KR101166349B1 - Apparatus and method for recognizing lane - Google Patents

Abstract

The present invention relates to a lane recognizing apparatus and method for stably determining a location of a lane according to a weight with reliability of left and right lanes even when a lane is not clear using a lane model. To this end, the present invention is a lane recognition device for lane departure warning of the vehicle, the image input unit for receiving a driving image of the vehicle; And extracting lane information including the lane recognition and the location of the lane through the lane recognition program from the driving image received from the image input unit, and extracting the left and right lane information and the left / right lane information of the extracted lane. It is determined whether the left / right lane information is input according to the reliability, and when the reliability of the left / right lane information is equal to or greater than a preset reference, the input is determined to determine whether the left / right double lane and the left / right are in the preset lane model. And an image processor configured to update location information of the right lane and determine whether to output a lane departure warning by determining whether the vehicle leaves the lane according to the updated lane model information. Therefore, the present invention has the advantage that the position of the lane can be grasped even in the dotted line section or the section where the lane is hard to see.

Description

Lane Recognition Apparatus and Method {APPARATUS AND METHOD FOR RECOGNIZING LANE}

The present invention relates to lane recognition, and more particularly, a lane recognizing apparatus using a lane model so that the position of a lane can be stably determined according to a weight with reliability of the left and right lanes even when the lane is not clear. It is about a method.

 In general, ASV (Advanced Safety Vehicle) is an advanced vehicle that applies advanced electronic and control technology to improve the safety of the vehicle. It reduces traffic accidents, increases traffic volume, saves energy and promotes driver convenience.

The Lane Departure Warning System (LDWS) detects the image of the road ahead from the camera attached to the vehicle and detects the current lane. It is a safety device that makes an alarm sound when you try to leave the lane by drowsy driving.

The lane departure warning system analyzes an image signal in front of the vehicle to determine whether the vehicle is out of the lane and a warning device that warns the driver of the fact that the vehicle is out of the lane as a result of the analysis of the lane detection apparatus. Include.

A general lane departure warning device for a vehicle is installed in front of a vehicle and receives an image of driving in advance to perform image processing for lane recognition, and then, based on the recognized lane, the width of the lane, the radius of curvature of the lane, the crossing time of the lane, and the like. After considering various variables, it is determined whether the vehicle has left the lane, and the lane departure warning and the situation are output to the driver according to the determination result.

However, in the conventional lane recognizing apparatus and method, it is difficult to determine the location of the lane in a section in which the lane is not clear, such as a dotted line section or a section in which one lane is hard to see.

In addition, the conventional lane recognizing apparatus and method has a problem in that it is possible to output a false alarm to the driver because a clear lane cannot be recognized on a busy road.

In order to solve this problem, the present invention provides a lane recognition apparatus and method for stably determining the position of the lane according to the weight with reliability of the left and right lanes even when the lane is not clear using the lane model. For the purpose of

In order to achieve the above object, the present invention provides a lane recognition device for lane departure warning of a vehicle, comprising: an image input unit configured to receive a driving image of the vehicle; And extracting lane information including the lane recognition and the location of the lane through a lane recognition program from the driving image received from the image input unit, and trusting the reliability and lane components of the extracted left and right lane information of the lane. Comparing the first reference value to determine whether there is a reliability of the lane information so that the current lane information is reflected if the reliability of the lane information is greater than or equal to the first reference value, and determining whether the reliability of the lane information and the lane component are unreliable. Compared with the second reference value, if the reliability of the lane information is less than or equal to the second reference value, previous lane information is reflected. If the reliability of the lane information is a value between the first and second reference values, the preset correction value is applied to the lane information. Determine the application of left / right lane information to be reflected, and trust the left / right lane information If is equal to or greater than a preset reference, the input is determined to update whether the left / right lanes are dual lanes and the location information of the left / right lanes in the preset lane model, and whether the vehicle leaves the lane according to the updated lane model. And determining an output of the lane departure warning.

The present invention may further include a lane departure warning unit configured to allow a driver to detect the lane departure warning when the lane departure warning is output from the image processing unit.

In addition, when the left / right lane is clearly input, the image processor according to the present invention increases the reliability of the left / right lane, updates the result by reflecting the result in the lane model information, and updates the left / right lane. If it is repeatedly input indefinitely, the reliability of the lane is reduced.

In addition, the present invention, if the reduced reliability is greater than or equal to the reference value for determining whether the lane component can be trusted reflects the result to the lane model, and if the reference value is less than the lane model is not updated. do.

In addition, the image processing unit according to the present invention is characterized by predicting and recognizing the lane through the previous lane model information when the reliability of both the left and right lanes is not updated to update the lane model.

The image processing unit may recognize a lane adjacent to the vehicle as a representative lane as a criterion for lane departure when the recognized lane is a double lane composed of two lines.

In addition, the image processor according to the present invention, when the lane of either the left or right side of the recognized lane is not recognized or is faintly recognized and the reliability of the lane is reduced, the image information of the other lane having high lane reliability is determined. It is characterized by predicting and recognizing the lane of the lower reliability through.

In addition, the present invention is a lane detection method for preventing lane departure of the vehicle, (a) extracting the driving image received from the image input unit through the lane recognition program in the image processing unit to extract lane information including the recognition of the lane and the location of the lane Making; (b) the image processing unit compares the reliability of the left and right lane information of the lane extracted in the step (a) with a first reference value for determining whether the lane component can be trusted. If the first reference value is greater than or equal to the first reference value, the current lane information is reflected, and if the reliability of the lane information is less than the second reference value compared with a second reference value for determining whether the reliability of the lane information and the lane component are unreliable, the previous lane Determining the application of the left / right lane information so that the information is reflected, and if the reliability of the lane information is a value between the first and second reference values, a preset correction value is reflected in the lane information; (c) if the reliability of the step (b) is greater than or equal to a preset criterion, determining whether to apply the lane and updating whether the left / right double lane and the location information of the left / right lane are preset in the preset lane model; (d) verifying whether or not the parallelism of the lane, the slope of the lane, and the width of the lane information are within a normal range based on the updated left / right lane information, and as a result, if the range is not within the normal range, the previous lane model information and the preset lane Updating to normal range lane model information using the information; And (e) recognizing left and right lanes according to lane information and reliability of the lane model updated in step (d).

In addition, the image processing unit according to the present invention further comprises the step of determining the output of the lane departure warning to the lane departure warning unit by determining whether the vehicle leaves the lane only for the lane recognized in step (e). It is done.

In addition, the reliability of the lane information of step (b) according to the present invention is that when the left and right lanes are clearly inputted, the reliability of the left and right lanes is increased, and the left and right lanes are unclearly repeated. When input, the lane reliability is reduced.

In addition, the reliability of the lane information of step (b) according to the present invention is characterized in that the reliability is reduced when the slope information of the lane does not change within a certain range.

In addition, the application of the lane information of the step (b) according to the present invention compares the reliability of the lane information and the first reference value for determining whether the lane component can be trusted, the reliability of the lane information is the first reference value If it is, the current lane information is reflected, and if the reliability of the lane information is less than the second reference value, the previous lane information is reflected in comparison with a second reference value for determining whether the reliability of the lane information and the lane component are unreliable. When the reliability of the lane information is a value between the first and second reference values, a preset correction value is reflected and applied to the lane information.

In addition, the step (c) according to the present invention, if the lane reliability of the lane is not recognized or faintly recognized in the left or right of the lane is reduced, the model information of the other lane with high reliability of the lane It is characterized by updating to predict and recognize the lane of the lower reliability through.

In addition, in the step (c) according to the present invention, if the step (c) is not recognized or faintly recognized in both the left and right lanes of the lane, the reliability of the lane is less than or equal to a second reference value at which the lane component is unreliable. After the previous lane model information is maintained for a certain period of time to update to change to the lane unrecognized state.

In the step (c) according to the present invention, when the lane is a double lane composed of two lines, the lane adjacent to the vehicle is set as the representative lane as a criterion for the departure of the lane.

The present invention has the advantage that it is possible to clearly determine the position of the lane even in the dotted line section, or the section is difficult to see the lane.

In addition, the present invention has a merit that can provide a clear lane recognition even in a road crowded with a lane model to prevent the false alarm output of lane departure.

In addition, the present invention has an advantage of improving the reliability of the lane departure warning device to improve the convenience and safety of the driver.

1 is a block diagram showing the configuration of a general lane departure warning system.
2 is a flowchart illustrating a lane recognition process according to the present invention.
3 is an exemplary view showing a lane model for lane position determination in a lane recognition process according to the present invention.
4 is an exemplary diagram showing a table used for updating a lane position in a lane recognition process according to the present invention.

Hereinafter, exemplary embodiments of a lane recognizing apparatus and method according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing the configuration of a general lane departure warning system.

As shown in FIG. 1, a lane recognizing apparatus for lane departure warning of a vehicle according to the present invention includes an image input unit 100 receiving a driving image of a vehicle and a lane from a driving image received from the image input unit 100. After determining the whether to apply the left / right lane information according to the reliability of the left and right lane information and the left / right lane information of the recognized lane, if the application of the left / right lane information is determined in advance An image processing unit 200 for determining an output of a lane departure warning by determining whether the vehicle leaves the lane according to the updated lane model information and updating the set lane model, and the image processing unit 200 to the lane If the departure warning is output, the lane departure warning unit 300 to enable the driver to detect the lane departure warning.

The image input unit 100 uses a CCD or CMOS as a camera installed in a vehicle, and is mainly installed in a room mirror installed inside the vehicle to detect a front driving image of the vehicle and provide the image to the image processing unit 200.

The image processor 200 is configured using a microprocessor, and recognizes the lane from the driving image of the vehicle through an embedded lane recognition program, and extracts lane information including position information of the recognized lane. The lane recognition program may use a known technique.

In addition, the image processor 200 determines whether to apply (input) the left / right lane information according to the reliability of the left and right lane information and the left / right lane information of the extracted lane.

When the image processor 200 clearly provides the left and right lanes recognized by the lane recognition program from the driving image received through the image input unit 100, the image processor 200 increases the reliability of the lane information and increases the increased lane information. The lane model information is updated with the current lane information including the reliability.

That is, in a section in which a lane, such as a center line or a bus-only lane, is clearly recognized, the reliability of the input lane information is increased, thereby enabling stable lane recognition.

The lane model has a value according to the left lane, the right lane, the slope of the lane, the reliability of the left lane and the reliability of the right lane, and the reliability of the lane is basically determined by the degree of certainty of the lane.

In addition, the image processing unit 200 cannot clearly recognize the position of the lane, such as a section in which the lane is a dotted line or a section in which the lane is difficult to see, and thus lane information having low reliability for the lane is repeatedly inputted. The lower reliability is compared with the predetermined reference value to determine whether to apply to the lane model.

If the reduced reliability is equal to or greater than the preset reference value, the lane information is given a predetermined weight, and then the lane model is updated with the lane information having the weight.

In addition, if the lowered reliability is less than or equal to the preset reference value, the lane information is not updated, and the lane model is updated with previous lane information.

In addition, the image processor 200 determines an input so that current lane information is applied when the reliability of the left / right lane information is greater than or equal to a preset reference, and updates the left / right lane information in the preset lane model.

In this case, it is determined whether the left / right lane is a double lane, and in the case of the double lane consisting of two solid lines, the lane adjacent to the vehicle is set as the representative lane for the lane departure determination.

In other words, there are various lanes on the left and right side of the road, such as the center line, the bus-only lane, and the lane change line for entering and exiting the vehicle at an interchange or intersection, so when two lanes are detected, one lane is selected to determine whether the lane is out of position. Set as the representative lane for.

In addition, when the reliability of the lane is reduced because the lane of either the left or right side of the recognized lane is not recognized or is faintly recognized, the image processing unit 200 uses the reliability of the other lane model information. Recognizes by predicting the lower lane.

That is, since most lanes are arranged in parallel, when one lane of both lanes is clearly recognized and the other lane is not clearly recognized, a certain weighted reliability is given to the lanes that are not clearly recognized. Based on this, the lane is predicted and recognized, and the lane departure warning is performed only for the recognized lane.

Meanwhile, if the reliability of both the left and right lanes is less than or equal to a preset reference value, the image processor 200 does not update the lane model and predicts and recognizes the lane through previous lane model information, and recognizes only the recognized lanes. Will trigger an alarm.

In addition, the image processor 200 may have normal lane information when the parallelism, the slope of the lane, the width of the lane, and the like exist within a preset normal range by using left / right lane information of the updated lane model. If the lane parallelism, lane inclination, lane width information, etc. are out of the normal range, the previous lane model information and the preset lane information are applied to the lane model, and the parallelism of the lane, the slope of the lane, The width information and the like are updated to be a lane model having a normal range.

In addition, the image processing unit 200 determines whether the vehicle leaves the lane according to the updated lane model information, and outputs a lane departure warning to the lane departure warning unit 300 when the vehicle leaves the lane.

The lane departure warning unit 300 may display a display function through a display means and a sound output means and fasten a seat belt such that the lane departure warning signal output according to the determination result of the image processor 200 may be recognized by a driver. Perform haptic functions.

2 is a flowchart illustrating a lane recognition process according to the present invention, FIG. 3 is an exemplary view illustrating a lane model for determining a lane position of the lane recognition process according to the present invention, and FIG. 4 is a view illustrating a lane recognition process according to the present invention. It is an example figure which shows the table used for updating a lane position.

A lane recognition method according to the present invention will be described with reference to FIGS. 1 to 4.

When the driving image is input by the image input unit 100 (S100), the image processing unit 200 extracts the lane information including the left and right lane recognition and the position of the left / right lane through the lane recognition program. (S200).

As shown in FIG. 3, the lane model is a virtual lane that is always present for lane recognition and includes dual lanes consisting of two solid lines and slope information of a lane.

That is, a double lane composed of a left-outer lane 1 and a left-inner lane 2 is disposed on the left side of the vehicle in the traveling direction, and a right-inner lane 3 and a right-outer lane ( 4, a double lane is arranged, and the left-inner lane 2 and the right-inner lane 3 are used as representative lanes for lane recognition, and the left-inner lane 2 and the right-inner lane ( The lane slope 5 is disposed between 3).

The image processor 200 determines whether to apply the left / right lane information to the lane model according to left and right lane information, that is, reliability, extracted in the step S200 (S300).

In step S300, if the left / right lane information satisfies a preset reference value, it is determined to apply to a lane model. If the left / right lane is a double lane, the lane adjacent to the vehicle is represented as a criterion for departure of the lane. By setting the lane, the location information of the left / right lane is updated in the lane model (S400).

In the step S300, if the left / right lanes are repeatedly inputted clearly, the reliability of the left / right lanes increases, and if the left / right lanes are inputted indefinitely, the reliability of the lanes decreases.

That is, the lane information on the left and the right has a certain range of reliability, respectively, and when the lanes inputted in the image are clearly and repeatedly inputted, the reliability of the lanes increases, and the lanes inputted through the image are repeatedly When input, the reliability of the lane is reduced, thereby updating the lane model.

In addition, when the reliability of the lane falls below a predetermined reference value at which the lane cannot be recognized, the lane is not recognized.

When the update of the lane model according to the reliability is described in more detail at step S400, the image processor 200 may detect information on the left and right lanes, and may trust the lane component based on the detected left and right lane information. The first reference value, which is the present value, is compared.

As a result of the comparison, if both the left and right lane information are greater than or equal to the first reference value, the current lane information is updated in both lanes to be reflected in the lane model information.

In addition, when the left / right lane information has a value between the first reference value and a second reference value that is a value of which the lane component is unreliable, the lane information is reflected in the lane model information by reflecting a preset correction value.

That is, when any one of the left and right lane information is equal to or greater than the first reference value and the other left and right lane information has a value between the first reference value and the second reference value, the left and right lanes in the current image frame. Information on the left and right lane information in one frame before the frame, a weight of a predetermined size is applied to the average interval between lanes, and a predetermined interval between lanes to update lane information, and then generate lane model information, and the condition is repeated. In this case, the reliability decreases.

In addition, when any one of the left and right lane information is greater than or equal to the first reference value, and the information of the other left and right lanes has a value less than or equal to the second reference value, left and right lane information and one frame in the current video frame. The lane information is generated after updating the lane information by giving a predetermined amount of weight to the left / right lane information, the average interval between lanes and the preset interval between lanes in all images, and generating the lane model information, and the reliability is decreased when the condition is repeated. do.

Meanwhile, a value obtained by subtracting the left / right lane information from the left / right lane information from the left / right lane information is compared with a boundary value of a range within which a lane component can move within one frame of the image, and the subtracted value is equal to the subtracted value. If the case below the threshold is not satisfied, the lane information in question is not updated, and if repeated, the reliability decreases.

As shown in FIG. 4, the lane model may be updated so that current lane information is reflected in both lanes according to the reliability of the left and right lanes, or the first to fifth correction values given to have a certain weight in one lane are reflected. The previous lane information may be maintained without updating the lanes or updating both lanes.

The first correction value is, for example, a value corrected by giving a constant weight a to current lane information and average lane width information, and the second correction value is a weight b to current lane information, average lane width, and preset lane width information. The third correction value is a value corrected by giving a weight c to the current lane information and the average lane width information, and the fourth correction value is a weight to the current lane information, the average lane width, and the preset lane width information. The fifth correction value is a value corrected by giving a weight e to current lane information, average lane width, and preset lane width information.

In addition, the inclination 5 information of the lane may change within a predetermined range, and if it does not change within the predetermined range, the lane model information is not updated, and when the inclination 5 of the lane does not change within a predetermined range, Reliability decreases.

In addition, two lanes are always parallel, and when the parallelism between the two lanes differs by more than an error range between the upper and lower points of the two lanes, the less reliable lane is given the average interval information between the lanes based on the highly reliable lane. The lane model information is updated according to the given information, and the reliability of the lane is reduced when the condition is repeated.

In addition, when one of the lanes on the left or right side of the lane is not recognized or is faintly recognized, and the reliability of the lane is decreased, the lane of the lower reliability is predicted through the model information of the other lane having the high reliability of the lane. The lane model information to be recognized.

In addition, if both the left and right lanes of the lane are not recognized, or both lanes are dimly recognized so that the reliability of the lane is less than or equal to a second reference value at which the lane component is unreliable, after maintaining the previous lane model information for a predetermined time, The lane model information is updated by changing to the recognition state.

After performing the step S400, based on the updated left / right lane information, an error is verified whether or not the parallelism, the slope of the lane, and the width of the lane are in a normal range, and as a result, the previous lane model information is not in the normal range. Using the preset lane information and to update the lane model information of the normal range (S500).

Recognizing left and right lanes according to the lane information and the reliability of the lane model updated in step S500 (S600), the lane departure warning unit 300 by determining whether the vehicle leaves the lane only for the lanes recognized in step S500 Determines whether the lane departure alarm is output.

Therefore, even when the lane is not clear, it is possible to stably determine the position of the lane according to the weight with reliability of the left and right lanes using the lane model, and it is possible to alert only the recognized lane according to the determination. .

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. It can be understood that

1: left-outer lane 2: left-inner lane
3: right-side lane 4: right-side lane
5: slope of the lane 100: image input unit
200: image processing unit 300: lane departure warning unit

Claims (15)

A lane detection device for lane departure warning of a vehicle,
An image input unit configured to receive a driving image of the vehicle; And
The lane image including the recognition of the lane and the position of the lane is extracted from the driving image received by the image input unit through a lane recognition program, and whether the reliability and lane components of the extracted left and right lane information of the lane are reliable. Comparing a first reference value for determining whether or not the reliability of the lane information is greater than the first reference value to reflect the current lane information, and a second for determining whether the reliability of the lane information and the lane component is unreliable Compared with a reference value, if the reliability of the lane information is less than or equal to the second reference value, previous lane information is reflected. If the reliability of the lane information is a value between the first and second reference values, a preset correction value is reflected in the lane information. The application of left / right lane information is determined, and reliability of the left / right lane information is determined. If it is equal to or more than a preset reference, the input is determined to update whether the left / right lane is a double lane and the location information of the left / right lane in the preset lane model, and whether the vehicle leaves the lane according to the updated lane model. And a video processor configured to determine whether to output a lane departure warning. The method of claim 1,
And a lane departure warning unit configured to allow a driver to detect the lane departure warning when the lane departure warning is output from the image processing unit. The method according to claim 1 or 2,
If the left / right lane is clearly input, the image processor increases the reliability of the left / right lane, updates the result by reflecting the result in the lane model information, and if the left / right lane is repeatedly inputted in an unclear manner. The lane recognizing apparatus, characterized in that for reducing the reliability of the lane. The method of claim 3, wherein
And if the reduced reliability is greater than or equal to a reference value for determining whether to trust a lane component, the result is reflected in the lane model and updated, and if less than the reference value, the lane model is not updated. The method of claim 3, wherein
And the image processor predicts and recognizes the lane based on previous lane model information when the reliability of both the left and the right lanes is decreased and the lane model is not updated. The method according to claim 1 or 2,
And the image processing unit recognizes a lane adjacent to the vehicle as a representative lane that is a criterion for lane departure when the recognized lane is a double lane composed of two lines. The method according to claim 1 or 2,
If the reliability of the lane is reduced because one of the lanes on the left or right side of the recognized lane is not recognized or is faintly recognized, the image processor may have a lower reliability level through model information of the other lane having higher lane reliability. Lane recognition apparatus, characterized in that for predicting the lane of the recognition. As a lane recognition method for preventing lane departure of a vehicle,
(a) extracting lane information including the lane recognition and the location of the lane through the lane recognition program from the image processing unit using the driving image received from the image input unit;
(b) the image processing unit compares the reliability of the left and right lane information of the lane extracted in the step (a) with a first reference value for determining whether the lane component can be trusted. If the first reference value is greater than or equal to the first reference value, the current lane information is reflected, and if the reliability of the lane information is less than the second reference value compared with a second reference value for determining whether the reliability of the lane information and the lane component are unreliable, the previous lane Determining the application of the left / right lane information so that the information is reflected, and if the reliability of the lane information is a value between the first and second reference values, a preset correction value is reflected in the lane information;
(c) if the reliability of the step (b) is greater than or equal to a preset criterion, determining whether to apply the lane and updating whether the left / right double lane and the location information of the left / right lane are preset in the preset lane model;
(d) verifying whether or not the parallelism of the lane, the slope of the lane, and the width of the lane information are within a normal range based on the updated left / right lane information, and as a result, if the range is not within the normal range, the previous lane model information and the preset lane Updating to normal range lane model information using the information; And
and (e) recognizing left and right lanes according to lane information and reliability of the lane model updated in step (d). The method of claim 8,
and (f) the image processing unit further comprises the step of determining whether the vehicle leaves the lane only for the lane recognized in step (e) and determining the output of the lane departure warning to the lane departure warning unit. Recognition method. 10. The method according to claim 8 or 9,
The reliability of the lane information of step (b) is that the reliability of the left / right lanes is increased if the left / right lanes are clearly repeated, and the reliability of the lanes is input if the left / right lanes are unclearly repeated. Lane recognition method characterized in that the decrease. 10. The method according to claim 8 or 9,
And the reliability of the lane information of step (b) decreases when the slope information of the lane does not change within a predetermined range. delete 10. The method according to claim 8 or 9,
In the step (c), when the lane of either the left or the right of the lane is not recognized or is faintly recognized and the reliability of the lane is decreased, the reliability is lower through the model information of the other lane having the higher reliability of the lane. And recognizing and updating the lane to predict and recognize the lane. 10. The method according to claim 8 or 9,
In step (c), if both the left and right lanes of the lane are not recognized or are faintly recognized, and the reliability of the lane is less than or equal to a second reference value at which the lane component is unreliable, the lane may be maintained after maintaining the previous lane model information for a predetermined time. And updating to change to an unrecognized state. 10. The method according to claim 8 or 9,
In the step (c), if the lane is a double lane consisting of two lines, the lane detection method characterized in that the lane adjacent to the vehicle is set as the representative lane based on the determination of lane departure.

Images