KR101603293B1 - Lane recognition method - Google Patents

Abstract

The method includes the steps of converting a shot image provided by a camera into a binary edge image, image processing the edge image, determining a position of a horizon line passing through a vanishing point and a vanishing point, processing the edge image, Selecting one left lane straight line and one right lane straight line by using a slope of a straight line and a brightness of a pixel located between a horizon position and a bonnet position in an edge image; Placing the left lane straight line and the right lane straight line between the horizon position and the bonnet position to grasp the first main point intersecting the horizon position and the second and third main points intersecting the bonnet position, Accumulating the first to third main points for the image, if the accumulated number of times is a preset number, Calculating a mean of one principal point, a mean of a plurality of second principal points and a plurality of third principal points, a first distance between an average of the first principal points and an average of the second principal points, Calculating a second distance between an average of the first principal points and an average of the third principal points, and if the first and second distances are less than or equal to a preset threshold value, Determining a straight line between the averages and a straight line between an average of the first principal points and an average of the third principal points as both lanes.

Description

Lane recognition method [0002]

The present invention relates to a lane recognition method using a photographed image.

Since the spread of automobiles, there has been constant interest in safety during driving. The lane of the road or the vehicle ahead is one of the most noticeable objects on the road. As various infrastructure technologies have developed, various derivative technologies for road safety such as lane recognition and vehicle recognition have been attracting attention.

Lane Departure Warning system and Forward Collision Avoidance Warning system have been recently put into practical use for smart car as an additional function to navigation or vehicle black box.

However, these driver aids have only been installed in some high-end vehicles to date, and are still not yet available in most vehicles, largely due to cost problems.

Korean Registered Patent No. 10-0975749 (registered on August 06, 2010) Korean Registered Patent No. 10-1169338 (registered on July 23, 2012)

SUMMARY OF THE INVENTION Accordingly, the present invention has been made to solve the above-mentioned problems occurring in the prior art, and it is an object of the present invention to provide a lane recognition method which has reliable accuracy and low cost.

A method of recognizing a lane according to an embodiment of the present invention includes the steps of capturing continuously a camera located in a vehicle with a time interval in a driving direction, receiving a photographed image from the camera, converting a received photographed image into a binary edge image, The method of claim 1, further comprising: processing the edge image to identify a position of a horizon line that is a straight line passing through the vanishing point and the vanishing point; image processing of the edge image to identify a bonnet position where the road meets the bonnet of the vehicle; Selecting one left lane straight line and one right lane straight line using the slope of the straight line and the brightness of the pixel located between the horizon position and the bonnet position and selecting the one left lane straight line and the one right lane straight line Is located between the horizon position and the bonnet position and intersects the horizon position The method comprising the steps of: identifying a first main point, second and third main points intersecting the bonnet position, accumulating the first to third main points for one photographed image, Calculating an average of a plurality of first main points accumulated in the number of times, an average of a plurality of second main points, and an average of a plurality of third main points, calculating an average of the first main points and an average Calculating a second distance between an average of the first principal points and an average of the third principal points, and if the first and second distances are less than or equal to a preset threshold value, And a straight line between an average of the second main points and a straight line between an average of the first main points and an average of the third main points as both lanes.

In the step of grasping the horizon position, a straight line which can be estimated as a lane in an image within a set range is subjected to Hough transform to the edge image, and one straight line having a slope of 30 to 90 degrees and a slope Extract one straight line from -30 to -90 degrees, and identify the intersection of the extracted two straight lines as a vanishing point.

In the step of grasping the bonnet position, a straight line in a horizontal direction is extracted from a plurality of edge images for each of a plurality of shot images, and a straight line without a position change among the extracted straight lines is grasped as a bonnet position.

The step of selecting one left lane straight line and one right lane straight line using the slope of the straight line and the brightness of the pixel may be performed by using one straight line made up of pixels having a slope of 30 degrees to 90 degrees and having the highest pixel brightness as the left lane And selects one straight line having a slope of -30 degrees to -90 degrees and pixels with the highest pixel brightness as the right lane.

According to the embodiment of the present invention, it is possible to improve the accuracy of the lane recognition and to use the lane recognition function at low cost by being mounted on the navigation device, the black box, or the like.

1 is a diagram showing the configuration of a lane recognizing apparatus installed in a vehicle according to an embodiment of the present invention.
2 is a flowchart of a lane recognition method according to an embodiment of the present invention.
FIG. 3 is a diagram showing main elements to be grasped through a lane recognition method according to an embodiment of the present invention.
4 is a diagram showing images generated by the lane recognition method according to the embodiment of the present invention.
5 is a diagram illustrating an image data accumulation method in a lane recognition method according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, which will be readily apparent to those skilled in the art. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Hereinafter, a lane recognition method according to an embodiment of the present invention will be described with reference to the accompanying drawings.

1, the lane recognizing apparatus 200 according to the embodiment of the present invention is used in connection with a camera 100 for the interior of the vehicle A. At this time, the camera 100 photographs the road running on the vehicle A, which is installed so as to face the front of the vehicle and is traveling at a constant speed or more. The lane recognizing device 200 is mounted in the interior of the vehicle A or used in a navigation or black box.

The lane recognizing apparatus 200 includes an edge image generating unit 210, a horizon line determining unit 220, a bonnet area determining unit 230, a main point extracting unit 240, an accumulated storage managing unit 250, And an extracting unit 260.

The edge image generating unit 210 receives the sequentially captured (i.e., sequenced) shot images from the camera 10 at intervals of time, that is, sequentially captured shot images, and performs canny edge detection detection method is used to generate an edge image. The canyon edge detection method is an edge detection method proposed by John F. Canny and is a publicly known technique.

The horizon line determining unit 220 grasps the horizon line which is the uppermost part where the road appears in the edge image generated by the edge image generating unit 210 and determines the position of the horizon line. 3, the horizon line determining unit 220 grasps the horizon line h1 appearing in a straight line shape and determines the position of the horizon line.

The bonnet area determining unit 230 determines the boundary between the bonnet and the road of the vehicle, that is, the bonnet position, from the photographed image photographed by the camera 100. [ For example, taking the shot image shown in FIG. 3 as an example, the bonnet area determination unit 230 grasps the bonnet position h2 appearing in a straight line.

The main point extracting unit 240 extracts two predicted lanes that determine the width of the lane in the edge image generated by the edge image generating unit 210, and extracts the two predicted lanes and the horizon line determining unit 220 Three main points are extracted using the determined horizon position and the bonnet position determined by the bonnet area determination unit 230, and an image frame in which the extracted three principal points are displayed is generated. Here, as shown in FIG. 3, the first principal point Pv is located at the horizon position, and the second and third principal points Pl and Pr are located at the bonnet position. Here, the first principal point Pv is a vanishing point for the road lane.

The cumulative storage management unit 250 cumulatively stores image frames (hereinafter referred to as "cumulative image frames") in which image frames generated by the main point extracting unit 240, i.e., three main points Pv, , And causes the cumulative storage amount (storage number) to be stored for the set storage amount (storage number). If the main point extracting unit 240 does not provide the three main points Pv, Pl and Pr as image frames but provides information on the three main points Pv, Pl and Pr, And the set storage amount is the set number of times.

5, the cumulative image management unit 250 sequentially accumulates the cumulative image frames according to the time sequence and stores the cumulative image frames as much as the set cumulative storage amount. When the cumulative image frames are stored as the set storage amount, , The new image frame (frame n) is stored and the oldest image frame (frame 1) is deleted.

The lane extracting unit 260 calculates each average value of the three main points displayed in each image frame of the set storage amount when the accumulated amount of the accumulated image frames is equal to or greater than the set storage amount by the accumulated storage management unit 250, A straight line between the main point Pv and the main point Pl and a straight line between the main point Pv and the main point Pr are extracted as a lane.

A lane recognizing method performed by the lane recognizing apparatus according to the embodiment of the present invention configured as described above will be described.

The camera 100 installed in the vehicle shoots forward while being driven at a constant speed to generate a photographed image (S201). At this time, the photographed image is, for example, as shown in FIG. 4 (a), including the shape of the front road, and is composed of digital data.

The photographed image is input to the edge image generating unit 210 of the lane recognizing apparatus 200. The edge image generating unit 210 performs the image processing of the photographed image with the canyon edge detection method to generate an edge image (S202) . 4 (b) shows an edge image obtained by image processing the photographed image shown in Fig. 4 (a) by the canyon edge detection method.

The edge image generated by the edge image generation unit 210 is input to the horizon line determination unit 220, the bonnet area determination unit 230, and the main point extraction unit 220.

Horizon line determination unit 220 performs Hough transformation on the received edge image to extract a straight line that can be guessed as a lane. At this time, the horizon line determination unit 220 extracts a straight line using a slope. In general, the left lane and the right are directed toward the center of the screen when shooting from the front. That is, the slope of the left lane is within 90 degrees, and the slope of the right lane is greater than 90 degrees and smaller than 180 degrees.

Therefore, the horizon line determination unit 220 extracts a straight line having an angle between 30 degrees and 90 degrees with one straight line corresponding to the left lane, and extracts a straight line corresponding to the right lane from an angle between -30 degrees and -90 degrees . Here, the horizon determining unit 220 determines whether the left side is set to the left from the pixel located on the center line (vertical line) of the screen or the vertical line of the setting position to extract one straight line corresponding to the left lane and the straight line corresponding to the right lane The first range is scanned by the first range and the right is scanned by the first range set in the right direction. Then, during the scan, edge pixels that meet first for each row are extracted. The shape of the straight line appears when the extracted edge pixels of each row are combined. If a straight line is not extracted, the scan range is expanded from the first range to the second range and the scan is performed.

4 (c) shows an image generated by using a straight line corresponding to the left lane and a straight line corresponding to the right lane extracted from the edge image shown in (b) of FIG.

Then, the horizon line determining unit 220 extends the straight line corresponding to the extracted left lane and the straight line corresponding to the right lane, as shown in FIG. 4 (d), and grasps the intersection point of the two straight lines, And determines a horizontal straight line passing through the vanishing point (horizontal straight line on the screen) as a horizontal line h1 (S203).

On the other hand, the horizontal line determination unit 220 extracts the right lane and the left lane for each of the plurality of shot images (i.e., the plurality of image frames) sequentially input, searches for the intersection of the extracted right lane and left lane, The intersection with the largest number of intersecting straight lines can be determined as the vanishing point.

The bonnet area determination unit 230 extracts a straight line in a horizontal direction (i.e., a horizontal direction) from a plurality of edge images for each of a plurality of shot images, and grasps a straight line without a position change among the extracted straight lines as a bonnet position ).

Here, the horizontal axis and the horizontal edge histogram are obtained for each edge image, and when the maximum value is obtained, the maximum value coincides with the bonnet position. Of course, if the maximum value of the histogram is less than the set value, the average of the maximum value of the histogram can be calculated and the straight line corresponding to the average value can be set as the bonnet position.

The horizon position h1 of the horizon line determining unit 220 and the bonnet position h2 of the bonnet unit determining unit 230 are provided to the main point extracting unit 240. [ The main point extracting unit 240 selects one of the two lane pixels for each row from the edge image (S205). At this time, the pixels of the detected lane are extracted from the pixels located between the horizon position h1 and the bonnet position h2, and the pixels to be extracted are searched toward the left and right around the coordinates of the vanishing point, It is the brightest pixel among the brightest pixels.

The main point extracting unit 240 selects both lane pixels and extracts straight lines by applying Hough transform (S206). The main point extracting unit 240 extracts one straight line having a slope of 30 to 90 degrees from the extracted straight lines (a straight line corresponding to the left lane And a straight line having a slope of -30 degrees to -90 degrees (a straight line corresponding to the right lane) (S207).

The main point extracting unit 240 then extracts the straight line of the extracted left lane and the straight line of the right lane from the horizon line position h1 obtained from the horizon line determining unit 220 and the bonnet position Pl and Pr of the first to third main points Pv, Pl and Pr are detected in step S208 and the detected first to third main points Pv, Pl and Pr are stored in the cumulative storage management part 250 (S209).

The cumulative storage management unit 250 determines whether the cumulative storage amount is equal to or greater than the preset storage amount (S210). If the cumulative storage amount is equal to or greater than the preset storage amount, the cumulative storage management unit 250 stores the cumulatively stored first to third main points Pv, Pl, (260) to instruct the extraction of the lane.

The lane extracting unit 260 calculates an average value of the accumulated first to third main points Pv, Pl and Pr and calculates a first major point average AVGPv, a second major point average AVGPl, The point average AVGPr is grasped (S211). Then, the lane-extracting unit 260 extracts a lane between the first main point average AVGPv and the second main point average AVGPl, the first main point average AVGPv and the third main point average AVGPr, (Left lane) between the first main point average (AVGPv) and the second main point average (AVGPl) if the calculated first and second distances are smaller than the threshold, (Right lane) between the first principal point average (AVGPv) and the third principal point average (AVGPr) is a correct lane (S212).

It will be understood by those skilled in the art that the foregoing description of the present invention is for illustrative purposes only and that those of ordinary skill in the art can readily understand that various changes and modifications may be made without departing from the spirit or essential characteristics of the present invention. will be. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive. For example, each component described as a single entity may be distributed and implemented, and components described as being distributed may also be implemented in a combined form.

The scope of the present invention is defined by the appended claims rather than the detailed description and all changes or modifications derived from the meaning and scope of the claims and their equivalents are to be construed as being included within the scope of the present invention do.

100: camera 200: lane recognition device
210: edge image generating unit 220: horizon determining unit
230: bonnet area determining unit 240: main point extracting unit
250: cumulative storage management unit 260: lane-

Claims (4)

The camera located in the vehicle continuously photographs at a time interval in the running direction,
Receiving a photographed image from the camera,
Converting the received photographed image into a binary edge image,
Processing the edge image to obtain a vanishing point and a horizon line position passing through the vanishing point,
Image processing of the edge image to grasp the bonnet position, which is a straight line where the road meets the bonnet of the vehicle,
Selecting one left lane straight line and one right lane straight line using the slope of the straight line and the brightness of the pixel located between the horizon position and the bonnet position in the edge image,
Placing the one left lane straight line and the one right lane straight line between the horizon position and the bonnet position to form a first main point intersecting the horizon position and a second main point intersecting the bonnet position Step to grasp,
Cumulatively storing the first to third principal points for one shot image;
Calculating a mean of a plurality of cumulative first stored points, an average of a plurality of second significant points, and an average of a plurality of third significant points when the accumulated cumulative number of times is a preset number,
Calculating a first distance between an average of the first principal points and an average of the second principal points and a second distance between an average of the first principal points and an average of the third principal points,
A straight line between an average of the first principal points and an average of the second principal points and a straight line between an average of the first principal points and an average of the third principal points, A step of judging as a lane
The lane recognition method comprising: The method of claim 1,
The step of grasping the horizon position
A straight line having a slope of 30 to 90 degrees and a straight line having a slope of -30 to -90 degrees among the extracted straight lines are extracted from the extracted image, Respectively, and recognizing the intersection of the extracted two straight lines as a vanishing point. 3. The method according to claim 1 or 2,
The step of grasping the bonnet position
A lane recognition method for extracting a straight line in a horizontal direction from a plurality of edge images for each of a plurality of shot images and grasping a straight line without a position change among the extracted straight lines as a bonnet position. 4. The method of claim 3,
The step of selecting one left lane straight line and one right lane straight line using the slope of the straight line and the brightness of the pixel
One straight line made up of pixels with a slope of 30 degrees to 90 degrees and having the highest pixel brightness is selected as the left lane and one straight line made up of pixels having a slope of -30 degrees to -90 degrees and the brightest pixel is called a right lane A lane recognition method.

Images