KR100355993B1 - An apparatus for warning lane deviation and a method for warning the same - Google Patents

Abstract

The present invention relates to a lane departure warning device and method for promptly tracking a lane and warning a lane departure to prevent an accident. The present invention relates to an inadvertent operation by monitoring a driver's driving status (lane departure). By detecting and alerting in advance, the driver can take appropriate action, thereby preventing accidents.

Description

An apparatus for warning lane deviation and a method for warning the same}

The present invention relates to a lane departure warning device and method, and more particularly to an apparatus and method for alerting according to the degree of departure by lane tracking.

If you drive longer, the driver chats or sleeps with the person next to you or the person in the back seat. These carelessness and other forms of carelessness cause an accident by leaving the lane slightly off or completely off.

As a result, many methods for monitoring the driver's driving status and informing the driver are suggested.

In particular, when a computer processes a road driving environment in real time and informs a driver, a processor having a large processing capacity is required. When moving at a high speed, such as a car, there is a time constraint that the environment must be quickly determined and processed.

Real-time image processing devices that need to process such fast images are essential for various applications such as mobile robots and unmanned vehicles, but processing a large amount of data in real time requires a very high-performance computer, which is necessary for visualizing mobile robots and unmanned vehicles. There are many challenges in manufacturing and designing the device.

Accordingly, the present invention has been proposed in view of the above-described conventional situation, and an object thereof is to provide a lane departure warning device that promptly tracks a lane and provides an alarm according to the lane departure degree to prevent an accident in advance.

It is another object of the present invention to provide a lane departure warning method for promptly tracking a lane and providing an alarm according to a lane departure degree to prevent an accident in advance.

In order to achieve the above object, a lane departure warning apparatus according to a preferred embodiment of the present invention comprises: image input means for inputting an image of a driving front;

Processing means for performing road modeling by extracting lane information from the image signal from the image input means and informing a lane departure of the vehicle; And

Image output means for displaying a lane image and a lane departure degree from said processing means,

The processing means comprises: preprocessing means for extracting edges of a predetermined size and a predetermined length or more from the video signal from the video input means; And extracting only an edge corresponding to the lane from the extracted edge, finding a valid lane from the extracted edge, calculating the width of the lane, and checking the lane departure of the vehicle.

The lane departure warning method according to the preferred embodiment of the present invention receives an image of a driving front, extracts a lane component for each frame, and uses the optimal lane component from the extracted lane component to present the current driving road for each frame. Performing a modeling on the first step; And

And a second process of determining a lane departure of the vehicle in the modeled road and alerting according to the result.

1 is a block diagram of a lane departure warning apparatus according to an embodiment of the present invention;

2 is a view showing an example of the installation position of the image input unit shown in FIG.

3 is a view for explaining a lane recognition method in an embodiment of the present invention;

4 is a view showing an example of the installation position of the image output unit shown in FIG.

5 is a flowchart illustrating a lane departure warning method according to an embodiment of the present invention;

6 is a view employed for explaining the lane width inspection operation according to the embodiment of the present invention;

7 is a diagram employed in describing an operation of removing an inappropriate slope and noise component according to an embodiment of the present invention;

8 is a diagram of an operation of removing an outer deflection edge according to an embodiment of the present invention;

9 to 12 are views employed for explaining the road modeling operation according to an embodiment of the present invention,

13 to 17 are graphs showing lane recognition rates by a lane recognition algorithm employed in an embodiment of the present invention.

<Description of Symbols for Major Parts of Drawings>

10: image input unit 12: amplification / filter unit

14: synchronization separator 16: A / D conversion unit

18: preprocessing unit 20: first memory unit

22: second memory unit 24: post-processing unit

26: memory unit 28: OSD generation unit

30: synchronization generator 32: D / A converter

34: filter / amplifier 36: video output unit

Hereinafter, with reference to the accompanying drawings will be described an embodiment of the present invention in more detail.

1 is a block diagram of a lane departure warning apparatus according to an embodiment of the present invention.

Reference numeral 10 is an image input unit which captures and inputs an image (ie, a lane on a road and an obstacle present on a road) in front of a vehicle that is driving.

The image input unit 10 may be configured as a small video camera. The small video camera sets the input pixel to 256 pixels horizontally and 256 pixels vertically, and sets the brightness of one pixel to 256 levels and displays it in 8 bits.

The image input unit 10 is fixedly installed near the rearview mirror installed in the vehicle interior as illustrated in FIG. 2, and is installed to face the road on which the vehicle is traveling.

Reference numeral 12 is an amplifier / filter unit for amplifying the video signal input by the video input unit 10 to a predetermined level and removing noise components included in the video signal.

Reference numeral 14 is used to extract a synchronization signal (horizontal synchronization signal, vertical synchronization signal, composite synchronization signal, field signal, etc.) in a standard NTSC system from an image signal output from the amplification / filter unit 12, that is, an analog video signal. Synchronous separation.

The synchronization signal extracted by the synchronization separator 14 is used as a synchronization reference, an interface synchronization between the preprocessor 18 and the post processor 24, and a memory change synchronization signal in the operation of the preprocessor 18 described later.

Reference numeral 16 is an A / D converter which quantizes the video signal output from the amplifier / filter unit 12, that is, the analog video signal, and converts the digital signal into digital data. The digital image data converted by the A / D converter 16 is stored in the memory unit 20 or 22 of the preprocessor 18 described later.

Reference numeral 18 finds only the feature points of the image data input through the A / D conversion unit 16 using a filtering element such as a differential filter, and selects only the points having a predetermined length or more and a predetermined length from the feature points. It is a preprocessing unit classified into an edge.

In the process of selecting the edge of the predetermined size or more in the preprocessing unit 18, a predetermined size is set. A threshold level is used for the setting, and the threshold value is variably changed according to the brightness of the road.

Reference numeral 20 is a first memory unit for the preprocessor 18, and reference numeral 22 is a second memory unit for the preprocessor 18. The first memory unit 20 and the second memory unit 22 alternately perform data memory operations by using a synchronization signal so that the preprocessor 18 can perform high-speed image processing.

In more detail, the preprocessor 18 has an image preprocessing and edge processing function, and an interface and preprocessing memory management function.

Referring to the image preprocessing and edge processing functions, the data through the A / D converter 16 is an original image, thereby reducing the processing burden of the post processor 24 and performing high-speed pixel-by-pixel processing. Here, the emphasis and pixel correlation of the edges are written to the first memory unit 20 in real time using differential and digital filters. The first memory unit 20 is divided into two banks, and when the first bank stores the preprocessing results, the second bank performs a read operation for display. The second memory unit 22 also has the same structure as the first memory unit 20.

The interface and preprocessing memory management functions will be described. In the standard NTSC video signal, one frame is composed of two field signals (even field signals and odd field signals). This field signal is used as a synchronization signal with the post processor 24. If the image preprocessing and edge processing functions are performed in the first and second banks of the first memory unit 20 for one field (eg, even field) time, the post-processing unit 24 may execute the first memory of the second memory unit 22. The result of image preprocessing and edge processing of the previous frame stored in one bank is taken, and the result is stored in the second bank of the second memory unit 22. The pre- and post-processes then perform the same function for another field (e.g., odd field) time, but the memory changes. In other words, the preprocessor 18 stores the result of the current frame in the first bank of the second memory 22 and stores the result of the current frame in the second bank of the second memory 22 for display. Read the result.

In this way, the memory is continuously changed and used by a synchronization signal, thereby performing a high performance image processing function in real time. In addition, the preprocessing unit 18 also performs signal conversion of signals for automobiles (SPEED, BRAKE, TRNR / L, ST1 / 2 / N).

TRNR / L is an abbreviation of Turn Right / Left as a turn signal, and ST1 / 2 / N is an abbreviation of Steering Angle 1/2 / Neutral as a rotation angle of a handle.

The preprocessor 18 has a built-in ROM that stores a drive program.

Reference numeral 24 is a post-processing unit that selects and smoothes only the edges that are determined to be lanes among the edges classified by the preprocessing unit 18, finds a valid lane, calculates the width of the lanes, and determines the degree of departure, and outputs a warning message. to be.

If the lane is broken or erased, it is difficult to identify the lane. In addition, the distinction between lanes is not accurate at the time of weather condition or at night or in rainy weather.

Accordingly, the post-processing unit 24 employs a method of easily finding the boundary line of the lane by emphasizing only the information of the lane which continues in the direction of travel (in the embodiment of the present invention, a method of superimposing the direction direction frame).

The direction limit frame superimposition method is a method in which a person easily finds a lane on a road. The person judges that the line which is white among the various objects on the road and the long continuous line in the driving direction is the lane, and that part is already learned in the brain and can be easily found. In other words, a person can easily find a lane with a characteristic of a white line that continues in the direction in which the car is traveling.

When the vehicle is traveling, lanes are drawn in the driving direction, and the obstacles around it are reflected in the shape of changing with time. That is, as illustrated in FIG. 3, the obstacles or the landscape on the road do not have this time difference because the lanes on the road continue to appear continuously as compared with the vehicle moving gradually. This phenomenon can be obtained by collecting the image coming through the image input unit 10 in time.

In the present invention, the direction limit frame superimposition method inserts the result of the difference in the memory in the continuous image input from the image input unit 10 and superimposes the differences of the continuous frames, resulting in the difference being added in the driving direction of the vehicle. On the other hand, the image of the surrounding environment or obstacle is shifted in time so that the difference is not different and the data is lower than the overlapped highlighted lane so that only the lane can be easily selected. .

The post-processing unit 24 has a built-in ROM that stores a drive program.

On the other hand, the post-processing unit 24 not only informs the driver by using a buzzer (not shown) and a flashing element (for example, LED) (not shown) according to the lane departure degree, but also controls the OSD generator 28. Outputs OSD text message for alarm. That is, the post-processing unit 24 differentially sets and outputs the frequency of occurrence of the buzzer sound or the blinking time difference of the flashing device according to the lane departure of the vehicle, and outputs the alarm OSD text along with the flashing operation of the buzzer sound and the flashing device. Print a message.

Reference numeral 26 is a memory unit that temporarily stores and outputs data generated during signal processing in the post-processing unit 24.

Reference numeral 28 denotes an OSD generator which generates an OSD text message for the alarm by the control signal of the post-processor 24 and provides it to the post-processor 24.

Reference numeral 30 denotes a synchronization generator which generates a synchronization signal (horizontal / vertical synchronization signal) necessary for generating a display image signal and provides the same to the preprocessor 18. In FIG. 1, the synchronization generating unit 30 is configured separately from the preprocessing unit 18, but may be embedded in the preprocessing unit 18.

Reference numeral 32 is a D / A converter for converting an image signal output through the preprocessor 18, that is, a digital video signal into an analog video signal.

Reference numeral 34 denotes a filter / amplifier for filtering noise components from the analog video signal for display output from the D / A converter 32 and amplifying them to a predetermined level.

Reference numeral 36 denotes an image output unit which displays an image by using the display analog video signal output from the filter / amplifier 34.

The image output unit 36 is installed above the driver's seat front panel as illustrated in FIG. 4.

Next, the operation of the lane departure warning apparatus according to the embodiment of the present invention will be described with reference to the flowchart of FIG. 5.

When the vehicle travels on the road, the image input unit 10 photographs the lanes on the road and obstacles existing on the road and inputs them to the amplification / filter unit 12 (step 100). The amplifying / filtering unit 12 amplifies the input video signal to a predetermined level, removes noise components of the video signal, and sends the same to the sync separator 14 and the A / D converter 16.

The sync separator 14 separates the field signal and the vertical / horizontal sync signal of the video necessary for preprocessing the video signal from the video signal (that is, the analog video signal) and provides the preprocessor 18 to the preprocessor 18. The / D converter 16 quantizes the analog video signal, converts the analog video signal into a digital signal, and provides the same to the preprocessor 18.

The preprocessing unit 18 finds only a feature point using a differential filter (not shown) for each frame with respect to the input video signal (that is, a digital video signal), and the point becomes more than a certain length and a predetermined length from the feature point. Only the bays are picked and classified as edges (step 110).

Here, when selecting the edge of predetermined size or more, it is determined that it is more than a predetermined size based on the threshold value preset in advance. The threshold value is automatically changed to an optimal value according to the road brightness.

In other words, to emphasize the vertical component (which is vertical when the lane is seen from the driver's side), a 2 * 1 mask and threshold are applied to extract the edge with the vertical component. The calculation method simply performs P (n, m) -P (n + 1, m) operation on the luminous intensity value of P (n, m) and P (n + 1, m) and includes the sign. Stored in D (n, m) pixels of differential image. If the sign of this value is +, it means that the brightness decreases from left to right, and-means the opposite. Where P represents an arbitrary function of the position of the pixel (applies to all frames), n represents the position of the pixel in X coordinates, m represents the position of the pixel in Y coordinates, and D Denotes the differential luminance value of the coordinate with respect to the differential image.

Subsequently, the post processor 24 first selects a search area for the edge having the vertical component extracted for each frame as described above, so as to select only the edge determined as the lane (step 120).

In the embodiment of the present invention, the lane equation extracted from the previous frame is defined as an edge search area of the current frame within 20 pixels of the left and right centers of the trajectory passing through the image plane. And, vertical positions where the trajectory of the equation is outside the size range of the image are given the basic search range. The basic search area is to find a new lane by dividing the left and right search areas based on the center line of the lane recognized in the previous frame when the lane is not found in the previous frame or when the lane is changed. This is because, in the conventional search area designation method, the road is designated by dividing the road from side to side. When the lane is changed to the other side, the lane tracking is stopped and the lane is not detected while searching for a new lane. Because.

When the search region designation for the current frame is completed, the post-processing unit 24 separates an object estimated as the lane width from the frame (step 130).

That is, when lanes appear for each vertical position, the edge points of the object corresponding to the widths are recorded in the lane edge candidate list by using a pre-estimated pixel width appearing on the screen. The lane width has a left point and a right point. In an embodiment of the present invention, an object is defined as an object when the luminance between the left and the right is brighter than the surrounding luminance.

When the object image is applied to the edge extraction method in step 110, as illustrated in FIG. 6, an edge with a-sign appears on the left side of the object and an edge with a + sign on the right side of the object. Therefore, by simply checking from the right to the left in the differential image, it can be concluded that the pixel distance from the + sign to the-sign is the width of the object. If the width is similar to the lane width previously estimated for each y-axis position, the width is recorded in the lane edge candidate list. At this time, record only two positions, one for left and one for each y-axis position.

Thereafter, the post processor 24 adjusts a threshold value used in the edge extraction step to remove noise edges and unnecessary edges (step 140). This is specified as the average differential value within the specified search area. It was conceived that the reflective properties of the paint used to apply the lanes of the road were tailored to appear clearly in the driver's field of vision, matching the distribution of ambient light.

In order to separate the object estimated by the lane width, the derivative values of the edge points referred to are added while referring to the edge points included in the search region described in step 120. This is done in the following order:

FOR [All Edge in Search area] ------- Loop 1

{

(y, x) = coordinates of the next search edge in the search area.

IF Edge (y, x) < Threshold THEN continue; -------- Treatment condition 1

IF (Edge (y, x) <11) --------------------------- Processing Condition 2

Total_Diff + = Edge (y, x);

Else continue;

Number_of_Edge ++;

Detect_Width (Edge (y, x));

}

At the end of this loop, object separation is complete, and Total_Diff contains the sum of the derivatives of the edges used in the search. And, Number_of_Edge contains the number of searched edge points.

Next, the following calculation is performed.

The threshold is used as a threshold in processing condition 1 when loop 1 is applied to the derivative of the next frame. The reason for subtracting 1 from Equation 1 is because the result of -1 from the average derivative is an appropriate threshold in the next frame.

According to the threshold setting method as described above, if the brightness difference of the objects in the search area in the current frame is large, the threshold goes up, and if it is small, that is, if the average of the brightness differences is small, the threshold goes down. This is called an automatic threshold adjustment method in the present invention. Since the automatic threshold adjustment method uses the threshold determined in the current frame as the threshold in the next frame, the threshold value is changed in real time according to the change of the lighting environment (eg, daytime, nighttime, in a tunnel, when it is cloudy or rainy) according to the road driving. It keeps changing.

Assuming that there is no processing condition 2 in the above manner, if there is a safety sign with a reflective tape in the search area at night, the contrast of the headlights will increase due to the reflection of the headlights, thereby causing the average threshold value to be excessive. There is a problem going up. In this case, there is a processing condition 2. Edges with excessively high derivatives are excluded from the calculation of the average brightness as well as from the object width calculation.

Thereafter, the post processing unit 24 groups the edge points associated with each other and removes the group having no lane feature (step 150). The criterion for grouping is as one group of edge points gathered within a predetermined pixel (for example, 4 pixels) from each other in the lane edge candidate list. Then, the groups having 10 or less edge points are removed. Finally, we extract the linear equations that can be explained by the minimum error for each group, and remove the group whose slope is negative in the left edge list and the positive slope in the right edge list from the lane edge candidate list (see FIG. 7). ).

In this state, in order to extract the driving lane among the lanes, the post-processing unit 24 removes only the groups having the first-order equation located inward and removes the groups located outside (step 160). As a result, the state is as shown in FIG.

The post-processing unit 24 groups the last remaining edge points in the lane edge candidate list into one and extracts an equation that can explain them with a minimum error (step 170). In the embodiment of the present invention, a Least square method is used to extract an equation that can be explained with the minimum error. The final calculated lane equation is obtained by the operation result of step 170.

Thereafter, the post processor 24 performs road modeling based on the extracted final lane equation (step 180). The road modeling operation is as follows.

Referring to FIG. 9, it is assumed that the EOF and the extracted lane equation are the lanes in which the vehicle is currently driving, and it is determined in which position in the screen area of the road. Whether the vehicle is separated is determined based on the center of the vehicle and the center distance of the extracted driving lane. At this time, a problem arises in which part of the lane is to be used for judging.

In FIG. 9, the center position S of the screen is a fixed value. The center of the driving lane is the position a when extracted from the bottom of the screen, and the position of the center becomes b, c, d, and e as it goes up.

When the degree of departure is S- (the driving lane center position), the degree of departure in case of a is S-a and the sign is negative, so it is determined that the deviation is to the right. In case of b, the degree of departure is S-b and the sign is negative. In case of c, the degree of departure is S-c and the sign is negative. In case of d, the degree of departure is S-d and the sign is negative. In the case of e, the degree of departure is S-e and the sign is positive.

The largest deviation degree in each case is a. Thus, the deviation value calculated in the same frame depends on the reference position of the lane determination. In each case, the calculated value of a is the value closest to the actual deviation. Accordingly, the reference position for deviation determination is defined as the lowest part of the screen (ie, y = 191).

The next thing to consider is the current lane width. The lane width is extracted as in FIG. 10.

If the direction of travel of the vehicle is parallel to the center line of the driving lane, and the vehicle moves in a straight line along the center line, the road width projected on the screen may be used as it is. However, it is different at the time of lane change or when driving near one side of a traveling lane. Referring to FIG. 11, a straight line A 'appearing on the screen becomes a straight line A on an actual road. This is not the width of the driving lane. The width of the traveling lane is straight line B. The appearance of this line on the screen is a straight line B '. Therefore, in order to calculate the exact road width of the driving lane, the following several processes should be added.

As shown in FIG. 12, in order to calculate the width of the driving lane, a point A and a point B where the straight line of y = 191 and the equations of the left and right lanes meet are designated as the positions of the lanes, respectively, and an intermediate point therebetween. Set as the lane center point in the screen coordinate system. Next, considering the phase of the vehicle with respect to the lane, the equation of a straight line passing through the center point of the lane perpendicular to the straight line passing through the EOF and the lane center point is obtained.

The two points where this equation meets the left and right lane equations are the left phase correction point (point A '') and the right phase correction point (point B ''), respectively. Is the distance between the current lane width (ie, the road width).

In this way, when the road modeling is completed (Yes in step 180), the post-processing unit 24 determines whether to activate the alarm logic by calculating the vehicle width calculation and the deviation degree using the equation of the finally extracted lane (step 190). The deviation calculation is as follows. Departure level = driving lane width / 2-vehicle width / 2-| screen center * coordinate-lane center point * coordinate |

On the basis of the departure level, it is determined whether the currently running vehicle is leaving a lane, and an alarm is issued when the departure level decreases below the alarm width. Then, if the lane faces outward or only one side is extracted, and the lane width calculated by the extracted equation exceeds the possible range (step 200).

At this time, the post-processing unit 24 not only informs the driver by using a buzzer (not shown) and a flashing element (for example, LED) (not shown) according to the lane departure degree, but also controls the OSD generating unit 28. The OSD text message for alarm is output to the video output unit 36. For example, measure the width of both lanes, calculate whether the car is close to which lane, and if there is a risk that the lane is close enough to leave the lane, give a first alarm, and then leave the lane completely. If there is a second alarm.

In addition, when the lane departure degree of the vehicle is severe at this time, the frequency of occurrence of the buzzer sound is suddenly increased, and the blinking operation of the blinking element is performed quickly. Then, an alarm OSD text message is output together with the blinking operation of the buzzer or the flashing device.

On the other hand, the post-processing unit 24 checks whether the system operation is abnormal, restarts the system if there is no abnormality in the system, and performs the speed signal processing.

As described above, when the vehicle stops driving while performing the alarm operation according to the lane departure degree of the vehicle (No in step 210), the lane departure determination and the alarm operation are terminated.

13 to 17 are graphs showing the results of experiments using the lane recognition method according to the present invention under various conditions. FIG. 13 is a graph showing experimental results of a straight section immediately after the driving test site. FIG. Fig. 15 is a graph of the experimental results when it rains slightly around 5 pm on the highway, Fig. 16 is a graph of the experimental results of the curve section at night at the driving test site, and Fig. 17 is a graph of the experimental results when the road rains during the daytime to be. 13 to 17, it can be seen that the lane recognition rate improves as the frame progresses.

According to the present invention as described above, by monitoring the driver's driving state (lane departure degree) to detect the situation caused by inadvertent driving and alert in advance, the driver to take appropriate action, thereby preventing accidents in advance Done.

On the other hand, the present invention is not limited to the above-described embodiment, but can be modified and modified within the scope not departing from the gist of the present invention, such modifications and changes should be regarded as belonging to the following claims. will be.

Claims (27)

Image input means for photographing a single image in front of the travel and outputting it as a video signal; Preprocessing means for extracting a predetermined size and a predetermined length or more edges in real time from the video signal from the video input means; Post-processing means for extracting only an edge corresponding to the lane from the extracted edge in real time, finding a valid lane at the extracted edge, calculating the width of the lane, and outputting lane information of a lane departure degree of the vehicle; And a video output means for displaying a lane image and a lane departure degree in real time based on the lane information extracted from the processing means. delete The method of claim 1, The image input means is a lane departure warning device, characterized in that installed in the vicinity of the rearview mirror in the cabin facing the road running. The method of claim 1, And the preprocessing means extracts an edge having a predetermined size and a predetermined length or more based on a threshold value that is variably set according to the brightness of an input video signal. The method of claim 1, The post-processing means is a lane departure warning device, characterized in that to emphasize only the information of the lanes running in the running direction to the boundary of the lane, and recognize the boundary of the lane as a valid lane. delete delete delete The method of claim 1, And the post-processing means displays an OSD text message on the video output means according to the lane departure degree of the vehicle. The method of claim 1, And the post-processing means differentially sets the frequency of occurrence of a buzzer sound according to the lane departure degree of the vehicle and displays an OSD text message on the image output means. The method of claim 1, And the post-processing means differentially sets the flashing time difference of the flashing element according to the lane departure degree of the vehicle and displays an OSD text message on the image output means. The method of claim 1, And the post-processing means differentially sets the frequency of occurrence of a buzzer sound and the flickering time difference of the flashing device according to the lane departure degree of the vehicle, and displays an OSD text message on the image output means. The method of claim 1, The image output means is a lane departure warning device, characterized in that installed on the driver's front panel. A first step of extracting a lane component for each frame by receiving a single image of a driving front and performing modeling of the current driving road in real time for each frame using only the optimal lane component from the extracted lane component; And And a second process of determining a lane departure of the vehicle in the modeled road and alerting according to the result. The method of claim 14, The first process includes a first step of setting an edge search region for each frame while extracting edges having vertical components for each frame of the input lane image; A second step of separating only an object estimated to be lane width in each frame edge search area and designating it as a lane edge candidate; A third step of extracting a lane edge candidate having lane characteristics from each of the lane edge candidates for each frame and grouping the extracted lane edge candidates into one frame to derive an optimal lane equation; And And a fourth step of performing road modeling based on the derived optimal lane equation. The method of claim 15, The first step may include setting the edge search region for the current frame within 20 pixels of the lane equation extracted from the previous frame, based on the trajectory passing through the image plane. Alarm method. The method of claim 15, The first step is to search for the vertical positions where the trajectory of the lane equation is out of the size range of the image. The lane departure warning method comprising the step of setting as a basic search area to find a new lane separated by. The method of claim 15, In the first step, the lane edge candidate for each frame is selected based on the threshold value in the corresponding frame, and the average value of the edge points in the search area in the current frame is designated as the threshold value in the next frame. Lane departure warning method comprising the step of adaptively auto-adjusting. The method of claim 15, The third step includes defining edge points within a predetermined pixel of the lane edge candidates as a group; Removing a group having less than a certain number of edge points among the groups; Extracting each linear equation that can be described as a minimum error for each remaining group, thereby preserving only the groups whose slopes satisfy certain conditions; And And extracting a final lane equation using groups having a first-order equation located in the innermost one of the preserved groups. The method of claim 14, In the first process, the lane of the image ahead of the driving stores the result of the difference in the continuous image inputted by the image input means in the memory, and overlaps the differences of the continuous frames so that the difference is added to the driving direction of the vehicle. And setting the image of the surrounding environment and the obstacle so that the image is overlapped without temporal difference so that the data is at a lower level than the highlighted lane. The method of claim 14, The second process is a lane departure warning method comprising the step of alarming by differentially setting the frequency of occurrence of the buzzer sound according to the lane departure degree of the vehicle. The method of claim 21, And displaying an OSD text message according to the lane departure degree of the vehicle. delete The method of claim 14, The second process of the lane departure warning method characterized in that for displaying the OSD text message further in accordance with the lane departure degree of the vehicle. delete delete The method of claim 1, And the preprocessing unit comprises a plurality of memory units, and performs a memory operation alternately with a process of the post-processing unit to process an image in real time.