KR100773870B1 - Method for recognizing traffic lines by using two cameras - Google Patents

Abstract

A traffic lane recognizing method using two cameras is provided to display clearly traffic lanes of a long range and to show the lane of a short range through a wide view angle by photographing the road with two cameras and combining images obtained by the cameras, as one. A method for recognizing traffic lanes from images obtained by photographing the road with two cameras comprises the steps of: positioning two cameras at the upper and lower sides in the center of a vehicle and photographing the image of the road with dividing the image according to short and long ranges, by making the focal distances of the cameras different(S410); combining the divided road images as one(S420); converting the combined image to be digitally processed(S430); filtering the image by a digital filter to divide clearly a boundary between an object and the background on the digital image(S440); calculating a lane specific point from geometrical information of the filtered image(S450); and recognizing the lane from the calculated lane specific point(S460).

Description

Lane recognition method using two cameras {Method for recognizing traffic lines by using two cameras}

1 is a view showing a lane recognition range of the lane recognition method using a single camera according to the prior art.

2 is a view showing a lane recognition range of the lane recognition method using two cameras according to a first embodiment of the present invention.

Figure 3 is a block diagram showing the configuration of a lane recognition device to which the lane recognition method according to the present invention is applied.

4 is a flowchart illustrating a lane recognition method using two cameras according to a first embodiment of the present invention.

FIG. 5 is a diagram illustrating a process of processing an image through the lane recognition method illustrated in FIG. 4.

6 is a view showing a lane recognition range of the lane recognition method using two cameras according to a second embodiment of the present invention.

<Description of the symbols for the main parts of the section>

120: image collecting unit 130: digital image conversion unit

140: image filter unit 150: lane recognition unit

160: display unit 170: central processing unit

100, 100a, 100b, 200a, 200b: camera

The present invention relates to a lane recognition method using two cameras, and more particularly, to capture a video of a driving road through two cameras provided in a vehicle, and to recognize the lane by collecting them as an image of one driving road. Therefore, the present invention relates to a lane recognition method that can further expand the angle of view of the camera to reduce blind spots and recognize lanes more clearly.

In general, when a moving object moves by tracking a line, in order to detect an out of line and alert the outside, an image recognition device or the like should first be used to recognize the line.

That is, the recognition apparatus using an image acquires an image of a target to be recognized by a moving object using a camera, and then extracts a feature of the target by using a digital image processing technology, and identifies the target by using the extracted feature. In order for the image recognition apparatus to perform its original function smoothly, it is necessary to be able to accurately extract the target line from the moving object.

Such image recognition device can be widely applied in the fields of automobile, robot technology (RT), automated guided vehicle (AGV), and the like, especially in the automobile field, it is necessary to ensure accurate lane recognition even at high speeds. Difficulty is high. Here, the lane is a reference line for driving, and is a standard for all driving behaviors such as forward, reverse, lane change, course change, forward parking, reverse parking, and row parking.

Advanced Safety Vehicles (ASVs), which are currently being highly functionalized and intelligent, are the main targets for lane recognition methods, lane departure warning devices to prevent drowsy driving, rear parking guides and rear parking guides to help beginners park. Applied to a lane keeping device that maintains a lane by applying a small torque to the handle in a dangerous situation, the range of application is gradually expanded.

Looking at the lane recognition method according to the prior art that acquires the driving road as an image as described above, and distinguishes and recognizes the lane from the obtained image, most of the methods to obtain the image of the driving road using one camera, It distinguishes the lanes from the image and recognizes the road. Using two cameras, the distance between the cars rather than the lanes is obtained by using the two cameras as a stereo camera, which is a kind of three-dimensional camera. It has the feature of recognizing it.

In the case of recognizing a driving road as an image using a camera, the angle of view, which is the sharpness according to the perspective of the object recognized as the image and the viewing angle of the image recognized by the camera, is determined according to the focal length of the lens provided in the camera. . In other words, if the focal length of the camera is short depending on the lens provided in the camera, the field of view of the image taken by the camera, that is, the angle of view, becomes wider, so that the near-lane lane can be widely recognized, while the remote object is not clearly recognized. On the contrary, when the focal length of the camera is long, the far point is clearly seen in the image of the driving road, while the angle of view of the image recognized by the camera is reduced, which makes it impossible to recognize all near lanes. have.

Therefore, when recognizing a lane as a straight line, it is recommended to use a lens with a relatively short focal length of the camera, and to use a camera having a relatively focal length when recognizing the lane as a curve to calculate the radius of curvature of the lane. It is advantageous.

To compare the characteristics of the camera focal length and the lane recognition method according to the prior art, in order to capture the image of the driving road with one camera, and to grasp the characteristics of the lane on the recorded driving road, the distance of the driving road The area must be covered, in which case a long focal length camera is used. However, when photographing a driving road with a long focal length camera, there is a problem in that a blind spot that is not captured by the camera is generated because the angle of view of the image photographed by the camera is narrowed.

In addition, if the image of the driving road is taken by using a camera having a short focal length to reduce the blind area, the blind area of the lane is reduced but the remote area of the driving road is not clearly captured, making it difficult to distinguish the characteristics of the lane. There is also a problem.

The present invention is to solve the above problems. That is, by taking images of the driving road using two cameras provided in the vehicle and collecting the images obtained from each camera into one image, the lanes of the distance are clearly displayed and the near lanes are displayed with a wide angle of view. The purpose is to recognize the lane more accurately by acquiring the image of the driving road.

In order to achieve the above object, the present invention as a technical concept, in the method for recognizing the lanes from the image obtained by photographing the image of the driving road with two cameras, the image of the driving road by the camera in the near and far Photographing by dividing the image into images; collecting images of the divided roads into one image; converting the collected image into digital processing; and converting the collected image into a digital image; An image filtering step of performing image processing through a digital filter so that the boundary between the object and the background becomes clear; calculating lane feature points from geometric information of the filtered image; and lanes from the calculated lane feature points Two cameras comprising the step of recognizing Provided by the lane recognition method.

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

In general, two types of cameras for image recognition are used, a CCD camera and a CMOS camera. Briefly, the advantages and disadvantages of the two types of cameras are that CCD cameras, which accumulate and transfer charge generated by light energy and recognize images, have superior image quality and high sensitivity compared to CMOS cameras. Expensive and complicated peripheral circuits cause severe heat generation. CMOS cameras that read images generated by light energy with semiconductor switches generate more noise and have lower sensitivity than CCD cameras. It is possible to make one chip with peripheral IC because of high degree, and it has low power consumption and low price.

Therefore, in the present embodiment, the image of the driving road is photographed using a CMOS camera.

1 is a view showing a lane recognition range of the lane recognition method using a single camera according to the prior art.

Referring to FIG. 1, the lane recognition method using one camera according to the related art includes one camera 100 at a center portion of a vehicle to photograph a driving road in front of the vehicle.

The image of the driving road photographed by the camera 100 is divided into a photographing area R through the camera 100 and a blind area which is a non-capturing area outside the photographing area R. Here, the blind spot means a part that cannot be recognized by the camera.

In addition, the lane of the driving road is divided into a recognition lane C included in the photographing area R and an unrecognized lane U included in the blind area.

Whether to recognize the lane through the camera 100 as described above is determined according to the focal length of the lens provided in the camera 100.

In other words, when the focal length of the lens used in the camera 100 is long, the angle of view that can be recognized by the camera 100 is narrowed, so that objects and backgrounds far from the vehicle are not widely recognized and are not widely recognized on both sides of the angle of view. Although blind spots are generated, objects and backgrounds far from the vehicle can be captured clearly. On the contrary, when the focal length of the lens used in the camera 100 is short, the angle of view is widened to recognize the object and the background in a short distance, but the object and the background in the far distance are blurred. Therefore, there is a problem in that it is difficult to capture a near distance at a wide angle of view and to obtain an image having a clear distance.

2 is a diagram illustrating a lane recognition range of a lane recognition method using two cameras according to a first embodiment of the present invention.

FIG. 2A is a view showing a state in which two cameras 100a and 100b for recognizing the front of the vehicle are positioned on both sides of the vehicle, respectively, and photographed by dividing the driving road. FIG. FIG. 2A shows the images 110a and 110b collected together.

In each of the cameras 100a and 100b shown in FIG. 2A, the focal length is set longer than the focal length of the camera 100 shown in FIG. 1, so that the focal length is longer than that of the one camera 100 shown in FIG. 1. Although the angle of view is narrow, the focal length is determined to clearly photograph a far object and a background.

As described above, the divided images obtained by taking a panorama method from each camera are combined into one, as shown in the image of FIG. 2 (b), to perform a lane recognition process on the combined images.

In this case, the panorama method is a method of taking a long sideward or upward direction than a normal picture to capture a lot of scenery in a picture that contains one screen. The height of the boundary of the periphery as indicated in the

The two cameras 100a and 100b which photograph a driving road by dividing a predetermined shooting area R have the same focal length, and thus combine the driving roads, lanes and backgrounds that appear in each divided image into one image. This is possible.

As described above, referring to the image of the driving road shown in FIG. 2B, the angle of view is wider than the image of the driving road obtained by using one camera 100 shown in FIG. You can see that the area is reduced, and the distant lanes are shot more clearly.

3 is a block diagram showing the configuration of a lane recognition apparatus to which the lane recognition method according to the present invention is applied.

Referring to the configuration of the lane recognition device to which the lane recognition method according to the present invention is applied with reference to FIG. 3, the lane recognition device is provided in a vehicle and receives two cameras 100a and 100b to receive an image of a driving road. An image collector 120 that receives an image obtained by the cameras 100a and 100b and collects the image into a single image, and digitalizes the image collected by the image collector 120 to enable input and output of a computer. An image conversion unit 130, an image filter unit 140 which performs image processing so that a specific object and a boundary are clear in the digital image converted by the digital image conversion unit 130, and the geometric information of the filtered image. It is composed of a lane recognition unit 150 for calculating a lane feature to recognize the lane.

The image collecting unit 120 is a part which collects the images obtained by dividing images through the two cameras 100a and 100b into one image. That is, the image collecting unit 120 finds an object or a background that is common in the divided images respectively obtained by the two cameras 100a and 100b, and combines the object or the background into one image so that the object or the background completely overlaps.

In addition, the digital image converter 130 converts an analog image into a digital image. That is, the digital image converting unit 130 converts the analog image obtained by collecting the single image in the image collecting unit 120 into a digital image capable of numerical measurement and control of the image.

The image filter 140 filters the digital image obtained by the digital image converter 130 to clearly display the outlines of the driving road, the vehicle, the lane, and the buildings around the road, which are included in the image. It is the part that separates the boundary.

Meanwhile, the lane recognizing unit 150 is a part for identifying and recognizing lanes by calculating a geometrical characteristic of a specific object obtained by the image filter 140, that is, a lane distinguished from other objects and a background.

4 is a flowchart illustrating a lane recognition method using two cameras according to the first embodiment of the present invention, and FIG. 5 is a view illustrating a process of processing an image through the lane recognition method shown in FIG. 4.

Hereinafter, a lane recognition method using two cameras will be described with reference to FIGS. 4 and 5.

First, the two cameras 100a and 100b provided in the vehicle divide and photograph an image of a driving road seen in front of the vehicle (S410).

As described above, the image of the driving road divided by the two cameras 100a and 100b is collected into one image as shown in FIG. 5A through the image collecting unit 120 (S420).

Thereafter, the collected image is converted into a digital image as shown in FIG. 5B through the digital image converter 130 to enable digital processing.

Thereafter, the converted digital image is subjected to image processing through the digital filter so that the boundary between the object and the background becomes clear through the image filter unit 140. As such, the image processed through the digital filter is displayed as shown in (b) of FIG. 5 (S440).

The image processing performed by the image filter unit 140 is performed using a Sobel Edge Filter method. In other words, when the Sobel edge filter method is used, the boundaries of the object and the background are clearly distinguished from each other in the image, and each object and the background are represented by lines.

Thereafter, lane feature points (CP) are calculated from the geometric information of the filtered image.

That is, as shown in (c) of FIG. 5, an image obtained through the image filtering step is represented by a line in which both an object and a background expressed in the image are divided. The geometric singularity showing the characteristic is found and calculated as the lane characteristic point (CP).

Finally, the lane is recognized as shown in FIG. 5D from the lane feature point CP calculated as described above (S460).

6 is a diagram illustrating a lane recognition range of a lane recognition method using two cameras according to a second embodiment of the present invention.

Referring to FIG. 6, two cameras 200a and 200b are provided up and down in the center of the vehicle and different focal lengths are set differently from each other in the installation method of the first embodiment. Ra) is taken to shoot the driving road at a long distance, and the other one camera is configured to shoot a wide range of near-road driving road having a wide shooting area (Rb).

That is, the two cameras use a long focal length lens to clearly capture a long distance, but the camera 200a of a narrow shooting area Ra and a short focal length lens are used to make the long distance not clear but wide. The camera 200b has a photographing area Rb.

In the divided images captured by the two cameras 200a and 200b having different focal lengths as described above, the spots having the same width W are found, and the slopes of the lanes Ca and Cb appearing in the divided regions are equal. By overlapping the two divided regions into one image.

The present invention described above is not limited to the above-described embodiments and the accompanying drawings, and various substitutions, changes, and modifications are possible in the technical field of the present invention without departing from the technical spirit of the present invention. It will be clear to those of ordinary knowledge.

As described above, the lane recognition method using two cameras reduces the blind spot of the road image and recognizes the lane image more broadly and clearly than the conventional method of recognizing the lane using one camera. There is an effect that can be recognized more accurately.

Claims (2)

(delete) In a method for recognizing lanes separately from an image obtained by photographing an image of a driving road with two cameras, Positioning the two cameras vertically in the center of the vehicle, and dividing an image of a driving road into near and long distances by varying a focal length of the cameras; Collecting the divided images of the traveling road into one image; A digital image conversion step of converting the collected image to enable digital processing; An image filtering step of performing image processing through a digital filter so that a boundary between an object and a background becomes clear in the converted digital image; Calculating lane feature points from the geometric information of the filtered image; Recognizing a lane from the calculated lane feature point; Lane recognition method using two cameras, characterized in that comprises a.

Images