KR20010096265A - Method for lane sencing of vehicle - Google Patents

Abstract

PURPOSE: A traffic lane detecting method is provided to improve traffic lane detecting performance by detecting a yellow traffic lane left to a first lane of a road. CONSTITUTION: A traffic lane is detected by the following steps. Firstly, image signals of a traffic lane are inputted from a camera(201). An edge is sampled by applying a mask and a limitation to the input image signals(202). If the traffic lane sampled in the edge sampling step, a search region for a new traffic lane is set up(203). Traffic lane edge lists are recorded by separating the edge points of an object matched with the width of a traffic lane(204). For removing noises and unnecessary edges, application limit is adjusted(205). Correlated edge points are grouped, and the groups non-related with the width of a traffic lane are removed(206). For sampling a travel lane, outer groups except innermost groups having a simple equation are removed(207). A polynomial expression is sampled by grouping the remaining edge points into one group(208). If the sampled traffic lane equation is appropriate, a road is modeled(209). Using the sampled traffic lane equation, displacement of a vehicle out of the traffic lane is decided(210). Then, if the vehicle is out of the traffic lane, an alarm is issued. If a traffic lane is not detected in the modeling step, existence of the leftmost traffic lane is decided(211). If the leftmost traffic lane is not detected, the pictured road and the traffic lane gray level are differentiated and a leftmost traffic lane is detected by inverting the gray level to the differentiated road and the leftmost traffic lane(212). According to the procedure, traffic lane detecting performance is improved.

Description

Lane detection of vehicle {METHOD FOR LANE SENCING OF VEHICLE}

The present invention relates to a method for recognizing a lane of a vehicle. More particularly, the present invention relates to a method for recognizing a lane of a vehicle to detect a yellow line drawn on a concrete road while driving a road while detecting whether the vehicle is out of a lane.

Conventionally, a lane detecting apparatus for a vehicle driving on a road while detecting whether a lane is recognized is determined by using the same filter when binarizing to recognize and recognize a lane.

In other words, the lane is recognized by finding a boundary of a portion in which a dark gray value (asphalt road) is changed from a dark gray value (asphalt road) to a light gray value (lane) with respect to an image signal (binarization signal) captured by the camera.

However, the lane recognizing apparatus recognizes lanes when driving on the road. As for the asphalt road, the dark gray value and the light gray value are distinguished from each other. However, in the case of the concrete road, the yellow lane marking the center line of the road has no problem. Had

Of course, when the camera of the apparatus for recognizing a lane is a color, the gray values due to the colors are distinguished, which is not a problem.

However, most lane-recognizing camera cameras are mainly black and white cameras in consideration of economic problems, so they are binarized when taking road lanes, so the value input through the filter does not differ between the dark and light gray values. You will not be able to distinguish lane boundaries.

Therefore, when driving on the concrete road, the lane recognizing apparatus may not function again.

As described above, the conventional lane recognizing apparatus easily distinguishes a boundary between a road and a lane when driving an asphalt road, but in the case of a concrete road, the yellow line cannot be distinguished from the road. In consideration of being drawn, there is a problem in deterioration of recognition capability and reliability of the lane recognizing apparatus.

Accordingly, an object of the present invention is to improve lane recognition ability by detecting a yellow line drawn on the left side of one lane of a concrete road when driving a concrete road while driving while determining a lane departure by recognizing a lane while driving. It is.

1 is a schematic control block diagram of a lane recognizing apparatus used in the present invention.

Figure 2 is a flow chart for the lane detection method of the present invention.

Figure 3a is a concrete asphalt road in the lane recognition method of the present invention.

3b is a concrete road to the lane recognition method of the present invention.

In order to realize the above object, the present invention provides a lane recognition method for recognizing a lane as a change in a binarized gray signal while driving on a road. Determining whether a lane has been recognized; If the final left lane is not recognized in this step, the gray level between the road background color and the lane is differentiated by the differential filter, and it is determined whether the road is a concrete road by this differential value. It features.

Therefore, according to the present invention, the road lanes and lanes are determined by determining whether the road is a concrete road by differentiating the gray level signals of the road and the lane image signals photographed while driving while determining whether there is a lane departure by recognizing the lane of the road. By reversing and recognizing the gray level of, it is possible to improve the capability and reliability in lane recognition.

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

1 is a control block diagram of a lane recognizing apparatus used in the present invention, comprising: a camera 10 photographing a lane of a driving road; An image processor 11 for amplifying the image signal photographed by the camera 10 and filtering the binarized gray signal and outputting the filtered signal; Control means (12) for comparing and controlling the image signal processed by the image processing unit (11) as a binarized gray signal by a predetermined program, and determining whether the lane is separated while storing the controlled signal; A monitor (13) for displaying the control signal output from the control means (12); The control means 12 is configured to the alarm unit 14 to alert the driver when the lane departure determination.

2 is a flowchart of a lane recognition method of the present invention, comprising: an image signal input step 201 for inputting a front lane image signal from a camera; An edge extraction step (202) for extracting an edge by applying a mask and a limit to the image signal input in the video signal input step (201); A search region designation step (203) for designating an area for searching for a new lane from the center of the screen when the lane extracted in the edge extraction step (202) is not found or the lane is changed; An estimated lane width separation step of recording a lane edge candidate list by dividing edge points of an object corresponding to the lane width by a pre-estimated value of the pixel width appearing on the screen in the lane search area designated in the search area specifying step 203 ( 204); An application limit adjustment step (205) of adjusting the application limit for noise removal and unnecessary edge removal in the lane edge candidate list in the estimated lane width separation step (204); An edge grouping step 206 of grouping the mutually related edge points adjusted in the application limit adjustment step 205 and removing the group having no lane feature; In the edge grouping step 206, the outer deflection edge group removing step removes only the groups having the innermost linear equation to remove the driving lane among the lanes of the grouped edges. 207); A polynomial extraction step (208) for extracting a lane equation that can be described as a minimum error by grouping the edge points remaining in the outer deflection edge group removal step (207); A road modeling step (209) of extracting lanes after determining whether the lane equation extracted in the polynomial extraction step (208) is suitable for the shape of the road; A lane departure determination step 210 of determining lane departure based on whether the lane departure is within a lane departure warning range using the extracted lane equation when the lane is extracted in the road modeling step 209; An alarm step 211 for generating an alarm upon lane departure in the lane departure determination step 210; A final left lane recognition determination step 211 for determining whether the final left lane of the road is recognized if the lane is not extracted in the road modeling step 209; If the last left lane is recognized in the final left lane recognition determination step 211, if the last left lane is not recognized, the photographed road and the lane gray level are differentiated by a differential filter and the differentiated road and the last left lane A lane recognition step 212 is performed in which the gray level is inverted to recognize the final left lane.

According to the present invention made as described above, first, a camera 10 for photographing a lane on a vehicle is mounted at a randomly set position (center) of the vehicle, and the image signal photographed by the camera 10 is a lane installed inside the vehicle. It is installed to be input to the recognition device.

When the vehicle starts to be started in this state and the lane recognition device is operated, the camera 10 photographs the driving lane to output an image signal, and the output image signal is an image of the lane departure determining apparatus. The image processing unit 11 is input to the processing unit 11 and sets the frame processing, line processing, and capture / display sections based on the frame, vertical, and horizontal synchronization with respect to the input image signal, and then the control unit 12. When inputted, the control means 12 constitutes a memory and a filter, and performs lane detection and lane departure determination by using a program preset in the memory with respect to the input video signal. In addition to outputting an image through the alarm unit 14, an alarm is issued when the lane departs, and the control unit 12 describes the lane recognition process. In operation 201, a lane image signal photographed through the camera 10 and a signal processed by the image processor 11 are input (step 201).

Subsequently, when an image processed signal is input, the control means 12 applies a 2 × 1 mask and a threshold value to extract an edge using a vertical edge image (vertical component) that is well represented without damaging the directionality of the lane as a lane detection component. (Step 202).

In this way, the extracted edge designates the edge search section of the current frame within 10 pixels of the left and right centers on the trajectory of the lane extracted from the previous frame in the image plane (step 203), where the trajectory of the equation is the image. Vertical positions outside the size range of are given a default search area.

The basic search area is for finding a new lane from the center of the screen when the lane is not found in the previous frame or when the lane is changed.

When the search area is specified, the control means 12 uses the estimated value of the pixel width appearing on the screen when lanes appear for each vertical position of the y-axis on the screen to select the edge points of the object corresponding to the width. Record on the list.

That is, the pixel width has a left point and a right point, and finds any two points among a number of edge points appearing on a line in the image and corresponds to each other to find a width between them.

In this case, when the luminance between the left and the right is brighter than the surrounding luminance, the object is defined as an object. The object is subjected to an edge extraction process, and an edge with a-sign appears on the left side, and an edge with a + sign on the right side.

Therefore, in the differential image, it is assumed that the pixel distance from the right sign to the left sign is the width of the object and is similar to the lane width estimated by the y-axis position. In the lane edge candidate list, only two positions, one for each of the left and one for each y-axis position, are recorded (step 204).

When the object fits the estimated lane width as described above, the control means 12 adjusts the threshold value used in the edge extraction step for noise removal and unnecessary edge removal at the lane edge, which is an average in the designated search area. It is specified as a derivative.

The reason is that the application limit of the lane is adjusted in consideration of the reflection characteristics of the paint used to apply the lane of the road to the driver's field of vision, clearly matching with the ambient light distribution (step 205).

As such, when the lane application limit value is adjusted, the control means 12 groups the edge points that are related to each other among the objects shown in the lane width and removes the group having no lane feature. In the grouping criterion, edge points gathered within 4 pixels from each other in the lane edge candidate list are formed into one group, and then groups having 10 or less edge points are removed. Finally, the linear equations that can be explained with the minimum error for each group are extracted, and the group having the negative slope in the left edge list and the positive slope in the right edge list is removed from the lane candidate list (step 206).

The control means 12 removes only the groups having the first-order equation located inward, for the purpose of extracting the driving lane, among the several lanes (step 207).

Subsequently, when the driving lane is extracted, the control means 12 extracts an equation using a least square method that can group the last remaining edge points in the lane edge candidate list into one and explain them with a minimum error (step 208). .

When the equation of the lane is extracted, the control means 12 extracts the left and right lanes while determining whether the extracted lane equation is suitable for the shape of the road and modeling the road (step 209).

Subsequently, the control means 12 determines whether there is a lane departure by a predetermined lane departure determination algorithm with respect to the left and right lanes extracted from the road modeling (step 210). If the departure, the control means 12 controls the alarm unit 14 to warn the driver of lane departure.

However, if both the left and right lanes are not extracted from the road modeling, the control means 12 determines whether the final left lane (center line) is recognized. At this time, it is determined that the driving lane is asphalt. Because there is a road, there is a concrete road.

That is, when the driving road is an asphalt road, as shown in FIG. 3A, an image signal photographed and input by the camera 10 is converted to a road surface (black) when the image processing unit 11 converts a digital signal. Since the gray level is significantly lower than the gray level of the lane (white or yellow), the lane can be easily determined.

On the other hand, as shown in FIG. 3B, when driving a concrete road, the image signal photographed and input by the camera 10 is converted to gray on the road surface (white) when the image processing unit 11 converts the digital signal. The level will appear no difference than the gray level of the left and right lanes (yellow or white).

In other words, looking at the gray level of the image signal around the final left lane (yellow) on the asphalt road, the gray level signal of the left edge of the lane, the center of the lane, and the right side of the lane center in the direction of travel are 50, 150, 150, respectively. If it appears, it can be seen that the smaller the gray level signal as described above, the closer the background is to black, and the larger the gray level signal, the closer the background is to white.

Therefore, the smaller the gray level signal processed for the road and lane, the blacker the background, and the larger the gray level, the whiter the background. The centerline will appear darker.

Therefore, in the direction of travel, the gray level signal of the outside of the left edge of the center line, the center of the lane, and the right side of the center of the lane is represented as 100, 80, 80, respectively, so that the gray level of the road is large, whereas the center line has the gray level small Since the control device 12 cannot distinguish the center line by the gray level shown as described above, it is determined whether the final left lane is recognized (step 211).

So, if the last left lane is recognized, the return is made. On the other hand, if the last left lane is not recognized, the control device 12 recognizes that the current road is not an asphalt road but a concrete road. As shown in FIG. 3A, when the asphalt road is differentiated with the filter (-1,0.1) for gray levels 50, 150, and 150, 50 (-1) + 150 (0) + 150 (1) = 100 3b, 100 (-1) + 80 (0) + 80 (1) = − when differentiating with a filter (-1,0,1) for gray levels 100, 80, and 80 when driving on a concrete road Will be 20.

The control means 12 recognizes whether the traveling road is an asphalt road or a concrete road by the value differentiated by the filter. The control means 12 drives the asphalt road as shown in FIG. 3A. In the case of filtering signals for roads (black) and lanes (yellow), the roads are filtered to low gray levels and the lanes to high gray levels.

On the contrary, as shown in FIG. 3B, in the case of a concrete road, the road (white) is recognized as a high gray level and the lane (yellow line) is regarded as a low gray level. You will make a mistake.

Therefore, when the control device 12 determines that the concrete road is driven, the low gray level is recognized as the lane for the filtered road and the lane, while the high gray level is determined by the inverse filtering to determine the road. When driving the concrete road, it is possible to travel while detecting the yellow lane (step 212).

As described above, the present invention determines whether the left lane is modeled in the lane that is recognized during the one-lane driving while recognizing the lane as the gray level change while driving the road, and the gray of the road and the lane that is filtered when the left lane is not modeled. By reversing and recognizing the level, it is possible to greatly improve the lane recognition ability to drive while detecting a yellow line drawn on the left side of the first lane of the road even when driving on a concrete road.

Claims (2)

Inputs the lane image signal in front of the road, extracts the edge from the input image signal, designates the area to search for a new lane with the extracted edge, and in this designated lane search area, After separating, the lane edge candidate list is recorded, and after removing noise and eliminating unnecessary edges in the lane edge candidate list, grouping edge points that are related to each other and leaving the groups having only the innermost first-order equation, leaving the outer group After removing them, the remaining lane points are grouped into one to extract a lane equation that can be explained with a minimum error, and the lane recognition that determines whether the extracted lane equation is suitable for the shape of the road and determines lane departure. The method, wherein if the lane is not extracted, the last left lane of the road Determining whether this is recognized; And if the last left lane is recognized in the step, if the last left lane is not recognized, inverting the gray levels of the photographed road and the last left lane to recognize the last left lane. Lane Recognition Method. The method of claim 1, wherein the gray level inversion recognition is performed by filtering the vehicle to be inversely recognized as a road when the gray level is high and a lane when the gray level is low.