KR101840974B1 - Lane identification system for autonomous drive - Google Patents

Abstract

The present invention relates to a system for identifying a lane of a road which comprises: a camera configured to photograph a road from a driving vehicle to generate an image; an image correcting unit configured to delete unnecessary pixels of the image to simplify the image; a lane detecting unit configured to track pixels of the image to detect a lane; a lane determining unit configured to analyze pixels configuring the lane of the image to determine a kind of the lane and a type of the road; and a lane verifying unit configured to filter an error of a lane identified by the lane determining unit. According to the present invention, a photographed image is corrected to increase processing speed and identification power. Colors of pixels configuring a lane, a kind of a lane and a type of a road are identified in a dense type to rapidly and accurately cope with a variety of road environments.

Description

[0001] Lane identification system for autonomous drive [

The present invention relates to a lane identification system applied to a vehicle for autonomous driving. More specifically, the present invention relates to a lane identification system, The present invention relates to a lane identification system for an autonomous lane that identifies types of dotted lines, solid lines, center lines, and dedicated lanes, thereby changing the lanes and promoting overall safe driving.

Autonomous running means that the driver automatically operates the vehicle such as running, stopping, turning, accelerating or decelerating without operating the vehicle directly, even if there is no driver or the driver. The main tasks of these intelligent vehicles are to maintain driving lanes, to establish safety distances to adjacent vehicles, to detect and avoid collision obstacles, to control collision avoidance, and to control vehicle speed in response to traffic conditions and road conditions.

In recent years, development of technology has led to the development of a safe driving assistance system such as a lane departure warning system (LDWS) or a lane keeping system, a vehicle automatic control system, and the like. Especially, the detection of the driving lane is one of the core technologies for solving the main problems in autonomous vehicles, and many researches are actively pursued due to international interest.

That is, since the detection of the driving lane has a great influence on the safe driving, the accurate driving lane is detected by using various sensors in order to estimate and determine the position of the lane. For example, various sensors such as an image sensor, a RADAR, or a LIDAR sensor are used to implement an autonomous vehicle control system in a form of a lane detection or an object recognition in front of the vehicle, either alone or in a fused form.

In particular, vision - based systems using image sensors have been widely used because they can extract a lot of information at low cost and can utilize various vision processing algorithms. In other words, it is very important to detect the color of the driving lane or the correct lane in the vision-based lane identification system. In recent years, in order to improve the discrimination power of lanes, Korean Registered Patent No. 10-1121777 (entitled "Lane Detection Method") and Korean Patent Registration No. 10-1094752 (entitled "Statistical Lane classification method using the model).

The proposed document extracts feature information from an image, and approximates it by application of update algorithms such as a parametric model and matching, Kalman filter or particle filter for lane detection, And non-parametric model matching by conversion such as HT (Hough Transform). However, since the lane is identified in the captured perspective image, there is a problem that the throughput is large and the error rate is high because the lane becomes narrower as the distance lengthens.

Korean Registered Patent No. 10-1121777 (Title: Lane Detection Method) Korean Patent Registration No. 10-1094752 (entitled "Lane Classification Method Using Statistical Model of HIS Color Information)

SUMMARY OF THE INVENTION Accordingly, the present invention has been made to solve the above-mentioned conventional problems, and it is an object of the present invention to improve the discrimination power of an overall lane by eliminating unnecessary pixels of a road image photographed by a camera, It is an object of the present invention to provide an autonomous lane identification system that not only changes lanes but also ensures safe driving.

In order to achieve the above object, the present invention provides a system for identifying a lane on a road, comprising: a camera for photographing a road from a driving vehicle to generate an image; The image is copied to two or more images in order to simplify and delete unnecessary pixels of the image, and the duplicated color image is converted into a gray image, and then the horizontal direction sole edge is found and the intensity of the edge gradient is set as a threshold value The binarized image is converted into the HLS color image, and the S channel is set to the threshold value to fuse the achromatic image with the chroma removed, and the binarized image and the achromatic image are merged into one, An image correction unit for converting a region of interest into a back projection to process the image as a bird's-eye image with the perspective removed; A lane detection unit for tracking a pixel to be compared with an average pixel in the image to obtain a cumulative histogram in a vertical direction and detecting a lane at a maximum peak position of the histogram around the boundary of the region of interest; The lane type is determined by the color of the pixels forming the lane of the image, and the RANSAC algorithm is applied to the pixels forming the lane of the image to calculate the curvature corresponding to the neighboring pixels between the pixels A lane determining unit for determining the type of road; And a lane verifying unit for loading the information of the color and the shape of the lane derived by the lane determining unit and for comparing the color and shape of the lane that are currently derived to filter the error to thereby make a final judgment .

delete

delete

delete

delete

delete

delete

delete

delete

delete

It should be understood, however, that the terminology or words of the present specification and claims should not be construed in an ordinary sense or in a dictionary, and that the inventors shall not be limited to the concept of a term It should be construed in accordance with the meaning and concept consistent with the technical idea of the present invention based on the principle that it can be properly defined. Therefore, the embodiments described in the present specification and the configurations shown in the drawings are merely the most preferred embodiments of the present invention, and not all of the technical ideas of the present invention are described. Therefore, It is to be understood that equivalents and modifications are possible.

As described above, the present invention improves the processing speed by eliminating unnecessary pixels of the road image photographed from the camera, and subsequently improves the discrimination power of the lane by switching the plane to the plane by removing the perspective of the image, It identifies the types of roads such as straight roads and curved roads along with types of lanes such as dotted lines, solid lines, center lines, and exclusive lanes based on the color and density of the pixels forming the lane. Thereby providing an effect that can be achieved.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram generally showing a system according to the present invention; FIG.
2 is a photograph showing an image of a frame unit generated by the camera of the present invention.
3 is a photograph showing a process of the image correction unit of the present invention.
4 is a photograph showing a process of the lane detecting unit of the present invention.
5 and 6 are photographs showing a process of the lane determining unit of the present invention.
Fig. 7 is a block diagram generally showing a modification of the system of the present invention; Fig.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The present invention relates to a lane identification system that is applied to a vehicle and is designed to make an autonomous driving. The lane identification system includes a camera 10, an image correction unit 20, a lane detection unit 30, Of the lane identification system for autonomous driving.

The system of the present invention improves the processing speed and discrimination power by correcting the photographed image and identifies the type of lane and the type of road by the color and density of the pixels forming the lane so that the driving control And the like.

First, the camera 10 according to the present invention is installed forward or backward in the interior or exterior of the vehicle as shown in FIG. 1 to photograph the road. That is, the camera 10 photographs a road from a driving vehicle as shown in FIG. 2 to generate an image. The image is used to identify a lane. In addition to the camera 10, a RADAR or a LIDAR may be used for identification of a lane or an object in front of the vehicle.

The image correcting unit 20 according to the present invention corrects the processing speed and discrimination power of the image photographed by the camera 10 as shown in FIG. 1, thereby eliminating unnecessary pixels of the image and simplifying it. As shown in FIG. 3, the image correcting unit 20 fuses the binarized image obtained by duplicating the image to two or more images and processing the Sobel mask, and the achromatic image obtained by removing the saturation.

That is, the original image of FIG. 2 is duplicated in two or more. In the first duplicate image of the duplicated image, the original color image is converted into a gray image as shown in FIG. 3A, and then the edge is searched for a horizontal direction. And the second duplicate image is set as a threshold value after converting the original color image into the HLS color image as shown in FIG. 3B.

Then, the binarized first replica image and the second replica image of the achromatic color are fused into a single image as shown in FIG. 3C. Then, only a region of interest (ROI) set in advance as shown in FIG. 3D is converted into a back projection as shown in FIG. 3E at a position where a lane can be displayed in the fusion image. The corrected image converted to the back projection is a bird's-eye view that removes the perspective and views the road from the top view.

Next, the lane detecting unit 30 according to the present invention tracks the pixels of the image corrected by the image correcting unit 20 as shown in FIG. 1 to detect lanes. As shown in FIG. 4, the lane detecting unit 30 detects a lane of a pixel to be compared with an average pixel of the image.

That is, a cumulative histogram is obtained in the vertical direction in the image using the slide (red block) window technique as shown in FIG. 4A, and the maximum peak position of the histogram is found around the boundary of the ROI, The initial position is detected.

Then, as shown in FIG. 4C, the image is shifted upward and continuously detected along the maximum peak position of the histogram. Here, the detection slides of the pair (left and right) do not overlap with each other, but move upward by the size of the preset slide.

Finally, the lane determining unit 40 according to the present invention analyzes the pixels constituting the lane of the image detected by the lane detecting unit 30 as shown in FIG. 1, and determines the type of the lane and the type of the road. As shown in FIG. 5, the lane determining unit 40 determines the type of the road by determining the type of lane in the dense form of the color of the pixels forming the lane, and calculating the curvature of the dense form.

When the lane is detected in all of the slides (red blocks) in the form of a lane, it is determined as a lane of a solid line shape as shown in FIG. 6A. Otherwise, a lane of a dotted line is determined as shown in FIG. 6B. Here, even if lanes are detected on all the slides, if the coordinates of the slides are based on the X axis, the lane is determined as a zigzag shape as shown in FIG. 6C.

For the color of the lane, the representative color (blue, yellow, white) of the lane is determined according to the following formula based on the coordinates (slides) detected in the corrected image.

That is, if the number of rB (blue) or rY (yellow) pixels among the RGB pixels in the slide window in which the lane is detected is absent or is close to 0, the lane is determined as white. If more than one pixel of rB (blue) or rY (yellow) is detected, the number of pixels of rB (blue) or rY (yellow) is compared to judge a large number of pixels as lane colors. The types of dotted lines, solid lines, center lines, and dedicated lanes can be finally identified in the color and shape of the lane.

Next, the road type is largely divided into a straight road and a curve road. The curvature value is calculated based on the following equation using the center coordinates of the slide detected in FIG. 7.

In this case, it is desirable to apply the RANSAC algorithm to improve the accuracy of the curvature of the lane by filling the margins between pixels adjacent to each other.

In this case, the lane determining unit 40 of the present invention detects the width of the lane on the center coordinates of the slide with respect to the left lane and the center coordinates of the slide with respect to the right lane and detects the position of the vehicle in the lane according to the detected width value do.

The system according to the present invention may further include a lane verifying unit 50 for filtering an error of the lane identified by the lane determining unit 40 as shown in FIG. The lane verifying unit 50 loads the identification information accumulated in the color and shape of the lane derived by the lane determining unit 40 as shown in FIG. 8, and compares the color and shape of the currently derived lane to filter the error.

If the camera 10 generates an image of 60 frames per second, 600 frames are generated for 10 seconds, and information identifying the color and shape of the lane is accumulated for each frame. If the filtering time for finally deriving the lane is set to 1 second, the cumulative information of 60 frames is combined to filter the errors of the identified lane to induce the final judgment.

For example, if the number of lanes identified by white and solid lines among the accumulated 60 frames for one second is 45 frames, and the lanes identified in the remaining 15 frames are white and dotted, the lanes identified in 15 frames are treated as errors, Of the lane.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit and scope of the invention as defined by the appended claims. It is therefore intended that such variations and modifications fall within the scope of the appended claims.

10: camera 20: image information section
30: lane detecting unit 40: lane determining unit
50: lane verification unit

Claims (10)

A system for identifying a lane in a road, comprising:
A camera for photographing a road from a driving vehicle to generate an image;
The image is copied to two or more images in order to simplify and delete unnecessary pixels of the image, and the duplicated color image is converted into a gray image, and then the horizontal direction sole edge is found and the intensity of the edge gradient is set as a threshold value The binarized image is converted into the HLS color image, and the S channel is set to the threshold value to fuse the achromatic image with the chroma removed, and the binarized image and the achromatic image are merged into one, An image correction unit for converting a region of interest into a back projection to process the image as a bird's-eye image with the perspective removed;
A lane detection unit for tracking a pixel to be compared with an average pixel in the image to obtain a cumulative histogram in a vertical direction and detecting a lane at a maximum peak position of the histogram around the boundary of the region of interest;
The lane type is determined by the color of the pixels forming the lane of the image, and the RANSAC algorithm is applied to the pixels forming the lane of the image to calculate the curvature corresponding to the neighboring pixels between the pixels A lane determining unit for determining the type of road; And
And a lane verifying unit for loading the information of the color and the shape of the lane derived by the lane determining unit and for comparing the color and shape of the lane that is currently obtained to filter the error to thereby induce a final judgment Autonomous lane identification system. delete delete delete delete delete delete delete delete delete

Images