map ( 8th ) the coordinates ( 9 ) of the potential parking area ( 12 ) between the coordinates ( 7a . 7b ) of the lines ( 6 ) of the particular binary value representing a potential parking obstacle. Method according to claim 2, characterized in that the search ( 8a ) to a potential parking area ( 12 ) in the area map ( 8th ) is carried out in two stages, wherein in a first stage by means of the equalization image processing unit ( 5 ) ( 7b ) a starting edge ( 11 ), which is transverse to a direction of travel of the vehicle and a minimum number of adjacent coordinates ( 7b ), and wherein in a second stage by means of the image processing unit ( 4 ) ( 7a ) in the area map ( 8th ) from the starting edge ( 11 ) in the direction of travel is examined, wherein the search ( 8a ), provided that a minimum number of coordinates ( 7a ) in a line that is transverse to the direction of travel of the relevant area is exceeded, and the search ( 8a ) is continued otherwise. Method according to at least one of the preceding claims, characterized in that an update of the map of the surroundings ( 8th ) with the discrete recording times of the images ( 2 ) by means of the monocamera ( 1 ) corresponds. Method according to at least one of the preceding claims, characterized in that the recorded image ( 2 ) as a gray image ( 2a ) is interpreted. Method according to at least one of the preceding claims, characterized in that the recorded image ( 2 ) of noise reduction ( 3 ). Method according to at least one of the preceding claims, characterized in that the binary coding of the cells ( 6 ) is carried out by statistical methods. Method according to at least one of the preceding claims, characterized in that only the coordinates ( 7a . 7b ) of the centers of the cells ( 6 ) to the map of the surroundings ( 8th ). Method according to at least one of the preceding claims, characterized in that in the environment map ( 8th ) the search ( 8a ) according to the coordinates ( 9 ) of the potential parking area ( 12 ) exclusively in a relevant area ( 10 ) he follows. Method according to claim 8, characterized in that both in the image processing unit ( 4 ) as well as in the equalization image processing unit ( 5 ) exclusively in the relevant field ( 10 ) the coordinates ( 7a . 7b ) of the cells ( 6 ) be determined. Method according to at least one of the preceding claims, characterized in that the environment map ( 8th ) both a list of coordinates ( 7a . 7b ) as well as an image of the coordinates ( 7a . 7b ) as a virtual image. Method according to at least one of the preceding claims, characterized in that the search ( 8a ) in the area map ( 8th ) takes place depending on driving data of the vehicle.