CN102303609A - Lane departure warning system and method - Google Patents

Abstract

The invention relates to a lane departure warning system and a lane departure warning method, wherein the system comprises an electrically connected camera unit for receiving and transmitting road condition signals; the lane line identification unit is connected with the camera unit, receives the road condition signal and performs Hough conversion on the road condition signal so as to identify a lane line; the deviation detection unit is connected with the lane line identification unit, calculates the distance between the lane line identification unit and the vehicle body according to the identified lane line, and sends out an early warning instruction when the distance exceeds a preset threshold value; and the early warning unit is connected with the offset detection unit, receives the early warning instruction and sends out a corresponding early warning signal. The method comprises the following steps: and calibrating the camera unit, setting an effective area, carrying out edge detection, searching lane lines, judging whether the lane lines deviate or not and sending an alarm signal.

Description

Lane departure warning system and method

【Technical field】

The invention relates to a lane departure warning system.

【technical background】

At present, lane departure accidents and car rear-end collisions caused by drivers' negligence, fatigue, drowsiness or lack of concentration account for about 60%-70% of the entire road traffic accidents, and about 70% of the fatalities. According to the estimates of the U.S. Federal Highway Administration, the use of a system that can detect the lateral position of a vehicle and a system that can detect the distance from a vehicle in front of a vehicle will greatly reduce the occurrence of these traffic accidents. Therefore, lane departure warning and front vehicle prevention The research of collision system has aroused great attention of various countries.

There are mainly the following methods for machine vision-based lane departure systems at home and abroad: feature-based recognition methods, model-based recognition methods, and methods based on fusion of vision and other sensors. The feature-based recognition method uses the gray-scale features of the road boundary and lane marking lines to complete the recognition of the road boundary and lane marking lines from the sequential grayscale images in front of the vehicle; the model-based recognition method is mainly based on different The road image model uses different recognition technologies to recognize road boundaries and lane markings; the method based on fusion of vision and other sensors fuses road information sensed by multiple sensors, and uses multiple image features to recognize roads. At present, some countries in the world have successfully developed some distinctive lane departure warning systems, such as: Mobileye_AWS system, AutoVue system, GOLD system, NavLab system, etc. These systems give the driver warning information in advance to remind the driver to take correct operation measures to prevent such accidents or reduce the degree of injury of such accidents.

However, due to the complex road environment and various types of vehicles, the lane departure warning and front collision avoidance warning systems must not only have good real-time performance, but also have strong robustness. However, the existing system cannot satisfy these requirements well. The present invention proposes a system with better real-time performance and robustness.

【Content of invention】

In view of the above problems, the present invention proposes a lane departure warning system, which is mainly used to prevent lane departure of vehicles in motion, and the system includes electrically connected

The camera unit is used to receive and transmit road condition image signals;

A lane line identification unit, connected to the camera unit, receives the road condition image signal, and performs canny edge detection and Hough transform on the signal to identify the lane line;

An offset detection unit, connected to the lane line recognition unit, calculates the distance between the lane line and the vehicle body according to the identified lane line, and sends an early warning command when the distance exceeds a preset threshold;

The early warning unit is connected with the offset detection unit, receives the early warning instruction, and sends out a corresponding early warning signal.

The present invention simultaneously proposes a method for realizing lane departure warning, said method comprising the following steps:

Step 1 calibrates the camera unit to receive and transmit road condition image signals;

Step 2: Set the effective area for detecting lane lines according to the driving state of the vehicle. The area of the effective area is larger than the area of the actual lane line in the image; the effective area range is set according to whether the lane line has been detected at this time: if the detection When the lane line is reached, the effective area is determined according to the camera unit calibrated in step 1 and the empirical value of the lane size of the Chinese road. This area only needs to contain the detected lane line; if no lane line is detected, or In the case of vehicle line pressing, the effective area at this time is the entire image. After the effective area is set, most of the invalid areas can be removed to improve the real-time performance of the algorithm;

Step 3: Carry out canny edge detection on the image signal in the effective area to obtain a binarized image;

Step 4 uses the binarized image to perform Hough transform to find and match lane lines;

Step 5 Calculate the distance between the vehicle body and the lane line according to the detected lane line, and determine whether to issue an early warning instruction;

Step 6: Send out an early warning signal according to the early warning instruction.

Said step 4 further comprises the following steps:

Step 41 detects the current driving state of the vehicle;

Step 42 searches for lane line parameters that meet certain constraints;

Step 43 determines the lane line according to the lane line parameters;

Wherein the constraints described in step 42 are: (thetaTh1<Theta<thetaTh2)&&(abs(Theta-PreTheta)<DeltaThetaTh)&&(abs(Distance-PreDistance)<DeltaDisTh),

Among them: Theta and Rho are Hough transformation, the current lane line corresponds to the straight line parameters in Hough space, thetaTh1 and thetaTh2 are the range of the lane line angle Theta, PreTheta is the effective lane line angle in the previous frame; DeltaThetaTh is continuous The change threshold of the angle between the lane lines in the middle of two frames; Distance and PreDistance are the X coordinate values of the current lane line and the lane line of the previous frame on the bottom horizontal line respectively, and DeltaDisTh is the change threshold of the distance between the two lane lines in the middle of the two frames before and after.

When the vehicle is running normally, when calculating the lane line parameters on one side, thetaTh1 can be a positive integer between 30-60, and thetaTh2 is 90; when calculating the lane line parameters on the other side, the thetaTh1 is -60 --Negative integer between 30, thetaTh2 is -90. When the vehicle deviates, when calculating the lane line parameters on one side, thetaTh1 can be a positive integer between 10-30, and thetaTh2 can be a positive integer between 60-80; when calculating the lane line parameters on the other side , thetaTh1 is a negative integer between -30-10, and thetaTh2 is -80-60.

When the vehicle deviates, middle lane line detection is performed, and the range of Theta is between 70-90.

The present invention designs a unique lane line recognition unit for lane recognition after Hough space transformation, and effectively realizes a system with better real-time performance and robustness.

【Description of drawings】

Fig. 1 is the connection schematic diagram of lane departure warning system of the present invention;

Fig. 2 is a schematic diagram of the effective areas of two lane lines of the present invention;

Fig. 3 is a schematic diagram of the effective areas of three lane lines of the present invention;

Fig. 4 is the early warning flowchart of the present invention;

Fig. 5 is a system connection diagram of another embodiment of the present invention;

Fig. 6 is the hardware connection circuit diagram of invention;

The main marks of the accompanying drawings are:

Camera unit 1 , lane line recognition unit 2 , deviation detection unit 3 , warning unit 4 , display unit 5 , control unit 6 , storage unit 7 , system power supply 8 , effective area 9 , and lane lines 10 .

【Detailed ways】

Referring to Figures 1-3, the present invention proposes a lane departure warning system, which is mainly used to prevent lane departure of a driving vehicle. The system includes an electrically connected camera unit 1, a lane line recognition unit 2, and a deviation detection unit 3 and early warning unit 4.

The camera unit 1 can be placed at the front of the car body, such as a car windshield or a car dashboard, and is used to record images of road conditions in front of the car body. The camera unit 1 may include a camera 11 and a video conversion chip 12 .

The lane line identification unit 2 is connected with the camera unit 1 , receives the road condition signal, and performs Hough transform on the signal to identify the lane line 10 . In order to improve the speed of signal processing, when detecting the lane line 10, it can be assumed that the lane line 10 is a straight line model, making full use of the characteristics of the width of the lane line, the gray value of the lane line is larger than the gray value of the road surface, etc., effectively improving the robustness and real-time performance of the system .

The offset detection unit 3 is connected to the lane line recognition unit 2, and judges whether a lane departure has occurred based on the distance between the middle position of the vehicle body and the lane line 10 according to the identified lane line 10. When the distance exceeds When the threshold is preset, an early warning command is issued.

The early warning unit 4 is connected with the offset detection unit 3, receives the early warning instruction, and sends out a corresponding early warning signal. The warning unit 4 may include an audio processing circuit 41 and a speaker 42 connected thereto.

The present invention also proposes a method for realizing lane departure warning, said method comprising the following steps:

Step 1 calibrates the camera unit 1 to receive and transmit road condition signals. The purpose of camera unit 1 calibration is mainly to obtain the internal and external parameters of the camera, and use these parameters to calculate the distance between the vehicle body and the left and right lane lines, the distance between the vehicle body and the vehicle in front, and the real height and width of the vehicle. The method of camera calibration is relatively mature, and there are a large number of documents that explain it in depth, so I will not introduce it in detail.

Step 2 is to set the effective area 9 for detecting the lane line 10 according to the driving state of the vehicle. If the vehicle is running normally at this time, and the detection of the lane line 10 is stable at this time (the lane line 10 is detected in several consecutive frames), then the effective area 9 of the lane line 10 is set to the top left and right width of the lane line 10 detected in the previous frame 10 pixels each, and a trapezoidal region with a left and right width of 15 pixels at the bottom, such as the region 9 marked in FIG. 2 . If we think that the lane line 10 is approximately a straight line, then the lane line 10 can be expressed as:

y=k*x+b

When the upper and lower ordinates of the region of interest are known, the four vertices of the trapezoidal region can be calculated to determine the effective region. If the lane marking 10 is in a deviation state, and the detection of the lane marking 10 is stable at this time, the effective area 9 of the lane marking 10 is a trapezoidal area respectively determined by the three lane markings, as shown in FIG. 3 , and the setting method is the same as above. If the detection of the lane marking 10 is unstable, the effective area 9 of the lane marking 10 is set to be the whole picture.

Without special instructions, the following operations on the lane line 10 will be performed in this effective area.

Step 3: Carry out conditional canny edge detection in the valid area 9 to obtain a binarized image. Lane markings 10 have the following inherent characteristics:

i. The main color of lane marking lines is white or yellow;

ii. Although due to the influence of perspective transformation, the width of the lane line 10 in the image is different, the lane line 10 still has a certain width in the image;

iii. The gray value of the lane line 10 is greater than the gray value of the road surface;

iv. The lane line 10 is composed of continuous or discontinuous lane markings, and it can be assumed that the lane line 10 is continuous to a certain extent;

v. The lane line 10 can be approximately described by a straight line: in general, the curvature of road curves is small, and the lane line 10 can be treated as a straight line within a certain range in front of the vehicle.

In lane line detection, the inherent characteristics of the above-mentioned lane line 10 can be used to remove points that do not belong to the edge of the lane line 10, which can not only improve the accuracy of lane line recognition, but also greatly improve the real-time performance of the system.

In edge detection, the constraints are:

(grad_right2left[x][y]>th1)&&(grad_left2right[x][y]<th2)&&(gray_right_average[x][y]>th3),

Where grad_right2left[x][y] is the difference between the gray value on the right side and the gray value on the left side of the currently detected frame image [x][y], and its calculation formula is:

grad_right2left[x][y]=gray[x+3][y]+gray[x+4][y]-gray[x-1][y]-gray[x-2][y],

The gray[x][y] is the gray value at the image [x][y]; th1 is the threshold of the difference between the gray value on the right side of the lane line 10 in the image and the gray value on the left side, and the value range of the threshold It can be a non-negative integer of 30-50; grad_left2right[x][y] is the difference between the gray value at a certain distance on the right side of the currently detected frame image [x][y] and the gray value on the left side, and its calculation formula is:

grad_left2right[x][y]=gray[x+L+1][y]+gray[x+L+2][y]-gray[x+1][y]-gray[x+2][y ],

The L is the width of the lane line 10; th2 is the threshold value of the difference between the gray value and the left gray value after a certain distance on the right side of the lane line 10 in the image, and the value range of the threshold can be a negative number of -10-0 ;gray_right_average[x][y] is the average gray value of pixels in a certain neighborhood on the right side of the currently detected frame image [x][y], and its calculation formula is:

gray_right_average[x][y]=(gray[x+1][y]+gray[x+2][y]+gray[x+3][y])/3,

Among them, th3 is the gray threshold value of the road surface. Since the gray value of the road surface is not the same in the whole image, the value of th3 can be given in the way of line scanning. The condition grad_right2left[x][y]>th1 uses the prior knowledge that the gray value of the lane line is greater than the gray value of the road surface; grad_left2right[x][y]<th2 uses the lane line 10 with a certain width and the gray level of the lane line The value is greater than the two prior knowledge of the gray value of the road surface.

In the case of meeting the above constraints, edge enhancement and edge binarization are performed according to the canny algorithm. Since the edge detected by canny is a single-pixel wide edge, the real-time and robustness of the program operation is improved.

Step 4 uses the binarized image to perform Hough transform to find and match the lane line 10;

Due to the fast speed of the vehicle, the position of the lane line 10 will not change greatly in the preceding and following frames. Using this feature, we can match the detected lane lines 10 in the preceding and following frames. Only a few consecutive frames When all match, we think that the lane line 10 detected at this time is indeed a lane line. After the Hough transformation, the steps to find the lane line 10 in the Hough space are:

Step 41. Detect the driving state of the vehicle at this time, including normal driving state and lane departure state.

Step 42, after knowing the driving state of the vehicle, search for the lane line 10. If it is a normal driving state at this time, only the left and right lane lines 10 of the running vehicle are searched; Press 10 on that lane line. Taking the lane line 10 on the left side of the driving vehicle as an example, the method of finding the lane line 10 is described in detail below: the left lane line parameter is the average value of Theta and Rho corresponding to the maximum value and the second maximum value in the Hough space satisfying certain constraints , the constraints are:

(thetaTh1<Theta<thetaTh2)&&(abs(Theta-PreTheta)<DeltaThetaTh)&&(abs(Distance-PreDistance)<DeltaDisTh),

Among them (Theta, Rho) is after the Hough transformation, the current lane line 10 corresponds to the straight line parameters in the Hough space; thetaTh1 and thetaTh2 are the range of the angle Theta between the left lane line, if the vehicle is in a normal driving state at this time, you can choose thetaTh1 as A positive integer around 45, preferably a positive integer between 30-60, thetaTh2 is 90, which is the maximum angle of the left lane line, if the vehicle deviates at this time, you can choose thetaTh1 to be a positive integer around 20, preferably 10 A positive integer between -30, thetaTh2 is a positive integer around 70, preferably a positive integer between 60-80; PreTheta is the angle between the left effective lane line 10 detected in the previous frame; DeltaThetaTh is two consecutive frames The change threshold of the angle between the middle lane line, DeltaThetaTh can take a positive integer between 5-8; Distance and PreDistance are the X coordinate values of the lane line 10 on the bottom horizontal line in this and the previous frame respectively, and DeltaDisTh is the two frames before and after The distance change threshold of the middle two lane lines 10, DeltaDisTh can take a positive integer between 10-15.

The detection method of the right lane marking 10 and the middle lane marking 10 is basically the same as that of the left lane marking 10, only the range of the angle between the lane markings is different. The angle constraint of the right lane line 10 can be as follows: thetaTh1 is about -45 during normal driving, and thetaTh2 is -90. When deviation occurs, thetaTh1 can be selected as -20, and thetaTh2 can be selected as -70; It is only detected when lane departure occurs, and the range of Theta satisfies that abs(Theta) is between 70 and 90.

Step 43. After detecting two straight lines that meet the conditions, if the distance between the two straight lines is not much different, calculate the average value of the parameters corresponding to the two straight lines, which is the parameter of the lane line 10 at this time; if the two straight lines The distance of the distance is quite different, and the line with the most pixels detected when Hough changes is taken as the lane line. Inside the Hough algorithm, the number of pixels on all lines detected is known. The detailed description of the Hough algorithm is as follows:

When there is a point in the XY coordinate system, there will be countless straight lines passing through this point, and each straight line can be represented by a unique set of polar coordinate system coordinates (ρ, θ) values. Therefore, a point in the XY coordinate system corresponds to a curve in the polar coordinate system. Then countless points on a straight line on the XY plane will correspond to countless curves on the polar coordinate plane, and the coordinates (ρ, θ) corresponding to the intersection points of these curves can be used to represent the straight line.

After the lane line is detected by canny edge, many edge points can be obtained, and these edge points are converted into a curve in Hough space by Hough transform. Create an accumulator array with different values of coordinate parameters Theta and Rho in Hough space. When the curve passes through the corresponding coordinate value, the corresponding coordinate value will add 1 to the accumulator value. In this way, after Hough transformation is performed on all edge points, when the value of the accumulator corresponding to the Hough space coordinate is greater than the set parameter, it is considered that the Theta and Rho values of the coordinate value correspond to a lane line.

Step 5 Calculate the distance between the vehicle body and the lane line 10 based on the detected lane line 10 , and judge whether lane departure occurs.

If the driving state of the vehicle is normal at this time, it is detected whether lane departure occurs at this time, and whether the degree of lane departure occurs can be determined by the following formula:

LeftDis<SafeDisanceTh or RightDis<SafeDisanceTh

Among them, LeftDis is the distance between the middle position of the car body and the left lane line 10; RightDis is the distance between the middle position of the car body and the right lane line 10; SafeDisanceTh is the warning distance threshold between the car body and the lane line 10, when the car body When the distance between the middle position and any lane line 10 is less than the warning distance threshold, it is considered that lane departure has occurred, and a corresponding warning command is issued.

Step 6: The early warning unit sends a corresponding early warning signal according to the alarm instruction, and the early warning signal can be an audible and visual signal.

According to Figure 4, the early warning process can be briefly described as follows: when the lane departure warning system is activated, the camera unit 1 needs to be calibrated in advance; then the effective area 9 for detecting the lane line 10 is set, and edge detection is performed in the effective area 9 to find out the lane Line 10: After finding out the lane line 10, compare the position of the vehicle body with the lane line 10 to determine whether the vehicle body deviates, and if the result is deviation, an early warning is given.

As shown in FIG. 5 , the system further includes a display unit 5 , which is connected to the deviation detection unit and displays one or more of the speed, distance and road condition signals of the vehicle body on the LCD display.

As shown in Figure 6, the system further includes a control unit 6 and a storage unit 7, the control unit 6 is connected with the camera unit 1 and the early warning unit 4, and includes the lane line recognition unit 2 and the deviation detection unit 3. It is used to control and schedule the overall operation of the system. It can be a high-performance DSP chip, which is used to control and schedule the overall operation of the system circuit, identify the lane line, and calculate the distance between the car body and the lane line according to the information such as the lane line parameters and position and the position information of the car body. If it is less than a certain set threshold, a lane departure warning command is output.

The storage unit 7 is used to store system operating programs, data and preset parameters, and is connected to and controlled by the control unit 6 . It can be stored in FLASH72 or memory 71.

The system power supply 8 provides power to each unit through the control unit 6 .

The present invention designs a unique lane line recognition unit for lane recognition after Hough space transformation, and effectively realizes a system with better real-time performance and robustness.

Claims (10)

1. A lane departure warning system, characterized in that: the system includes an electrically connected camera unit for receiving and transmitting image signals of road conditions; A lane line identification unit, connected to the camera unit, receives the image signal, and performs canny edge detection and Hough transform on the image signal to identify the lane line; An offset detection unit, connected to the lane line recognition unit, calculates the distance between the lane line and the vehicle body according to the identified lane line, and sends an early warning command when the distance exceeds a preset threshold; The early warning unit is connected with the offset detection unit, receives the early warning instruction, and sends out a corresponding early warning signal. 2. The lane departure warning system according to claim 1, characterized in that: the system further comprises a display unit, which is connected with the deviation detection unit, and displays the speed, distance and road condition image signal of the vehicle body One or more are displayed on the LCD display. 3. The lane departure warning system according to claim 1, wherein said system further comprises: a control unit, connected to the camera unit and the early warning unit, and including the lane line recognition unit and the deviation detection unit, for controlling and scheduling the overall operation of the system; The storage unit is connected to and controlled by the control unit, and is used for storing system operation programs, data and preset parameters. 4. A method for realizing lane departure warning, characterized in that: said method comprises the following steps: Step 1 calibrates the camera unit to receive and transmit road condition image signals; Step 2 is to set the effective area for detecting the lane line according to the vehicle driving state, the area of the effective area is greater than the area of the actual lane line in the image; Step 3: Carry out canny edge detection on the image signal in the effective area to obtain a binarized image; Step 4 uses the binarized image to perform Hough transform to find and match lane lines; Step 5 Calculate the distance between the vehicle body and the lane line according to the detected lane line, and determine whether to issue an early warning instruction; Step 6: Send out an early warning signal according to the early warning instruction. 5. the method for realizing lane departure warning as claimed in claim 4 is characterized in that: when described step 3 carries out canny edge detection, constraint condition is: (grad_right2left[x][y]>th1)&&(grad_left2right[x][y]<th2)&&(gray_right_average[x][y]>th3), Among them, [x][y] is the two-dimensional array index in the computer, and the coordinate system where [x][y] is located is: the upper left corner of the image is the origin, the horizontal direction to the right is the positive direction of the x-axis, and the horizontal direction is the y-axis In the positive direction, grad_right2left[x][y] is the difference between the gray value on the right side and the gray value on the left side of the image [x][y] of the current detection frame, and its calculation formula is: grad_right2left[x][y]=gray[x+3][y]+gray[x+4][y]-gray[x-1][y]-gray[x-2][y], The gray[x][y] is the gray value at the current detection frame image [x][y]; gray[x+3][y] is the distance from the current detection frame image [x] in the positive direction of X The pixel gray value at the 3 pixel position at [y], gray[x+4][y] is the pixel gray at the 4 pixel position from the current detection frame image [x][y] in the positive direction of X gray[x-1][y] is the pixel gray value at a pixel position from the current detection frame image [x][y] in the negative direction of X, gray[x-2][y ] is the pixel gray value at two pixel positions away from the current detection frame image [x][y] in the negative direction; th1 is the gray value on the right side of the current detection frame image [x][y] and The minimum value that must be satisfied by the difference of the left gray value of the previous detection frame image [x][y]; grad_left2right[x][y] is the gray value at a certain distance on the right side of the current detection frame image [x][y] The difference between the value and the gray value at a certain distance to the left, its calculation formula is: grad_left2right[x][y]=gray[x+L+1][y]+gray[x+L+2][y]-gray[x+1][y]-gray[x+2][y ], gray[x+L+1][y] is the pixel gray value at the position L+1 pixel width away from the current detection frame image [x][y] in the positive direction of X, gray[x+L+2 ][y] is the pixel gray value at the position of L+2 pixel width away from the current detection frame image [x][y] in the positive direction of X, and gray[x+1][y] is the distance in the positive direction of X The pixel gray value at the position of 1 pixel width at the current detection frame image [x][y], gray[x+2][y] is the distance from the current detection frame image [x][y] in the positive direction of X ] at the pixel gray value at the position of 2 pixel widths, L is the width of the assumed lane line in the effective area); th2 is the gray value of the distance greater than L in the positive direction of the current detection frame image [x][y]X The value must meet the maximum value of the difference between the gray value of the adjacent pixel in the positive direction of the previous detection frame image [x][y]X; gray_right_average[x][y] is the image [x][y]X of the current detection frame The average gray value within the range of four adjacent pixels in the positive direction, its calculation formula is: gray_right_average[x][y]=(gray[x+1][y]+gray[x+2][y]+gray[x+3][y])/3, where th3 is the road gray threshold . 6. The method for realizing lane departure warning as claimed in claim 4, characterized in that: said step 4 further comprises the following steps: Step 41 detects the current driving state of the vehicle; Step 42: After determining the driving state of the vehicle, search for lane lines that meet certain constraints; After step 43 detects the lane line that meets the conditions, determine the lane line; Wherein the constraint condition described in step 42 is: (thetaTh1<Theta<thetaTh2)&&(abs(Theta-PreTheta)<DeltaThetaTh)&&(abs(Distance-PreDistance)<DeltaDisTh), wherein: Theta and Rho are after Hough transformation , the current lane line corresponds to the straight line parameters in Hough space, thetaTh1 and thetaTh2 are the range of the lane line angle Theta, PreTheta is the lane line angle in the previous frame; DeltaThetaTh is the change threshold of the lane line angle in the middle of two consecutive frames ;Distance and PreDistance are the X coordinate values of the current lane line and the lane line of the previous frame on the bottom horizontal line, respectively, and DeltaDisTh is the change threshold of the distance between the two lane lines in the middle of the two frames before and after. 7. The method for realizing lane departure warning according to claim 6, characterized in that: when the vehicle is running normally, said thetaTh1 is a positive integer between 30-60, and thetaTh2 is 90; or said thetaTh1 is -60- Negative integer between -30, thetaTh2 is -90. 8. The method for realizing lane departure warning as claimed in claim 6, characterized in that: when the vehicle deviates, said thetaTh1 is a positive integer between 10-30, and thetaTh2 is a positive integer between 60-80; or The thetaTh1 is a negative integer between -30-10, and thetaTh2 is -80-60. 9. The method for realizing lane departure warning according to claim 6, characterized in that: said DeltaThetaTh is a positive integer between 5-8. The DeltaDisTh is a positive integer between 10-15. 10. The method for realizing lane departure warning as claimed in claim 4, characterized in that: when the vehicle deviates, middle lane line detection is performed, and the range of Theta after Hough transformation is between 70-90.

Images