JP6648553B2 - White line recognition device - Google Patents

Description

  The present invention relates to a technical field of a white line recognition device that recognizes a white line on a road from an image.

  As an apparatus of this type, for example, a luminance difference between a pixel to be determined as to whether or not it is an edge of a white line and a pixel adjacent to the pixel by 1 to K pixels in the scanning direction from this pixel is calculated, and the calculated difference is calculated. An apparatus has been proposed in which, when at least one of them is equal to or larger than a threshold value, a pixel to be determined is set as the edge of a white line (see Patent Document 1).

JP 2012-018531 A

  By the way, when an image of a road illuminated by strong sunlight is captured, the brightness value of a white line portion on the road in the captured image may become the upper limit saturation value (so-called overexposure). In this case, since the brightness value of the asphalt portion of the road is relatively large, the difference between the brightness value of the white line portion and the brightness value of the asphalt portion is relatively small. Alternatively, when a shaded road is picked up, the brightness value of the asphalt portion in the picked-up image may become the lower limit saturation value (so-called black crush). In this case, since the luminance value of the white line portion is also relatively small, the difference between the luminance value of the white line portion and the luminance value of the asphalt portion is relatively small.

  In such a case, the technique described in Patent Document 1 has a technical problem that the edge of the white line may not be correctly recognized.

  SUMMARY An advantage of some aspects of the invention is to provide a white line recognition device that can appropriately recognize a white line.

  In order to solve the above-described problem, a white line recognition device of the present invention is a white line recognition device that recognizes a white line on a road based on an image obtained by capturing an image of the outside of a vehicle. Extracting means for extracting an adjacent area adjacent to the area; and a condition that a difference between a luminance value of the extracted white line candidate area and a luminance value of the extracted adjacent area is larger than a threshold value. A recognition unit for recognizing as a white line, and when the first pixel value that is the pixel value of the extracted white line candidate region is the upper limit saturation value, the threshold is set to the first pixel value that is not the upper limit saturation value. If the second pixel value that is the pixel value of the extracted adjacent region is the lower limit saturation value, the threshold value is compared with the case where the second pixel value is not the lower limit saturation value. Setting means for reducing Provided.

  According to the white line recognition device of the present invention, when the first pixel value relating to the white line candidate region is the upper limit saturation value, or when the second pixel value relating to the adjacent region is the lower limit saturation value, the first pixel value Is not the upper limit saturation value, or the threshold value is smaller than when the second pixel value is not the lower limit saturation value. For this reason, the white line recognition device can appropriately recognize a white line even when so-called overexposure or crushed black occurs.

  Examples of the “first pixel value” and the “second pixel value” according to the present invention include, for example, a luminance value and a color value, but are not limited thereto. Other parameters may be used.

  The operation and other advantages of the present invention will become more apparent from the embodiments explained below.

It is a block diagram showing the composition of the white line recognition device concerning an embodiment. 9 is a flowchart illustrating a white line recognition process according to the embodiment. It is an example of the change of the brightness value of the white line part and the asphalt part of the road.

  An embodiment according to a white line recognition device of the present invention will be described with reference to the drawings.

(Overview of white line recognition device)
An outline of a white line recognition device according to the embodiment will be described with reference to FIG. FIG. 1 is a block diagram illustrating a configuration of a white line recognition device according to the embodiment.

  In FIG. 1, a white line recognition device 10 includes an image input unit 11, a white line candidate region extraction unit 12, a logical processing block or a physical processing circuit implemented therein to execute a white line recognition process. It is configured to include a whiteout determination unit 13, an adjacent region extraction unit 14, a blackout determination unit 15, a threshold setting unit 16, and a white line determination unit 17.

  Image information output from a camera 20 that captures an image of the outside of a vehicle (not shown) such as an automobile (for example, the front of the vehicle, the rear of the vehicle, etc.) is input to the image input unit 11. The camera 20 may be a camera mounted on the vehicle (that is, a camera mounted on the vehicle may be used) or a camera unique to the white line recognition device 10. .

(White line recognition processing)
Next, the white line recognition processing executed by the white line recognition device 10 will be described with reference to the flowchart in FIG.

In FIG. 2, after image information from the camera 20 is input to the image input unit 11 (step S1), the white line candidate region extraction unit 12 extracts one pixel to be determined from the image as a white line candidate pixel. Alternatively, the white line candidate region extraction unit 12 determines that one pixel and one pixel are present when there is a pixel having the same or similar pixel value as the one pixel around the one pixel to be determined. Are extracted as white line candidate areas. (Step S2)
Next, the adjacent area extraction unit 14 extracts the white line candidate pixel or the pixel adjacent to the white line candidate area extracted in the process of step S2 as an adjacent pixel or an adjacent area (step S3). Here, the adjacent area means a pixel group including a white line candidate pixel or a pixel adjacent to the white line candidate area, and a pixel located around the pixel and having the same or similar pixel value as the pixel value. I do.

  Next, the whiteout determination unit 13 determines whether or not the luminance value of the white line candidate pixel or the white line candidate region extracted in the process of step S2 matches the maximum luminance value (step S4). The whiteout determination unit 13 determines that whiteout has occurred when the luminance value of the white line candidate pixel or white line candidate region matches the maximum luminance value. On the other hand, when the luminance value of the white line candidate pixel or the white line candidate region does not match the maximum luminance value, the whiteout determination unit 13 determines that there is no whiteout.

  Note that the “maximum luminance value” is “255” for a camera that measures a luminance value using, for example, unsigned 8 bits. In the case of the white line candidate region, the overexposed determining unit 13 determines that the maximum luminance value or the mode of luminance is equal to the luminance maximum value among the plurality of luminance values respectively corresponding to the plurality of pixels constituting the white line candidate region. What is necessary is just to judge whether or not.

  In parallel with the processing in step S4, the blackout determination unit 15 determines whether or not the luminance value of the adjacent pixel or adjacent area extracted in the processing of step S3 matches the minimum luminance value (step S5). . The black crush determination unit 15 determines that a black crush has occurred when the luminance value of the adjacent pixel or the adjacent region matches the minimum luminance value. On the other hand, when the luminance value of the adjacent pixel or the adjacent region does not match the minimum luminance value, the blackout determination unit 15 determines that the image is not blackout.

  The “minimum luminance value” is “0” for a camera that measures a luminance value with, for example, unsigned 8 bits. In the case of the adjacent region, the blackout determination unit 15 determines that the minimum luminance value or the mode of luminance among the plurality of luminance values corresponding to the plurality of pixels forming the adjacent region matches the minimum luminance value. It is sufficient to determine whether or not there is.

  Next, the threshold setting unit 16 sets a threshold used for white line determination, which will be described later, according to the determination result by the overexposed determination unit 13 and the determination result by the black crush determination unit 15 (step S6).

  Here, the change in the luminance value of the white line portion and the asphalt portion of the road will be described with reference to FIG. In FIG. 3, a solid line indicates an example of a change in luminance value of a white line portion, and a broken line indicates an example of a change of luminance value of an asphalt portion.

  As shown in FIG. 3, the brightness value of the white line portion and the brightness value of the asphalt portion change in proportion to the brightness of the scene (image). When no whiteout or blackout occurs, the difference between the brightness value of the white line portion and the brightness value of the asphalt portion is “brightness difference 2”.

  However, when overexposure occurs (see the "sunlight area where the luminance of the white line is saturated upward" in FIG. 3), the luminance value of the white line portion is constant at the maximum luminance value (255 in this case). On the other hand, the brightness value of the asphalt portion increases as the scene becomes brighter. Then, the difference between the brightness value of the white line portion and the brightness value of the asphalt portion is smaller than “brightness difference 2”, for example, “brightness difference 3”.

  Similarly, when black crushing has occurred (see “shade where the brightness of asphalt is saturated downward” in FIG. 3), the brightness value of the asphalt portion is constant at the minimum brightness value (here, 0). On the other hand, the brightness value of the white line portion becomes smaller as the scene becomes darker. Then, the difference between the brightness value of the white line portion and the brightness value of the asphalt portion is smaller than “brightness difference 2”, for example, “brightness difference 1”.

  For this reason, if the threshold value used for the white line determination is set to a fixed value corresponding to, for example, a luminance difference of 2, the white line may not be able to be recognized when whiteout or blackout occurs. Therefore, the threshold setting unit 16 according to the present embodiment reduces the threshold value used for the white line determination when whiteout or blackout occurs, as compared with the case where no whiteout or blackout occurs.

  In other words, in the process of step S6, when the whiteout determination is made by the whiteout determination unit 13 or when the blackout determination is made by the blackout determination unit 15, the threshold setting unit 16 sets the threshold value used for the white line determination. Is set to be smaller than when the whiteout determination unit 13 determines that it is not whiteout and when the blackout determination unit 15 determines that it is not blackout.

  The threshold value used for the white line determination is, for example, a luminance difference between the white line portion and the asphalt portion in an image in which the white line portion is overexposed, and a white line portion and an asphalt portion in an image in which the asphalt portion is blackened. A plurality of samples having a difference in luminance from the sample may be collected and set appropriately based on the collected samples.

  Alternatively, using an image in which the white line portion is overexposed or an image in which the asphalt portion is blackened out, when the threshold used for the white line determination is changed, the image is detected as a white line even though it is not a white line portion. The amount of erroneous detection indicating a pixel that has not been detected as a white line despite being a white line portion or an undetected amount indicating a pixel that has not been detected as a white line is determined. A threshold value that achieves the design target value of 10 may be set as a threshold value used for white line determination.

Returning to FIG. 2 again, after the process of step S6, the white line determination unit 17 extracts the white line candidate pixel or the luminance value of the white line candidate region extracted by the white line candidate region extraction unit 12 and the luminance value of the white line candidate region extracted by the adjacent region extraction unit 14. Then, a luminance difference from the luminance value of the adjacent pixel or the adjacent area is calculated. Then, the white line determination unit 17 determines whether the obtained luminance difference is larger than the threshold set by the threshold setting unit 16. (Step S7)
The white line determination unit 17 determines a white line candidate pixel or a white line candidate region as a white line pixel when the obtained luminance difference is larger than the threshold value. On the other hand, when the obtained luminance difference is smaller than the threshold, the white line determining unit 17 determines that the white line complementary pixel or the white line candidate region is not a white line pixel. When the obtained luminance difference is equal to the threshold value, it may be handled by including one of them.

  Next, the white line recognition device 10 determines whether or not there is a pixel that is not a determination target of the white line recognition processing (Step S8). When it is determined that there is a pixel that is not a determination target of the white line recognition processing (step S8: Yes), the processing from step S2 described above is executed. On the other hand, when it is determined that there is no pixel that is not a determination target of the white line recognition processing (step S8: No), the white line recognition device 10 determines, for an image different from the image subjected to the current white line recognition processing, The above-described white line recognition processing is performed.

  Note that the pixels to be determined in the white line recognition processing are not limited to all the pixels constituting the image. For example, if a road area in an image can be specified in advance, only pixels forming the road area may be determined as white line recognition processing targets. Alternatively, the range in which the white line pixel exists as a result of the previous white line recognition processing is specified, and the image of the previous white line recognition processing is compared with the image of the current white line recognition processing. If the difference (that is, the change in the position or the change in the image due to the moving object) is specified, only the pixels constituting the estimated existence range of the white line pixel in the image subjected to the white line recognition processing this time are represented by the white line It may be a determination target of the recognition processing.

(Technical effect)
In the present embodiment, when overexposure or crushed black occurs, the threshold value used by the threshold value setting unit 16 for white line determination is made smaller than when no overexposure or crushed black occurs. For this reason, the white line recognition device 10 can appropriately recognize a white line even in an image in which whiteout or blackout occurs in which the difference between the luminance value of the white line portion and the luminance value of the asphalt portion is relatively small. Can be.

  The “white line candidate region extracting unit 12” and the “adjacent region extracting unit 14” according to the present embodiment are examples of the “extracting unit” according to the present invention. The “threshold setting unit 16” and the “white line determination unit 17” according to the present embodiment are examples of the “setting unit” and the “recognition unit” according to the present invention, respectively.

<Modification>
In the above-described embodiment, the occurrence of overexposure or crushed black is determined based on the luminance value. However, a color value may be used instead of the luminance value. In this case, the overexposure determination unit 13 determines that the overexposure is under the condition that the color value of the white line candidate pixel or the white line candidate area is, for example, (225, 225, 225) (for the RGB color model). Further, the black crush determination unit 15 determines black crush on condition that the color value of the adjacent pixel or the adjacent region is, for example, (0, 0, 0) (in the case of the RGB color model).

  The present invention is not limited to the above-described embodiment, and can be appropriately changed without departing from the spirit and spirit of the invention which can be read from the claims and the entire specification, and a white line recognition device with such a change Are also included in the technical scope of the present invention.

  DESCRIPTION OF SYMBOLS 10 ... White line recognition apparatus, 11 ... Image input part, 12 ... White line candidate area extraction part, 13 ... Whiteout determination part, 14 ... Adjacent area extraction part, 15 ... Black crush determination part, 16 ... Threshold setting part, 17 ... White line Judgment unit, 20 ... Camera

Claims (1)

A white line recognition device that recognizes a white line on a road based on an image of the outside of a vehicle,
Extracting means for extracting a white line candidate area and an adjacent area adjacent to the white line candidate area from the image,
A recognition unit that recognizes the white line candidate region as a white line, on condition that a difference between the luminance value of the extracted white line candidate region and the luminance value of the extracted adjacent region is larger than a threshold value;
When the first pixel value that is the pixel value of the extracted white line candidate region is the upper limit saturation value, the threshold value is made smaller than when the first pixel value is not the upper limit saturation value. Setting means for setting the threshold value to be smaller than when the second pixel value that is the pixel value of the obtained adjacent region is the lower limit saturation value, as compared with the case where the second pixel value is not the lower limit saturation value;
A white line recognition device, comprising:

Images