WO2013035951A1 - Apparatus and method for detecting traffic lane in real time - Google Patents

Abstract

Disclosed are an apparatus and method for detecting a traffic lane in real time. The disclosed apparatus for detecting a traffic lane in real-time includes: a candidate region setup unit for setting a region having an intensity corresponding to the brightness of a traffic lane, among the intensities of brightness in a color space of a color image, as a candidate region for detection of a traffic lane; and a traffic lane determination unit for determining a traffic lane in the candidate region using a straight-line component of the set candidate region. According to the present invention, it is possible to accurately and rapidly detect a traffic lane even in environments with various changes in the lighting of roads.

Description

Electronic Documents Available PCT

 [Appendix—Form 43]

[Name of invention]

 Real time lane detection device and method

 Technical Field

 Embodiments of the present invention relate to a real-time lane detection apparatus and method, and more particularly to a real-time lane detection apparatus and method that can detect the lane accurately and quickly even in various lighting changes environment of the driving road.

 Background Art

 As interest in intelligent cars, which are fused with IT technology, is increasing, research on this is being actively conducted. In particular, Advanced Safety Vehicle (ASV) technology, which provides drivers with lane departure warning, front / side vehicle crash warning, and pedestrian crash warning, is a foundation for intelligent vehicle technology. Research and technology development are underway.

 Detection of lanes from color images on roads is a key technology of advanced safety vehicle (ASV) technology, and vision based lane detection method using image sensor is most widely used.

 The vision-based lane detection method consists of three stages: feature extraction, outlier removal and post-processing, and lane tracking. The performance and characteristics of the lane detection algorithm are determined by the algorithm used in each step. .

 In Vision Based Lane Detection Method. The features used can be broadly divided into edges and colors. However, the vision-based lane detection method using color as a feature has a problem in that it is difficult to apply to a road environment in which a change in lighting is found because it is very vulnerable to a change in lighting.

 That is, since the road environment has various lights according to time and place, and the color of the lane also has various values according to the change of the illumination, reliability of lane detection may be reduced when using color.

 For example, in a tunnel or at night, the road environment is usually yellow by sodium lamps, so the color of the lane changes with yellow lighting. In the case of fluorescent or LED, the color of the lane is white. Will change. Similarly, the color of the lane changes according to each lighting change in the road environment of rainy weather and the road environment of dawn.

 On the other hand, the vision-based lane detection method using a boundary line as a feature has a problem in that the sharpness of the image decreases with increasing distance, so that the boundary line detection is difficult to be performed near the vanishing point.

 In order to solve this problem, an image sensor capable of obtaining high resolution data or a method of combining heterogeneous sensors with a long detection distance may be used.

210 醒 X297 誦 (Storage paper (2 types) 70g / m ² ) _

Electronic Document Usage for PCT

 Although the proposed method has been proposed, these methods have disadvantages of increasing implementation cost and difficulty in processing data in real time.

 [Detailed Description of the Invention]

 [Technical problem]

 In order to solve the problems of the prior art as described above, the present invention is to propose a real-time lane detection apparatus and method that can detect the lane accurately and quickly even in various lighting changes environment of the driving road.

 Other objects of the present invention may be derived by those skilled in the art through the following examples.

 [Technical Solution]

 According to an exemplary embodiment of the present invention, an area having a value corresponding to the brightness of a lane among brightness values in a color space of a color image is set as a candidate area for detecting a lane. A candidate region setting unit; And a lane determination unit configured to determine a lane in the candidate area by using the set straight line component of the candidate area.

 The apparatus may further include a color space converter configured to convert the color space of the color image into at least one channel of a YCbCr color space channel.

 The candidate region setting unit may set a region having a value greater than or equal to a threshold among brightness values of the Y channel of the color image as a candidate region for detecting a lane.

The threshold value is a value _obtained by multiplying the Y channel brightness value for the standard color of the white lane by _ηι % or when the plurality of pixels constituting the color image are accumulated in ascending order according to the Y channel brightness value. It may be a γ channel brightness value when the number exceeds the upper n ₂ %.

 The candidate region setting unit may set a candidate region for detecting lanes having a value less than or equal to a threshold value among brightness values of the Cb channel of the color image.

The threshold value is a value obtained by multiplying the Cb channel brightness value for the standard color of the yellow lane by the number or when the plurality of pixels constituting the color image are accumulated in ascending order according to the Cb channel brightness value. It may be the Cb null brightness value when the lower m ₂ % is exceeded.

 The apparatus may further include a binarized image generator generating a binarized image by using the set candidate region.

 The lane determination unit may determine the lane using a straight line component in the generated binarized image.

210 画 X297 腿 (Storage paper (2 types) 70g / m ² ) Electronic Document Usage for PCT

The lane determination unit may apply a Hough transform to the set candidate area, and determine a lane in the candidate area by using a plurality of straight lines generated by the Hough transform.

The lane determination unit may determine a lane in the candidate area by dividing the generated plurality of straight lines into left and right straight lines based on a slope ^' .

 A boundary extractor for extracting a boundary line from the color image; And a lane validator configured to verify the lane determined by the lane determiner using the extracted boundary line.

 The lane validator may set a search area within a predetermined range based on the determined lane, and verify the determined lane from the presence or absence of the extracted boundary line in the set search area.

 In addition, according to another embodiment of the present invention, the color space conversion unit for converting the color space of the color image to at least one channel of the YCbCr color space channel; And a candidate region setting unit configured to set a region having a value corresponding to the brightness of the lane among the brightness values of the converted color space channel as a candidate region for detecting the lane.

 In addition, according to another embodiment of the present invention, the color space conversion unit for converting the color space of the color image to the Y channel of the YCbCr color space channel; And a candidate region setting unit configured to set a region having a value greater than or equal to a threshold value among brightness values of the converted Y channel as a candidate region for detecting a lane.

 According to another embodiment of the present invention, the method may further include converting a color space of a color image into at least one channel of a YCbCr color space channel; And setting a region having a value corresponding to the brightness of the lane among the brightness values of the converted color space channels as a candidate area for detecting the lane.

 Advantageous Effects

 According to the present invention, there is an advantage that the lane can be detected accurately and quickly even in various lighting changing environments of the driving road.

 [Brief Description of Drawings]

 1 is a diagram schematically illustrating a process of detecting a lane by receiving an image from an image capturing apparatus according to an embodiment of the present invention.

 2A is a diagram illustrating a color path of white lanes under various light sources and brightness.

2101 腿 X297 腿 (Storage paper (2 types) 70g / m ² ) Electronic Document Usage for PCT

2B is a diagram showing a color path of yellow lanes under various light sources and brightness.

 3 is a diagram illustrating a detailed configuration of a real-time lane detection apparatus according to an embodiment of the present invention.

 4A is a diagram illustrating an input image according to an embodiment of the present invention. 4B is a diagram illustrating an image generated with a Y channel value converted by a color space converter according to an exemplary embodiment of the present invention.

 4C is a diagram illustrating an image generated with a Cb channel value converted by a color space converter according to an embodiment of the present invention.

 5 is the number of the plurality of stacked pixels is different according to an embodiment of the present invention

An area with a value greater than or equal to the Y channel brightness value when greater than 5% is shown.

 FIG. 6 shows Cb channel brightness characteristics of yellow in contrast to other colors. FIG.

 FIG. 7 illustrates an area having a value equal to or lower than the Cb channel brightness value when the number of accumulated plurality of pixels exceeds the lower 3% according to one embodiment of the present invention.

 8A is a diagram illustrating a candidate region for detecting a white lane binarized by a binarization image generating unit according to an embodiment of the present invention.

 FIG. 8B is a diagram illustrating a candidate region for detecting a yellow lane binarized by a binarization image generating unit according to an embodiment of the present invention.

 8C is a diagram illustrating a binarization image in which the binarization images of FIGS. 8A and 8B are combined, according to an embodiment of the present invention.

 9A illustrates a binarized image generated by a binarized image generating unit according to an embodiment of the present invention.

 9B is a diagram illustrating a result of applying a Hough transform to a binarized image according to an embodiment of the present invention.

 10 is a diagram illustrating a process of dividing a plurality of straight lines into left and right straight lines using a slope according to an exemplary embodiment of the present invention.

 11A is a diagram illustrating a process of grouping left and right straight lines by the lane determination unit according to an exemplary embodiment of the present invention.

 FIG. Lib is a diagram illustrating a lane determined by a lane determination unit according to an exemplary embodiment of the present invention. 12 illustrates a search area set based on a determined lane according to an embodiment of the present invention.

FIG Ba is a view showing an ^'image regarding the extracted boundary by the boundary extraction unit in accordance with one embodiment of the present invention.

210 薩 X297 画 (Storage paper (2 types) 70g / m ² ) Electronic Document Usage for PCT

 FIG. 13B is a diagram showing a boundary line existing in a search area according to an embodiment of the present invention. FIG.

 14 is a flowchart illustrating a real-time lane detection method according to an embodiment of the present invention over time.

 [Description of the code]

 10: imaging device 100 real-time lane detection cabinet

 110: color space conversion unit 120 candidate area setting unit

 130: binarization image generation unit 140 lane determination unit

 150: boundary extractor 160 lane verification unit

 162: determined lane 164 search area

 [Form for implementation of invention]

 As the invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention. In describing the drawings, similar reference numerals are used for similar elements.

 Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 illustrates a real-time lane detection apparatus 100 according to an embodiment of the present invention inputting a color image from an imaging apparatus 10. It is a figure which shows typically the process of receiving a lane detection.

 Referring to FIG. 1, the real-time lane detecting apparatus 100 according to the present invention is based on a vision-based lane detecting method, and may receive a color image of front information of a vehicle from an imaging device 10 mounted on a vehicle. (Hereinafter referred to as an 'input image'), the lane may be detected by analyzing the same.

 In the present invention, for convenience of description, it is assumed that the imaging device 10 exists outside the real-time lane detection apparatus 100, but the present invention is not limited thereto.

 The real-time lane detection apparatus 100 according to the present invention uses both the boundary line and the color as a feature for lane detection.

That is, the real-time lane detecting apparatus 100 may set a candidate area ^' that is predicted to have a lane in the input image using color and determine a lane, and additionally verify the lane by applying boundary information to the determined lane.

210 誦 X297 薩 (Storage paper (2 types) 70g / m ² ) Electronic Documents Available PCT

: The attached sheet 43th form].

 In more detail, the candidate region setting unit 120 and the binarization image generator 130 of the real-time lane detecting apparatus 100 analyze the color information of the input image to generate a binarized image with emphasis on the lane information, and the lane determination unit ( 140 determines the lane therefrom. The boundary extractor 150 and the lane validator 160 verify the determined lane using the boundary information.

 On the other hand, as described above, since the vision-based lane detection method using color is sensitive to a change in illumination, that is, a change in the light source, the real-time lane detection apparatus 100 according to the present invention uses YCbCr instead of RGB color space as color information. Use color space.

 The RGB color space represents an arbitrary color by using a combination of three colors of red (R), green (G), and blue (B), which are three primary colors of light, and a plurality of pixels constituting the input image are generally RGB. It is expressed in color space. In addition, the YCbCr color space considers that a human eye is more sensitive to brightness than color, and refers to a color expression method based on luminance Y, blue intensity Cb, and red intensity Cr.

 That is, as shown in FIG. 2A, the white lane always has a higher brightness value than the other colors in various light sources (A, Cool, Daylight, Horizon, TL84) and brightness, so that the YCbCr color space for the white lane The Y channel brightness value, that is, the intensity may be formed to be constantly high unlike other colors.

 In addition, as shown in FIG. 2B, the yellow lanes represent various light sources A, Cool,

Daylight, Horizon, and TL84) and brightness always have a lower blue intensity than other colors, so the Cb channel brightness value, ie intensity, in the YCbCr color space for the yellow lane will be consistently low, unlike other colors. Can be.

 Therefore, the real-time lane detection apparatus 100 according to the present invention uses the Y channel brightness value for the extraction of the white lane from the color image, and uses the Cb channel brightness value for the extraction of the yellow lane.

In short, the real-time lane detection apparatus 100 according to an embodiment of the present invention extracts a white lane using a high γ channel brightness value of a white lane and uses a low Cb channel brightness value of a yellow lane. By extracting the yellow lanes, the lanes can be accurately detected even in an environment where the light source changes. In addition, the real-time lane detection apparatus 100 according to an embodiment of the present invention uses the brightness characteristics in the Y channel of the white lane and the brightness characteristics of the Cb channel of the yellow lane, as described above. brightness, an area having a value in ilreul typically set by the candidate area for the detection of a white lane, and the ^'yellow lane areas uniformly with a Cb-channel brightness value equal to or less than a threshold value

210 画 X297 醒 (Storage Paper (2 Types) 70g / n) Electronic Document Usage for PCT

 By setting it as a candidate area for detection, the lane can be detected more quickly.

 In addition, the Y channel brightness value and the Cb channel brightness value from the RGB image may be quickly obtained through a simple conversion process, and may be used for lane detection. Meanwhile, in the present invention, the real-time lane detection apparatus 100 will be described as detecting the lane using the YCbCr color space, but the present invention is not limited thereto. It will be apparent to one skilled in the art that other color space channels may also be used that may maintain the < RTI ID = 0.0 >

 As described above, according to the present invention, the lane can be accurately and quickly detected even in various lighting changes. Hereinafter, the real-time lane detecting apparatus 100 according to the present invention will be described in detail with reference to FIGS. 3 to 13. 3 is a diagram showing a detailed configuration of the real-time lane detection apparatus 100 according to an embodiment of the present invention.

A, ttaron real-time lane detection apparatus 100 includes a color space conversion unit 110, a candidate area setting unit 120, a binary image generator ⁽¹³ 0) to one embodiment of the present invention as shown in Figure 3, It may include a lane determination unit (MO), a boundary line extraction unit 150 and a lane verification unit (160).

 First, the color space converter 110 according to an embodiment of the present invention may convert the color space for each of the plurality of pixels constituting the input image into at least one channel of the YCbCr color space.

 As described above, since a plurality of pixels constituting the input image is generally represented by an RGB color space, the color space converter 110 converts the RGB color space into a Y channel value or a Cb channel value according to the following equation. I can convert it.

 [Equation 1]

 Here, Y denotes a channel value, R, G, and B represent each channel value in the RGB color space, and 0.29900, 0.58700, and 0.11400 denote coefficients applied to the R, G, and B channels, respectively.

 [Equation 2]

 Cb = ᅳ 0.16874? ᅳ 0.33126 0.50000 ^ where Cb is the Cb channel value, -0.16874, -0.33126, 0.50000 is the R, G, B channel

210 誦 X297 瞧 (Storage paper (2 types) 70g / m ^! ) Use of electronic documents; ^ ^{7 "} PCT

 [Appendix 43] The coefficients to be applied are each applied. Equations 1 and 2 represent examples for converting an RGB color space into Y channel values and Cb channel values, and various algorithms for channel value extraction may exist.

 According to an embodiment of the present invention, the Y channel value is zero at

Since Y has a value up to 255 and Y is a value related to the average brightness of RGB, the Y channel value may be used as the Y channel brightness value in the candidate region setting unit 120 described below.

 In addition, since the Cb channel value has a value from -127.5 to 127.5 according to Equation 2, and Cb is a value related to the blue intensity, the Cb channel value is the Cb channel brightness value in the candidate region setting unit 120 described below. To be used as, 127.5 can be added and used as a value from 0 to 255.

 4 illustrates an input image and an image generated by channel values converted by the color space converter 110 according to an embodiment of the present invention. FIG. 4A illustrates an input image and FIG. 4B illustrates Y channel values. FIG. 4C is a diagram showing images generated with Cb channel values, respectively.

 Meanwhile, according to another exemplary embodiment of the present invention, color space conversion by the color space conversion unit 110 corresponds to a data processing process, and may be performed by the candidate area setting unit 120 to be described below. That is, the candidate region setting unit 120 may convert the color space of the input image into at least one of the YCbCr channels, and set a candidate region for lane detection, as described below.

 The candidate region setting unit 120 according to the present invention sets a region having a value corresponding to the brightness of the lane among the brightness of the channel converted by the color space converter 110 as a candidate region for detecting the lane.

 In more detail, the candidate region setting unit 120 according to an embodiment of the present invention may identify a region having a value greater than or equal to a threshold among the Y channel brightness values in the YCbQ color space with respect to the input image, as a candidate region for detecting white lanes. Can be set with That is, the candidate region setting unit 120 recognizes that the Y channel brightness value of the white lane is always formed high, and sets the region having a threshold value or more among the Y channel brightness values of the input image as the candidate area. The set candidate area may be used for detecting the white lane.

 In this case, the threshold value refers to a Y-channel brightness value in which the white lane can be distinguished from other colors. According to an embodiment of the present invention, the threshold value may be relatively determined from the brightness value of the input image.

 White lanes are displayed relatively brightly on bright input images and relatively brightly on dark input images.

210 画 X297 誦 (Storage paper (2 types) 70g / m ^! ) Electronic Document Usage for PCT

 The threshold value should be determined relatively from the brightness value of the input image.

 Accordingly, the candidate region setting unit 120 according to an embodiment of the present invention may allow the threshold value to be relatively determined according to the input image by using the number of pixels accumulated according to the Y channel brightness value.

 That is, when the candidate region setting unit 120 accumulates in ascending order according to the plurality of pixel c channel brightness values constituting the input image, the candidate region setting unit 120 may set the Y channel brightness value when the accumulated number of picks exceeds the upper n%. The threshold value is set, and an area having a random value or more is set as a candidate area.

 For example, the candidate region setting unit 120 detects a white lane in an area having a value greater than or equal to the Y channel brightness value when the number of accumulated plurality of pixels among the Y channel brightness values of the input image exceeds the upper 5%. It can be set as a candidate region for. 5 is the Y channel brightness. The vertical axis represents the number of pixels according to the Y channel brightness value, and the area indicated by the circle is when the number of accumulated pixels exceeds the upper 5% according to an embodiment of the present invention. A region having a value greater than or equal to the Y channel brightness value is shown.

 As shown in FIG. 5, an area having a value greater than or equal to the Y channel brightness value when the cumulative number of pixels causes the top 5% is an area having a Y channel brightness value that can be distinguished from other colors. At the same time, since the area is relatively determined according to the brightness value of the input image, when the candidate area is set according to an embodiment of the present invention, the white lane can be accurately detected even under various lighting changes. In addition, as shown in FIG. 5, since all the region cells having the Y channel brightness value greater than or equal to the threshold value correspond to candidate regions for detecting the white lane, additionally in setting the candidate region for the value Y for lane detection Since no calculation is required, the white lane can be detected more quickly according to the present invention.

On the other hand, according to another embodiment of the present invention to simplify the lane detection process _{: the} threshold value may be set absolutely from the standard color of the white lane.

 That is, in a driving environment in which the change of the light source is expected to be constant, the threshold value may be preset from the Y channel brightness value for the standard color of the white lane, and the area having the Y channel brightness value over the preset threshold value is selected as the candidate area. By setting, the white lane can be detected more quickly.

 For example, 204 corresponding to 80% of the Y channel value 255 for the standard color sRGB (255, 255, 255) of the white lane is set as a threshold, and an area having a Y channel brightness value of 204 or more is a candidate region. Can be set.

 Subsequently, the candidate region setting unit 120 according to an exemplary embodiment of the present invention selects a value less than or equal to a threshold value among the Cb channel brightness values in the YCb color space for the input image.

210 匪 X297 画 (Storage paper (2 types) 70g / m ² ) Electronic Documents Available PCT

 The region having the attached sheet 43 can be set as a candidate region for detecting yellow lanes.

 That is, as shown in FIG. 6, the candidate region setting unit 120 recognizes that the Cb channel brightness value for the yellow lane is always low, and thus the value below the threshold value among the Cb channel brightness values for the input image. It is also possible to set an area having a to be a candidate area, and use the set candidate area to detect yellow lanes.

 In this case, as described above, the threshold value refers to a Cb channel brightness value in which the yellow lane can be distinguished from other colors, and may be relatively determined from the brightness value of the input image. Uses the number of pixels accumulated according to the Cb channel brightness value.

 That is, when the plurality of pixels constituting the input image are accumulated in ascending order according to the Cb channel brightness value, the candidate region setting unit 120 according to an embodiment of the present invention accumulates the lower number of m%. The Cb channel brightness value when exceeding is regarded as a threshold value, and an area having a value below the threshold value is set as a candidate area. For example, the candidate region setting unit 120 may display an area having a value less than or equal to the Cb channel brightness value when the number of accumulated pixels among the Cb channel brightness values for the input image exceeds the lower 3%. It can be set as a candidate region for detection. The horizontal axis of FIG. 7 represents the Cb channel brightness value, and the vertical axis represents the number of pixels according to the Cb channel brightness value, respectively. The area indicated by the circle is the number of accumulated pixels according to an embodiment of the present invention. Shows an area with a value less than or equal to the Cb channel brightness value when is greater than the lower 3%.

 As shown in FIG. 7, an area having a value less than or equal to the Cb channel brightness value when the number of accumulated pixels exceeds the lower 3% is an area having a Cb channel brightness value that can be distinguished from other colors. At the same time, since the region is relatively determined according to the brightness value of the input image, when the candidate region is set according to an embodiment of the present invention, the yellow lane may be accurately detected even under various lighting changes.

In addition, as in the case of the white lane detection, all regions having a Cb channel brightness value below the threshold value correspond to candidate regions for detection of the yellow lane _: the candidate region for lane detection can be set quickly, and as a result, Yellow lanes can be detected more quickly. ^■

 And, according to another embodiment of the present invention, the threshold may be set absolutely from the standard color of the yellow lane, for example, the Cb channel brightness value 29 for the standard color sRGB 254, 193, 16 of the yellow lane. 23, which corresponds to 80% of the value, is set as a threshold value, and an area having a Cb channel brightness value of 23 or less may be directly set as a candidate area.

210 腿 X297 醒 (Storage paper (2 types) 70g / m ^! ) Electronic Document Usage for PCT

 For the convenience of explanation, the above description has been made on the assumption that the number of accumulated plurality of pixels is set to the upper 5% and the lower 3%, respectively, but this is only an example, and more accurate lane detection is performed. It will be apparent to those skilled in the art that for the sake of brevity, they may be set to the top 3% and the bottom 1%, respectively.

 In addition, although it has been described assuming that the input image includes both the white lane and the yellow lane, according to another embodiment of the present invention, the input image may include only the white lane, and in this case, the candidate region setting unit 120 ) May set both the candidate region for the detection of the white lane and the candidate region for the detection of the yellow lane.

 Next, the binarized image generating unit 130 according to an embodiment of the present invention may generate the binarized image by using the candidate region set by the candidate region setting unit 120.

 More specifically, the binarization image generating unit 130 displays a plurality of pixels constituting the candidate region set by the candidate region setting unit 120 as 1 and a plurality of pixels constituting the other region as 0. Can be generated.

 That is, the binarization image generating unit 130 according to an embodiment of the present invention, the pixel having a value greater than or equal to a threshold value among the Y channel brightness values for each of the plurality of pixels constituting the input image is 1, and the other pixels are 0. By setting to, it is possible to generate a binarized image for detecting all white lanes.

 In addition, the binarization image generating unit 130 according to an embodiment of the present invention, the pixel having a value less than or equal to the threshold value among the Cb channel brightness values for each of the plurality of pixels constituting the input image is 1, and the other pixels are Set to 0 to generate a binarized image for detecting the yellow lane.

In addition, the binarization image generating unit 130 according to an embodiment of the present invention combines each binarization image generated as described above into one binarization image, thereby referring to the vehicle by the lane determination unit 140 to be described below. The left lane and the right lane to be determined can be determined. ^'

 That is, the binarization image generator 130 may combine the candidate region of the white lane having the brightness characteristic in the different channels and the candidate region of the yellow lane into one image, and as a result, the left lane based on the vehicle And the right lane may be determined and used as information for driving the vehicle.

 8 illustrates a binarized image generated by the binarized image generator 130 according to an embodiment of the present invention. FIG. 8A illustrates a binarized image of a candidate region for detecting a white lane, FIG. 8B illustrates a binarized image of a candidate region for detecting a yellow lane, and FIG. Sc illustrates a binary image of FIG. 8A as a premise for determining a lane.

210腿X297腿(retention sheet ⁽² types) 70g / m ²⁾ Electronic Documents Available PCT Applications ·

— [Annex 43] Each of the images combined with the binarized image of Fig. 8B is shown.

 As illustrated in FIG. 8C, the binarization image generating unit 130 of the present invention may generate a binarized image for each candidate region for a white lane having a brightness characteristic and a candidate region for a yellow lane in different channels. .

Subsequently, the lane determination unit 140 according to an embodiment of the present invention may determine the lane by extracting the ^' normal ^' component from the candidate area set by the candidate area setting unit 120.

 In this case, since the candidate region is generated as the binarized image by the binarization image generating unit 130 as described above, the lane determination unit 140 according to the present invention determines a lane by extracting a linear component from the binarization image.

 First, the lane determiner 140 according to an embodiment of the present invention may use a Hough transform to extract a linear component.

 In this case, the Hough transform is an algorithm for finding a straight line that satisfies a specific condition among a set of random points on a plane, and the tacking condition is a line where the points exist when a certain number of points exist on an arbitrary line. It may be determined as.

 FIG. 9 is a diagram illustrating a result of applying a Hough transform to a binarized image according to an embodiment of the present invention. As shown in FIG. 9, a region of interest of a binarized image may be represented by a plurality of straight lines by the Hough transform. have.

 Next, the lane determination unit 140 according to an embodiment of the present invention analyzes a plurality of straight lines transformed by the Hough transform so that candidate regions having no characteristics as lanes are proposed in the lane determination process. At the same time, the left lane and the right lane based on the vehicle can be determined as the driving information. More specifically, the lane determination unit 140 divides the plurality of converted straight lines to the left and right, groups the plurality of straight lines divided to the left and right by .region, and then selects the area having the most straight lines in each grouped area. Decide on your lane

 To this end, the lane determination unit 140 first divides the plurality of straight lines converted by the Hough transform into left and right straight lines based on a sloping machine.

 For example, as illustrated in FIG. 10, the lane determination unit 140 applies coordinate axes X and y to the image shown in FIG. 9B, and an angle formed by a normal line lowered from a plurality of straight lines at the center point of the coordinate axis to the y axis. By using, a plurality of straight lines can be divided into left and right straight lines.

For example, if the angle Θ formed by the y-axis is 10 ^° <θ <80 ^° , the straight line exists on the left side of the vehicle. When 100 ^° <θ <170 ^° , the straight line exists on the right side of the vehicle. It can be divided into

210 匪 Χ297 醒 (Storage paper (2 types) 70g / m ^J ) Using Electronic Documents 7 ^ ^" PCT Applications

Next, as shown in FIG. 11A, the lane determination unit 140 groups the divided left and right straight lines into regions based on the slope and the starting point of the X-axis, according to one embodiment of the present invention. According to, the left straight line may be grouped into one region, and the right straight line may be grouped into two regions, respectively.

 Finally, the lane determination unit 140 determines the area including the most straight lines in each grouped area as the lane. As shown in FIG. Lib, according to an embodiment of the present invention, since the left straight line is one area, it may be immediately determined as a lane, and the right straight line may be determined as a lane as an area including many straight lines among the two areas. .

 As such, the candidate determining region or the binarized image including the information about the lane may be determined as the lane as the driving information by the lane determining unit 140 according to the exemplary embodiment of the present invention, and the area having no characteristic as the lane may be determined. It may be excluded from the suboptimal decision process.

 For example, according to another exemplary embodiment of the present disclosure, the candidate region setting unit 120 may set a region having a brightness value characteristic of a yellow lane as a candidate region even in an input image including only white lanes, and the region is determined as a lane. The unit 140 may be excluded from the lane detection process.

 Meanwhile, in the above, it is assumed that the lane determining unit 140 according to an embodiment uses the Hough transform as an algorithm for extracting the linear component, but the present invention is not limited thereto. Can use various algorithms for the extraction of linear components.

 In addition, according to one embodiment of the present invention, the lane determined by the lane determination unit 140 may be used directly as a data for detecting a lane in an arbitrary input image, and the boundary line extracting unit 150 and the lane described below Lane detection by the verification unit 160 corresponds to a process of verifying whether lane determination from a candidate area has been correctly performed. .

 Prior to describing the boundary extractor 150, the lane verification unit 160 will be described first for convenience of description.

 The lane verification unit 160 according to an exemplary embodiment sets a search area within a predetermined range based on the lane determined by the lane determination unit 140, and uses the boundary line existing in the set search area to set a lane. Can be verified. In more detail, the lane validator 160 sets a search area within a range of 10 to 20 pixels in the left and right directions based on the lane determined by the lane determiner 140, and determines whether a boundary line exists within the search area. Validate it. In this case, the lane verification unit 160 determines the slope of the boundary line and the determined lane and the X coordinate axis.

210 薩 x ² 97 mill (Storage paper (2 types) 70g / m ² ) Electronic Documents Available PCT

 The value of Annex 43 may be additionally used.

 For example, when the boundary line does not exist in the search area, the lane verification unit 160 processes the lane determined by the lane determination unit 140 as invalid, and when the boundary line exists in the search area, the lane determination unit ( The lane determined by 140 may be treated as valid /

Even when the boundary line ^' is present in the search area, the determined lane may be treated as invalid when the slope of the lane determined by the lane determination unit 140 and the slope of the boundary show a large difference.

 In addition, when only a part of the boundary line exists in the search area, the determined lane may be treated as valid only when the determined slope and the X coordinate axis value are within a certain error range with the boundary line.

 In this case, 10 to 20 pixels are only an example, and the search area may be narrower to improve the accuracy of lane detection.

 Meanwhile, the boundary line may be extracted by the boundary line extracting unit 150 according to an embodiment of the present invention, and the boundary line extracting unit 150 may use Canny edge detection as an algorithm for extracting the boundary line from the input image. Can be.

 12 illustrates a search area 164 set up based on the determined lane 162 according to one embodiment of the invention.

 FIG. 13A illustrates an image of a boundary line extracted by the boundary line extracting unit 150 according to an embodiment of the present invention, and FIG. 13B illustrates a region of the search area 164 according to an embodiment of the present invention. Show the border.

 As shown in FIG. 13, the lane validator 160 processes the lane 162 determined by the lane determiner 140 as valid when a boundary line exists in the search area 164.

 Meanwhile, according to another exemplary embodiment of the present disclosure, the lane validator 160 may remove the boundary line existing outside the search area 164 from the boundary image, and remove the boundary line existing in the search area 164 as one straight line. By combining, additional information may be provided for detection of the lane.

 That is, according to another embodiment of the present invention, the detection of the lane may be made from the average value of the boundary line existing in the search area 164.

 As described above, the real-time lane detecting apparatus 100 according to the present invention may accurately detect a lane even in an environment where a light source changes as compared with other vision-based lane detecting apparatuses. ·

 In addition, the real-time lane detection apparatus 100 according to the present invention, as described above, has an area having a brightness value above the threshold and a brightness value below the threshold.

210 匪 X297 腿 (Storage paper (2 types) 70g / m ² ) By setting the area as a lane candidate area, the lane can be detected more quickly, and the process of acquiring the brightness value from the color space conversion process is also faster

Can be processed and used for the detection of lanes. 14 is a diagram of a real-time lane detection method according to an embodiment of the present invention.

This is a flow chart showing the flow.

 As shown in FIG. 14, a real-time lane according to an embodiment of the present invention.

The detection method converts the color space (S1410) to set a candidate area.

In operation S1420, generating a binarized image in operation S1430, determining a lane in operation S1440,

Extracting a boundary line (S1450) and verifying a lane (S1460)

have.

 First, in operation S1410, the color space of the input image is converted into a color space.

 At least one of the YCbCr color space channels may be converted.

 According to an embodiment of the present invention, in step S1410, the RGB color of the input image,

Space can be converted to Y channel value and Cb channel value. In this case, the Y channel value is

Can be used as the Y channel brightness value in step S1420, and the Cb channel value is

Can be used after 127.5 is added to have a value from 0 to 255.

 Next, in step S1420 of setting the candidate area, in step S1410

The area of the brightness of the converted channel whose value corresponds to the brightness of the lane

It is set as a candidate area for detection.

 More specifically, in step S1420 according to an embodiment of the present invention, the force of the first-in, first-out

White difference between the Y-channel brightness of the image

As a candidate area for detection, the set area is detected as a white lane

Can be used.

 In this case, the threshold is the Y channel brightness in which the white lane can be distinguished from other colors.

As a value, it can be determined relatively from the brightness value of the input image.

have.

 In addition, in step (S1420) according to an embodiment of the present invention Cb of the input image

For the detection of the yellow lane, the area having the value below the threshold among the channel brightness values

By setting as a candidate area, the set area can be used for detection of yellow lanes.

have.

 Similarly, the threshold is Cb, where the yellow lane can be distinguished from other colors.

Meaning the channel brightness value, which is relatively from the brightness value of the input image.

Can be determined.

 Subsequently, in step S1430 of generating a binarized image, in step S1420

210 H1H1X297 翻 (Storage paper (2 types) 70g / irf) Electronic Documents Available PCT

 [Annex 43] A binarized image is generated using the set candidate region.

 More specifically, in step S1430 according to an embodiment of the present invention, binarization expresses a plurality of pixels constituting the candidate region set in step S1420 as 1 and a plurality of pixels constituting the other region as 0. An image can be generated. That is, in step S1430 according to an embodiment of the present invention, among the Y channel brightness values of each of the plurality of pixels constituting the input image, a pixel having a value greater than or equal to a threshold value is set to 1 and other pixels are set to 0. A binarization image for detecting a white lane may be generated.

 In operation S1430, a pixel having a value less than or equal to a threshold value among the Cb channel brightness values of each of the pixels constituting the input image is set to 1 and other pixels are set to 0. A binary image for detecting the yellow lane may be generated.

 In operation S1430, the candidate region and the yellow lane of the white lane having brightness characteristics in different channels are combined by combining the binarized images generated as described above into one binarized image. The candidate region for the is represented as one image, and the left lane and the right lane based on the vehicle may be determined in operation S164 described below.

 In the determining of the lane (S1440), the lane may be defected using a straight line component in the binarization image generated in the operation S1430.

At this time, according to an embodiment of the present invention _. In operation S1440, a Hough transform may be used to extract the linear components.

Then, the left of the vehicle all criteria as, and at the same time driving information such that the step (S1440) in the by analyzing a plurality of straight lines transformed by the Hough transformation having the lane as a characteristic candidate regions are ^"not excluded from the decision-making process of the lane the lane And the right lane can be determined.

 That is, the plurality of converted straight lines are divided into left and right, the plurality of straight lines divided by left and right are grouped by area, and the area having the most straight lines in each grouped area is determined as a lane.

 Finally, in the step of extracting the boundary line (S1450), the boundary line is extracted from the input image, and in the verifying lane (S1460), the boundary image extracted in the step (S1450) and the lane determined in the step (S1440) are input image. The lane can be verified at.

 The embodiments of the real-time lane detection method according to the present invention have been described so far, and the configuration of the real-time lane detection apparatus 100 described above with reference to FIGS. Thus, the more detailed description

210 匪 x ² 97 mm (Storage paper (2 types) 70g / n) Electronic Document Usage for PCT

 Annex No. 43 will be omitted.

Embodiments of the present invention is implemented in program instruction form that can be executed by various computer ^means, it may be recorded in computer-readable media. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. -Optical media (magneto-optical), and names (ROM) _: Examples of program instructions such as RAM, flash memory, etc., may be executed by a computer using an interpreter as well as machine code such as produced by a compiler. Contains high-level language codes. The hardware device described above may be configured to operate as at least one software module to perform the operations of one embodiment of the present invention, and vice versa. As described above, the present invention has been described by specific embodiments such as specific components and the like, but the embodiments and the drawings are provided only to assist the overall understanding of the present invention, and the present invention is not limited to the above embodiments ^. In other words, various modifications and variations are possible to those skilled in the art to which the present invention pertains. Accordingly, the spirit of the present invention should not be limited to the described embodiments, and all of the equivalents and equivalents of the claims, as well as the appended claims, will fall within the scope of the present invention.

210 薩 X297 腿 (Storage paper (2 types) 70g / m ² )

Claims

Electronic document use PCT use [attached sheet 43 form] [claim]

[1] claims ^"

 A candidate region setting unit configured to set a region having a value corresponding to the brightness of the lane among the brightness values in the color space of the color image as a candidate region for detecting the lane; And

 A lane determination unit that determines a lane in the candidate area using the set straight line component of the candidate area

 Real-time lane detection apparatus comprising a.

 [Claim 2]

 The method of claim 1,

 And a color space converting unit converting the color space of the color image into at least one of the YCbCr color space channels.

 [Claim 3]

 According to claim 12,

 The candidate region setting unit

And a region having a value greater than or equal to a threshold among brightness values of the Y channel of the color image as a candidate region for detecting a lane. ^'

 [Claim 4]

 The method of claim 3,

The threshold value is a value obtained by multiplying the Y channel brightness value for the standard color of the white lane by the number or when the plurality of pixels constituting the color image are accumulated in ascending order according to the Y channel brightness value. A real-time lane detection device, characterized in that the γ-channel brightness value when the upper n ₂ % exceeds.

[Claim 5]

 The method of claim 2,

 The candidate region setting unit

 And an area having a value equal to or lower than a brightness increase value threshold value for the Cb channel of the color image as a candidate area for detecting a lane.

 [Claim 6]

 The method of claim 5,

 The threshold is the Cb channel for the standard color of the yellow lane. The brightness value is multiplied by 1%, or a plurality of pixels constituting the color image are Cb channels.

210 匪 X297 匪 (Storage paper (2 types) 70g / m ² ) Use of electronic documents Ϊ ¥ ^~ PCT

[Appendix 43] The real-time lane detecting apparatus according to claim 43, wherein the accumulated number of pixels is a Cb channel brightness value when the number of accumulated pixels exceeds the lower m ₂ %.

 [Claim 7]

 The method of claim 1,

 And a binarization image generating unit generating a binarization image by using the set candidate region.

 [Claim 8]

 The method of claim 7,

 The lane determination unit

 And determining the lane using a linear component of the generated binarized image.

 [Claim 9]

 The method of claim 1,

 The lane determination unit

 And applying a Hough transform to the set candidate region, and determining a lane in the candidate region using a plurality of straight lines generated by the Hough transform.

 [Claim 10]

 The method of claim 9,

 The lane determination unit

 And determining a lane in the candidate area by dividing the generated plurality of straight lines into left and right straight lines based on a slope.

 [Claim 11]

 The method of claim 1,

 A boundary extractor for extracting a boundary line from the color image; And

 A lane verification unit that verifies the lane determined by the lane determination unit using the extracted boundary line

 Real-time lane detection apparatus further comprising a.

 [Claim 12]

 The method of claim 11,

 The lane verification unit

 A search area is set within a predetermined range based on the determined lane, and from the presence or absence of the extracted boundary line in the set search area.

210 mfflX297 醒 (Storage paper (2 types) 70g / m ² ) Electronic Document Usage for PCT

 [Annex 43] The real-time lane detection apparatus characterized by verifying the determined lane.

[Claim 13]

 A color space converter for converting the color space of the color image into at least one channel of the YCbCr color space channel; And

 A candidate region setting unit configured to set a region having a value corresponding to the brightness of the lane among the brightness values of the converted color space channels as a candidate region for detecting the lane;

 Real-time lane detection apparatus comprising a.

 [Claim 14]

 The method of claim 13,

 The candidate region setting unit

 Setting an area having a value greater than or equal to a threshold among brightness values of the converted Y channel as a candidate area for detecting a lane;

 The threshold value is a value obtained by multiplying the Y channel brightness value for the standard color of the white lane by%, or when the plurality of pixels constituting the color image are accumulated in ascending order according to the Y channel brightness value. Is a Y-channel brightness value when is greater than the upper%.

[Claim 15]

 The method of claim 13,

 The candidate region setting unit

 A region having a value less than or equal to a threshold value among brightness values of the converted Cb channel is set as a candidate region for detecting a lane;

The threshold value is a value obtained by multiplying the Cb channel brightness value for the standard color of the yellow lane by m ^< ¾ or when the plurality of pixels constituting the color image are accumulated in ascending order according to the Cb channel brightness value. And a Cb channel brightness value when the number of m exceeds a lower m ₂ %.

 [Claim 16]

 A color space converter for converting the color space of the color image into the Y channel among the YCbG color space channels; And

 And a candidate region setting unit configured to set a region having a value equal to or greater than a threshold value among the intensity values of the converted Y channel as a candidate region for detecting a lane.

 [Claim 17]

 The method of claim 16,

210 薩 X297 誦 (Storage paper (2 types) 70g / m ² ) Electronic Documents Available PCT

[Attachment 43] The threshold value is a value _obtained by multiplying the Y channel brightness value for the standard color of the white lane by _ηι % or by collecting a plurality of pixels constituting the color image in ascending order according to the Y channel brightness value. And the? Channel brightness value when the number of accumulated plurality of pixels exceeds the upper n ₂ %.

[Claim 18]

Converting a color space of a color image into at least one channel of a _YCbCr color space channel; And

 And setting an area having a value corresponding to the brightness of the lane among the brightness values of the converted color space channel as a candidate area for detecting the lane.

 [Claim 19]

 The method of claim 18,

 The candidate region setting step

 A region having a value greater than or equal to a threshold value among brightness values of the converted Y channel is set as a candidate region for lane detection;

The threshold is a value obtained by multiplying the Y channel brightness value for the standard color of the white lane by _ηι % or when the plurality of pixels constituting the color image are accumulated in ascending order according to the Y channel brightness value. Real-time lane detection method characterized in that the γ channel brightness value when the number exceeds the upper n ₂ % ᅳ

[Claim 20]

 The method of claim 18,

 The candidate region setting step

 A region having a value less than or equal to a threshold value among brightness values of the converted Cb channel is set as a candidate region for detecting a lane;

The threshold is a value obtained by multiplying the Cb channel brightness value for the standard color of the yellow lane by _mi % or when the plurality of pixels constituting the color image are accumulated in ascending order according to the Cb channel brightness value. Real-time lane detection method characterized in that the Cb channel brightness value when the number exceeds the lower m ₂ %

210 匪 X297 mm (Storage paper (2 types) 70g / m ² )