WO2011039977A1 - 横断歩道標示検出方法および横断歩道標示検出装置 - Google Patents
横断歩道標示検出方法および横断歩道標示検出装置 Download PDFInfo
- Publication number
- WO2011039977A1 WO2011039977A1 PCT/JP2010/005741 JP2010005741W WO2011039977A1 WO 2011039977 A1 WO2011039977 A1 WO 2011039977A1 JP 2010005741 W JP2010005741 W JP 2010005741W WO 2011039977 A1 WO2011039977 A1 WO 2011039977A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- pedestrian crossing
- image data
- imaging system
- crossing sign
- unit
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
- G01C21/36—Input/output arrangements for on-board computers
- G01C21/3697—Output of additional, non-guidance related information, e.g. low fuel level
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
Definitions
- the present invention relates to a pedestrian crossing sign detection method and a pedestrian crossing sign detection apparatus for detecting a pedestrian crossing sign on a road surface using an in-vehicle camera image.
- the crosswalk sign in Japan is a sign in which rectangular paint parts and non-paint parts are alternately repeated with the same width, and the paint part and the non-paint part are the same width and the width is standardized from 45 cm to 50 cm. Yes.
- a conventional pedestrian crossing sign detection device uses an image brightness edge that changes greatly in image brightness by using the fact that the image brightness of a pixel in a paint portion and a pixel in a non-paint portion differ greatly in image data captured by a camera. It was extracted. As a result, the conventional pedestrian crossing sign detection device calculates the interval and the number of adjacent image luminance edges and determines whether or not the pedestrian crossing is present (see Patent Document 1). Hereinafter, the prior art will be described.
- FIG. 1 shows a block diagram of a conventional pedestrian crossing sign detection apparatus 1500 described in Patent Document 1.
- a stereo image data acquisition unit 1501 acquires a pair of image data obtained by simultaneously capturing the front of the vehicle including the road surface using a stereo camera including two imaging systems.
- data of an image captured by one imaging system is used as reference image data
- data of an image captured by the other imaging system is used as reference image data.
- the distance data calculation unit 1502 calculates the distance of the real space from the stereo camera at each point in the image by stereo-matching the standard image data and the reference image data.
- the pedestrian crossing sign detection unit 1503 detects the presence or absence of a pedestrian crossing sign in the acquired reference image data, and detects the position of the pedestrian crossing sign in the image data.
- the pedestrian crossing sign detection unit 1503 includes a luminance edge detection unit 1504, a width determination unit 1505, an interval determination unit 1506, and a repetition number determination unit 1507.
- the luminance edge detection unit 1504 uses an arbitrary area in the reference image data as a setting range, and detects a luminance edge position in the setting range.
- the detection of the luminance edge position in an arbitrary setting range is performed for the entire range of the reference image data.
- the luminance edge position is detected by general image processing such as a Sobel filter.
- the width determination unit 1505 extracts, from the luminance edge positions detected by the luminance edge detection unit 1504, an image luminance edge that changes from high luminance to low luminance as a starting point, and an image luminance edge that changes from low luminance to high luminance. Is extracted as the end point. Then, the width determination unit 1505 extracts a combination of the start point and the end point, from the combination group of the start point and the end point, whose interval is the same as or close to the standard width of the paint part of the pedestrian crossing.
- the interval determination unit 1506 is located on the same horizontal axis as the end point of the first combination in the reference image from the end point of the first combination in the combination group of the start point and the end point extracted by the width determination unit 1505, and An interval to the start point of the second combination different from the start point of the first combination is calculated. Then, the interval determination unit 1506 extracts the first combination whose interval is the same as or close to the standard width of the non-painted portion of the pedestrian crossing.
- the repetition number determination unit 1507 calculates the number of repetitions of the combination of the start point and the end point extracted by the interval determination unit 1506 in the setting range, and when the result is a predetermined number of times or more, the region is set as a pedestrian crossing region.
- the image luminance edge detection method is executed on the assumption that the image luminance of the pixels in the paint portion and the image luminance of the pixels in the non-paint portion in the pedestrian crossing sign are greatly different. For this reason, when there is a situation where the brightness of the edge part, which is the boundary between the paint part and the non-paint part of the pedestrian crossing sign, does not appear strongly, the difference in image brightness between the paint part and the non-paint part becomes small. It is difficult to detect an image luminance edge at the boundary between a paint part and a non-paint part in a sidewalk sign.
- An object of the present invention is to provide a pedestrian crossing sign detection method and a pedestrian crossing sign detection apparatus capable of accurately detecting the presence or absence and the position of a pedestrian crossing sign from a captured image even when it is difficult to detect a luminance edge of a paint part. Is to provide.
- the pedestrian crossing sign detection apparatus has a first imaging system, and traverses an image data acquisition unit that acquires image data of a road surface imaged using the first imaging system, and a setting range of the image data.
- Period calculation unit that calculates the period of the paint part when the paint part of the sidewalk sign is included, based on the coordinates of the setting range, the geometric parameters of the first imaging system, and the standard value of the pedestrian crossing sign
- a frequency analysis unit that performs a frequency analysis of the image luminance in the set range based on a basis function with an integer multiple of the period of the paint portion as one cycle, and calculates a frequency power distribution from the set range
- An area detection unit that extracts a coordinate point having a power level greater than a predetermined threshold and detects an area where the coordinate point exists as a pedestrian crossing marking area;
- a position output section which outputs the data, and comprising a.
- the period calculation unit is set from a memory unit that stores in advance the geometric parameters of the first imaging system and the standard value of the pedestrian crossing sign, and the image data acquisition unit A parameter acquisition unit that acquires the coordinates of the range, acquires the geometric parameters of the first imaging system and the standard value of the pedestrian crossing sign from the memory unit, the coordinates of the setting range, and the geometry of the first imaging system A calculation unit that calculates the period of the paint portion based on the target parameter and the standard value of the pedestrian crossing sign.
- the period calculation unit is set from a memory unit that stores in advance the geometric parameters of the first imaging system and the standard value of the pedestrian crossing sign, and the image data acquisition unit It has a parameter acquisition unit that acquires the coordinates of the range, acquires the geometric parameters of the first imaging system and the standard value of the pedestrian crossing sign from the memory unit, and a sensor that detects changes in angular velocity, and the sensor detects When the detected change in angular velocity is greater than a predetermined threshold, the change in the angular velocity detected by the sensor when the turn detection unit for detecting the turn of the pedestrian crossing detection device and the turn of the pedestrian crossing detection device are detected.
- the cycle of the paint part Calculate A calculation unit that is one with a.
- the cycle calculation unit has an effect that the cycle of the paint portion in the set range when the paint portion of the pedestrian crossing sign exists in the set range can be calculated.
- the pedestrian crossing sign detection device shows the first imaging system and the setting range based on the coordinates of the setting range included in the image data and the geometric parameters of the first imaging system.
- a distance calculation unit that calculates the distance in real space with the road surface as distance data is further provided, and the cycle calculation unit calculates the cycle of the paint part when the set range includes the paint part of the pedestrian crossing sign, the distance data,
- the position output unit outputs coordinate data and distance data of the pedestrian crossing marking area in the image data.
- the pedestrian crossing sign detection device further includes a stereo matching unit and a distance calculation unit
- the image data acquisition unit further includes a second imaging system, and the first imaging system is provided.
- the standard image data obtained by imaging the road surface using the reference image data obtained by imaging the road surface using the second imaging system is acquired, and the stereo matching unit is configured to obtain a reference point included in the standard image data and the reference image data.
- 3D reference image data based on one distance data Calculates the ccel data, calculates the distance in the real space between the first imaging system and the road surface indicated by the setting range included in the reference image data as the second distance data based on the three-dimensional voxel data, and calculates the period
- the unit calculates the period of the paint part when the set range includes the paint part of the pedestrian crossing sign based on the second distance data and the standard value of the pedestrian crossing sign.
- the coordinate data of the pedestrian crossing marking area in the image data and the second distance data are output.
- the geometric parameters of the camera are parameters including a camera installation pitch angle, an installation height, a focal length, and a pixel pitch.
- the frequency analysis unit performs frequency analysis based on a basis function in which one cycle is twice or more the cycle of the paint portion.
- the area detection unit controls a predetermined threshold based on the illuminance around the vehicle.
- the pedestrian crossing sign detection method acquires the road surface image data imaged using the imaging system, and sets the cycle of the paint part when the paint part of the pedestrian crossing sign is included in the setting range of the image data. Calculated based on the coordinates of the setting range, the geometric parameters of the imaging system, and the standard value of the pedestrian crossing marking, and based on a basis function with an integer multiple of the period of the paint portion as one period.
- the frequency analysis of the image luminance in the image is performed, the frequency power distribution is calculated, the coordinate point whose power level in the frequency power distribution is greater than the predetermined threshold is extracted from the setting range, and the area where the coordinate point exists is used as the pedestrian crossing marking area It detects and outputs the coordinate data of the pedestrian crossing marking area
- the present invention by calculating the period of the pedestrian crossing sign in the set range in the image data and performing frequency analysis in the period, it is possible to extract the periodic characteristics of the pedestrian crossing sign, Since it is possible to detect without detecting the luminance edge of the paint portion, even when it is difficult to detect the luminance edge of the paint portion, it is possible to detect a pedestrian crossing sign.
- the figure which shows the block configuration of the conventional pedestrian crossing sign detection apparatus The figure which shows the block configuration of the pedestrian crossing sign detection apparatus in Embodiment 1 of this invention.
- the figure which shows the positional relationship of the image data of the road surface acquired with a vehicle-mounted camera, and a vehicle-mounted camera and a road surface Diagram showing image data taken from a pedestrian crossing sign Diagram showing image data taken from a pedestrian crossing sign The figure which shows the composition of the period calculation part Diagram for explaining frequency analysis method
- a figure used to explain how to detect areas where pedestrian crossings are present A figure used to explain how to detect areas where pedestrian crossings are present
- FIG. 2 shows a block configuration of the pedestrian crossing sign detection apparatus 100 according to Embodiment 1 of the present invention.
- a pedestrian crossing sign detection apparatus 100 shown in FIG. 2 includes a pedestrian crossing sign detection unit 101 and an image data acquisition unit 102.
- the pedestrian crossing sign detection unit 101 includes a period calculation unit 103, a frequency analysis unit 104, a region detection unit 105, and a position output unit 106.
- the image data acquisition unit 102 acquires road image data using an in-vehicle camera.
- the image data is a still image, and the horizontal axis is described as the X axis and the vertical axis is described as the Y axis. Further, a description will be given using one pixel on the screen as a coordinate point.
- the in-vehicle camera is described as a camera having one imaging system, but this is not restrictive.
- FIG. 3 shows the road surface image data acquired by the in-vehicle camera and the positional relationship between the in-vehicle camera and the road surface.
- FIG. 3A shows road surface image data acquired by the in-vehicle camera
- FIG. 3B shows a positional relationship between the in-vehicle camera and the road surface.
- the in-vehicle camera is installed in front of the vehicle with an installation height H and an installation pitch angle ⁇ , and there is no installation roll angle that is a rotation angle in a lateral direction with respect to the imaging direction of the in-vehicle camera. explain.
- this invention can be implemented.
- the period calculation unit 103 acquires the image data acquired by the image data acquisition unit 102, and as shown in FIG. 3A, the Y coordinate in the image data is an arbitrary Y coordinate (y1). A certain horizontal range is selected as the setting range.
- the cycle calculating unit 103 calculates the cycle of the paint portion in the image data when the set range includes the paint portion of the pedestrian crossing sign. This cycle calculation is performed uniformly on the image data regardless of whether or not the pedestrian crossing sign is actually captured in the image data.
- the period of the paint part is calculated based on the positional relationship between the in-vehicle camera and the road surface, the Y coordinate (y1), and the period of the paint part of the pedestrian crossing in the real space. Do not depend.
- the frequency analysis unit 104 performs frequency analysis of the image luminance in the setting range based on the basis function with the cycle of the paint portion calculated by the cycle calculation unit 103 as one cycle, and calculates the distribution of the frequency power P (x). To do.
- the region detection unit 105 detects, from the distribution of the frequency power P (x) calculated by the frequency analysis unit 104, a region where the power level exceeds a predetermined threshold as a region where a pedestrian crossing sign exists.
- the frequency power P (x) means a power distribution in a frequency setting range in which the period of the paint portion is one period.
- Each processing for the set range related to an arbitrary Y coordinate (y1) as described above is performed for all the Y coordinates in the image data, that is, the Y coordinate (y1) to the Y coordinate (ym) (m: integer). Then, a region where a pedestrian crossing sign exists in the image data is detected. Note that m (m: integer) is the number of pixels on the Y axis in the image data.
- the position output unit 106 outputs the coordinate data of the area where the detected pedestrian crossing sign exists.
- the period calculation unit 103 uses pixels located in a horizontal row whose Y coordinate is an arbitrary Y coordinate (y1) from the image data of the road surface acquired by the image data acquisition unit 102.
- the range to be configured is selected as the setting range. Note that the setting range is described as being composed of all coordinate points located in a horizontal row where the Y coordinate is an arbitrary Y coordinate (y1), but the setting range is located in a horizontal row where the Y coordinate is the Y coordinate (y1). Of the coordinate points, the coordinate points may be located in an arbitrary range.
- the cycle calculation unit 103 calculates the cycle of the paint portion when the paint portion of the pedestrian crossing sign is included in the setting range of the image data. Note that this cycle calculation is performed uniformly on the image data for subsequent processing regardless of whether or not the pedestrian crossing sign is actually captured in the image data.
- FIG. 4 shows image data when a pedestrian crossing sign is imaged.
- the function of the period calculation unit 103 will be described on the assumption that the setting range includes all coordinate points located in a horizontal row in the Y coordinate (y1).
- the cycle calculation unit 103 acquires the image data acquired by the image data acquisition unit 102, and selects the setting range as described above. Then, the cycle calculation unit 103 performs the painting in the set range on the image data when the set range includes a paint portion, as shown in FIG. 4, regardless of whether a crosswalk sign is present in the image data.
- the period Ty1 of the part is calculated.
- the reason for calculating the cycle Tyn (n: integer) is that when the pedestrian crossing sign is displayed in the image data, if the Y coordinate is different, the width and interval of the paint portion are different.
- the period Tyn (n: integer from 1 to m) of the paint part is defined by the width of the paint part of the pedestrian crossing marking at each Y coordinate (yn) (n: integer from 1 to m) and the interval between the paint parts.
- This is a sum value, and the width and interval of the paint part in real space (that is, the width of the non-paint part) are the same values in the Japanese regulations. Therefore, according to Japanese regulations, the period Tyn (n: an integer from 1 to m) is twice the width of the paint portion or the interval of the paint portions in the image data. That is, the cycle Tyn is the sum of the width of the paint portion and the interval between the paint portions (that is, the width of the non-paint portion) in the image data.
- the width of the paint part and the non-paint part of the pedestrian crossing sign is different as in other countries, and the sum of the width of the paint part and the non-paint part is constant, and the paint part and the non-paint part are separated.
- the period Tyn (n: an integer from 1 to m) is the sum of the width of the paint portion and the width of the non-paint portion in the image data.
- the width of the painted portion and the non-painted portion of the pedestrian crossing sign is different, the sum of the width of the painted portion and the non-painted portion is constant, and the painted portion and the non-painted portion are alternately repeated.
- image data when a pedestrian crossing sign is imaged is shown.
- the cycle Tyn is calculated by Equation 1 using the width or interval (W) of the paint part of the pedestrian crossing in real space, the coordinates (y) of the setting range, and the geometric parameters of the camera.
- the geometric parameters of the camera are camera installation parameters including the installation pitch angle ⁇ , the camera installation height H, the camera focal length f, and the Y-axis pixel pitch dpy.
- yc in Formula 1 is the center coordinate of the Y axis.
- Equation 1 the width or interval of the paint part of the pedestrian crossing sign (that is, the interval is the width of the non-paint part) is W, but the width of the paint part of the pedestrian crossing sign and the width of the non-paint part. If the sum of and W ′ is W ′, Equation 2 is obtained.
- the width of the painted part and the non-painted part of the pedestrian crossing marking is different as in other countries, and the sum of the width of the painted part and the width of the non-painted part is constant. Tyn can be calculated even in the case where the non-painted portion is alternately and marked.
- FIG. 6 shows a configuration diagram of the period calculation unit 103.
- the period calculation unit 103 includes a parameter acquisition unit 401, a calculation unit 402, and a memory 403.
- the memory 403 stores in advance the geometric parameters of the camera and the standard value of the pedestrian crossing sign.
- the standard value of the pedestrian crossing sign includes the width of the paint part and the width of the non-paint part of the pedestrian crossing sign. That is, the standard value of the pedestrian crossing sign includes at least one of the width of the paint part of the pedestrian crossing sign, the width of the non-paint part, and the sum of the width of the paint part and the width of the non-paint part. .
- the parameter acquisition unit 401 acquires the image data from the image data acquisition unit 102, selects the Y coordinate (yn) of the setting range, and reads the geometric parameters of the camera and the standard value of the pedestrian crossing sign from the memory 403. Then, based on the camera geometric parameters acquired by the parameter acquisition unit 401, the Y coordinate (yn) of the setting range, and the standard value of the pedestrian crossing sign, the calculation unit 402 uses Equation 1 or Equation 2 Tyn is calculated.
- the frequency analysis unit 104 performs frequency analysis of image luminance in a set range using an even function and an odd function of a rectangular wave whose cycle is Ty1 as a basis function. By performing the frequency analysis, the frequency power P (x) in the set range is calculated.
- FIG. 7 is a diagram for explaining the frequency analysis method.
- 7A shows the luminance distribution Ly1 (x) in the set range
- FIG. 7B shows the even function u (s) 501 among the basis functions used for frequency analysis in the set range
- FIG. 7C shows the odd function v ( s) 502 is shown.
- the frequency power P (x) is calculated from the inner product values Au (x) and Av (x) calculated using the even function u (s), the odd function v (s), and the following Equation 3.
- the powers of two orthogonal basis functions are calculated, and the frequency power P (x) becomes the same value at any position (x) of the pedestrian crossing sign.
- the area of the sidewalk sign is easily calculated.
- the frequency power P (x) is the same value at any position (x) of the pedestrian crossing sign, and the case where the width of the paint part and the non-paint part in the image data are equal will be described. To do.
- the frequency power P (0) here is 600.8 from Equation 3.
- the frequency power P (1) here is 600.8 from Equation 3, and is equal to P (0).
- the frequency power P (x) is the same value at any position (x) of the pedestrian crossing sign, and the case where the width of the paint part and the non-paint part in the image data are different will be described. To do.
- the case of 10, 10, 200, 200, 10, 10] will be described.
- the number of pixels, that is, the width of the paint portion having a luminance value of 200 and the non-paint portion having a luminance value of 10 are different.
- this function which is a function having the lowest frequency in the rectangular wave, is used, as in the case where the width of the paint portion in the image data is equal to the width of the non-paint portion.
- the frequency power P (0) here is 424.9 from Equation 3.
- the frequency power P (1) here is 424.9 from Equation 3, and is equal to P (0).
- the odd function and the even function are composed of a rectangular wave having the lowest frequency, that is, a rectangular wave having the same period, a high frequency component cannot be extracted with only one period of the odd function and the even function, and the frequency power becomes small. . Therefore, when the width of the paint part is different from the width of the non-paint part, it is necessary to change the threshold of the power level set by the area detection unit 105 in accordance with the width and the interval defined by the standard. .
- the threshold of the power level set by the area detection unit 105 is lowered, it is not often erroneously detected in an actual traffic environment. This is because when the frequency power increases, it is a repeating pattern in which the size of one cycle is equal to the standard value, and the width of the paint part and the width of the non-paint part are more equal. This is because there is nothing other than sidewalk marking.
- the frequency analysis using the rectangular wave is not affected by the orthogonal component of the luminance distribution of the image data, and only the frequency power of the rectangular wave having a cycle of Ty1 can be extracted. Therefore, even when the illuminance on the road surface changes, The frequency power of the sidewalk sign is unchanged and is a stable detection method.
- the invariance of the frequency power with respect to the illuminance on the road surface will be described.
- the frequency power here is 600.8 as a result of the square root of -190 and 570.
- the frequency power P2 (0) here is 600.8 from Equation 3, which is equal to P1 (0).
- the period of the basis function can be set to a period twice as long as Ty1, or can be set to a period three times as long as Ty1. This is not a limited range.
- the frequency analysis method using the basis function whose period is twice Ty1 is suitable for roughly detecting the position of the pedestrian crossing sign because the area determined as the pedestrian crossing sign is small.
- the region detection unit 105 extracts coordinates (x) whose power level exceeds a predetermined threshold from the distribution of the frequency power P (x) calculated by the frequency analysis unit 104, and the extracted x coordinates exist densely.
- the area is detected as an area where a pedestrian crossing sign exists.
- FIG. 8 is a diagram for explaining a method for detecting an area where a pedestrian crossing sign exists.
- FIG. 8A shows image data in which a pedestrian crossing sign is imaged in a method of detecting a region where a pedestrian crossing sign exists from the distribution of frequency power P (x) in the set range
- FIG. 8B shows frequency power P in the set range. The distribution of (x) is shown.
- the x-coordinate at which the frequency power P (x) exceeds a predetermined threshold is a coordinate point between x1 ⁇ x ⁇ x2, meaning that there is a crosswalk sign between x1 and x2 in the set range. To do.
- the frequency analysis unit 104 may determine a surrounding illuminance or the like, and may be a variable value that can be changed according to the determination result. Thereby, when the surroundings are dark and there is a lot of gain noise, such as at night, the pedestrian crossing sign can be detected regardless of the luminance edge by changing the threshold value low.
- the period calculation, frequency analysis, and area calculation for the set range of the arbitrary Y coordinate (y1) on the image data described above are performed for all Y coordinates in the image data, that is, from the Y coordinate (y1) to the Y coordinate.
- a series of processes are performed for all the Y coordinates, but a series of processes may be performed for an arbitrary range.
- the arbitrary range is a range from the vanishing point to the lowest end of the screen.
- the vanishing point means a point where parallel lines intersect in perspective.
- FIG. 9 is a diagram for explaining a method for detecting an area where a crosswalk sign is present.
- FIG. 9A shows the frequency power P (x) in the set range at the Y coordinate (y1) as three-dimensional data
- FIG. 9B shows a series of operations in the period calculation unit 103, frequency analysis unit 104, and region calculation unit 105 shown above.
- the range exceeding the predetermined threshold shown in FIG. 8B is shown in the sum P (x, y) of the frequency powers obtained by performing the processing from the Y coordinate (y1) to the Y coordinate (ym) shown in FIG. 8A.
- a region 701 indicated by a solid line in FIG. 9 is a range exceeding a predetermined threshold in the total sum P (x, y), and a crosswalk sign is present in the region 701.
- the position output unit 106 outputs the coordinate data of the region 701 detected by the region detection unit 105.
- the frequency analysis is performed based on the cycle of the pedestrian crossing sign in the set range in the image data, and the area where the frequency power exceeds the predetermined threshold is detected as the area where the pedestrian crossing sign exists. Since the pedestrian crossing sign can be detected without detecting the luminance edge of the pedestrian crossing, the pedestrian crossing sign can be detected even when the detection of the luminance edge is difficult.
- the entire horizontal row range positioned at the Y coordinate (y1) has been described as the set range.
- an arbitrary coordinate point positioned at the Y coordinate (y1) is set as the set range. Also good.
- the cycle calculation unit 103 calculates the cycle of the paint portion of the pedestrian crossing sign in the image data for each image data acquired by the image data acquisition unit 102.
- the cycle of the paint portion is calculated in advance. And may be stored in the memory.
- a square wave is used as the basis function, but a trigonometric function can also be used as the basis function.
- FIG. 10 shows a block configuration of the pedestrian crossing sign detection apparatus 800 in the second embodiment.
- the difference from Embodiment 1 is that a distance calculation unit 802, a cycle calculation unit 803, and a position output unit 804 are provided, and the cycle calculation unit 103 and the position output unit 106 are unnecessary.
- FIG. 10 the same components as those in the first embodiment are given the same reference numerals.
- the image data acquisition unit 102 acquires image data including a road surface. Since the method for acquiring the image data is the same as that in the first embodiment, the description thereof is omitted.
- the distance calculation unit 802 extracts all or an arbitrary range on the X axis at an arbitrary Y coordinate (y1) from the road surface image data acquired by the image data acquisition unit 102 as a setting range.
- the distance calculation unit 802 calculates a distance Dy1 in real space between the camera and the road surface indicated by the setting range.
- the distance Dyn (n: integer) is calculated by Equation 4 using the coordinates of the set range and the geometric parameters of the camera.
- the geometric parameters of the camera are the camera installation parameters including the installation pitch angle ⁇ , the camera installation height H, the camera focal length f, and the Y-axis pixel pitch dpy, as described in the description of the first embodiment.
- Yc means the coordinates of the center of the Y-axis image.
- the installation pitch angle ⁇ and the installation height H change depending on bouncing and pitch movement of the host vehicle, but the method of calculating the distance to the road surface using the installation pitch angle ⁇ and the installation height H is an example. It is not intended to limit the invention.
- the pitch angle variation and height variation can be measured and corrected using other sensors such as a vehicle height sensor and a gyro sensor.
- the period calculation unit 803 receives the distance Dy1 calculated by the distance calculation unit 802, and sets the range when the paint portion is imaged in the setting range regardless of whether or not the pedestrian crossing sign exists in the image data.
- the period Ty1 of the paint portion at is calculated.
- the cycle Tyn (n: integer) is calculated using the distance Dyn (n: integer) calculated by the distance calculation unit 802, the width or interval (W) of the paint part of the crosswalk in the real space, and the camera focal length f. Calculated by Equation 5.
- the width or interval of the paint part of the pedestrian crossing sign (that is, the interval is the width of the non-paint part) is W, but the width of the paint part of the pedestrian crossing sign and the width of the non-paint part. If the sum of and is W ′, Equation 6 is obtained.
- the width of the paint part and the non-paint part of the pedestrian crossing sign is different as in other countries, and the sum of the width of the paint part and the width of the non-paint part is constant. Tyn can be calculated even in the case where the non-painted portion is alternately and marked.
- the cycle calculating unit 803 can calculate the cycle Ty1 in the same manner as in the first embodiment, and thus the description thereof is omitted in the second embodiment.
- the frequency analysis unit 104 performs frequency analysis of the image luminance in the set range using the even function and the odd function of the rectangular wave with the cycle of Ty1 calculated by the cycle calculation unit 803 as basis functions. Since the function of the frequency analysis unit 104 is the same as that of the first embodiment, description thereof is omitted.
- the area detection unit 105 extracts the x coordinate whose power level exceeds a predetermined threshold from the distribution of the frequency power P (x) calculated by the frequency analysis unit 104, and extracts the area where the extracted x coordinates exist densely. Detect as an area where there is a pedestrian crossing sign. Since the function of the area detection unit 105 is the same as that of the first embodiment, description thereof is omitted.
- the position output unit 804 outputs the coordinate data of the region where the pedestrian crossing sign detected by the region detection unit 105 and the information of the distance Dyn (n: integer) calculated by the distance calculation unit 802 are output.
- the frequency analysis is performed based on the cycle of the pedestrian crossing sign in the set range in the image data, and the area where the frequency power exceeds the predetermined threshold is detected as the area where the pedestrian crossing sign exists. Because it is not necessary to detect the luminance edge of the pedestrian crossing, it is possible to detect the pedestrian crossing sign even when the luminance edge of the paint part is weak and to detect the distance in real space from the camera to the pedestrian crossing sign. it can.
- FIG. 11 shows a block configuration of the pedestrian crossing sign detection apparatus 900 according to the third embodiment.
- the difference from the first embodiment is a stereo camera in which the in-vehicle camera is equipped with two imaging systems, a stereo image data acquisition unit 902, a stereo matching unit 903, a distance calculation unit 904, a period calculation unit 905, and The position output unit 804 is provided, and the period calculation unit 103 and the position output unit 106 are unnecessary.
- the pedestrian crossing sign detection apparatus 900 has a pedestrian crossing sign in a set range by changing the pitch angle of the camera with respect to the road surface when there is bouncing or pitching of the own vehicle while the vehicle is traveling and the road surface is inclined. This solves the problem that the period calculation accuracy of the paint part in the image data in the case where the paint part is included deteriorates.
- the stereo image data acquisition unit 902 acquires road image data by an in-vehicle stereo camera equipped with two imaging systems. As shown in FIG. 3B, the in-vehicle camera is installed in front of the vehicle as an installation height H and an installation pitch angle ⁇ , and the installation roll angle is assumed to be zero. In addition, about the installation method of a vehicle-mounted stereo camera, it is an example and it can apply also when there exists an installation roll angle, and does not limit the scope of the present invention.
- the stereo matching unit 903 calculates distance data based on the principle of triangulation using a stereo camera having two cameras, a standard camera and a reference camera.
- FIG. 12 shows an outline of calculating the distance of the object 1001 using the principle of triangulation with a stereo camera.
- the distance between the stereo camera and the object is calculated by Equation 7 based on the principle of triangulation.
- the parallax d is a difference between the position of the target object in the standard image captured by the standard camera and the position of the target object in the reference image captured by the reference camera, and is calculated by stereo matching.
- the focal length f is a distance between the image sensor and the imaging surface.
- FIG. 13 is a diagram for explaining the parallax calculation.
- FIG. 13A shows a reference image captured by the reference camera in the image processing for calculating the parallax for the object
- FIG. 13B shows a reference image captured by the reference camera in the image processing for calculating the parallax for the object. Show.
- a reference point in the reference image shown in FIG. 13A is set, and a search area in the reference image shown in FIG. 13B is set corresponding to the reference point.
- FIG. 13A shows a reference point in the reference image shown in FIG. 13A
- a search area in the reference image shown in FIG. 13B is set corresponding to the reference point.
- an evaluation value distribution indicating an evaluation value distribution that is a similarity between the peripheral luminance value of the reference point and the peripheral luminance value of the corresponding point candidate in the search region is calculated, and the corresponding point having the smallest evaluation value is calculated.
- Candidates are calculated as corresponding points.
- SAD Sud of Absolute Differentities
- SSD Sud of Squared Differences
- NCC Normalized Cross-Correlation
- the distance data from the reference camera to the road surface captured in the reference image data is calculated.
- the distance calculation unit 904 calculates three-dimensional voxel data of the reference image using the distance data calculated by the stereo matching unit 903.
- the distance calculation unit 904 calculates a distance D from the reference camera to the road surface imaged in the reference image based on the calculated three-dimensional voxel data.
- FIG. 14 shows the three-dimensional voxel data of the road surface imaged in the reference image.
- the three-dimensional voxel data is data in a coordinate space in which an axis of depth is added to image coordinates having an X axis and a Y axis, and there is one distance data for each point of the image coordinates.
- the road surface is a shaded portion in the three-dimensional voxel data shown in FIG.
- the method of estimating the hatched portion can be calculated by general image processing. For example, since the shaded portion has the largest number of data points in the three-dimensional voxel data, the plane having the densest density distribution of the data points is the most dense. It can be estimated as a hatched portion by the likelihood estimation method.
- One distance data is calculated for the set range based on the estimated plane. In FIG. 14, the depth distance Dy1 of the road surface is determined for the setting range where the Y coordinate is y1, and the distance of the road surface in the setting range can be calculated.
- the distance Dy1 calculated based on the three-dimensional voxel data is the same as the Dy1 calculated by Expression 4 in the second embodiment. That is, in the third embodiment, Dy1 is calculated based on the three-dimensional voxel data without using Equation 4 in the second embodiment.
- the cycle calculation unit 905 selects all or an arbitrary range on the X axis at an arbitrary Y coordinate (y1) from the reference image data as a setting range. Then, the cycle calculation unit 905 calculates the cycle Tyn (n: integer from 1 to m) of the paint portion in the set range when it is assumed that the paint portion of the pedestrian crossing sign exists in the set range included in the reference image data. .
- the period Tyn (n: integer from 1 to m) of the paint part is the distance Dyn (n: integer) calculated by the distance calculation unit 904, the width or interval (W) of the paint part of the crosswalk in real space, and the camera focus Using the distance f, it is calculated according to Equation 5 shown in the second embodiment.
- the period Tyn (n: integer from 1 to m) of the paint part is the distance Dyn (n: integer), the sum (W ′) of the width of the paint part and the width of the non-paint part of the crosswalk in real space, Using the camera focal length f, it is calculated according to Equation 6 shown in the second embodiment.
- the numerical value of Tyn is the same whether using Equation 5 or Equation 6.
- the frequency analysis unit 104 and the region detection unit 105 perform information processing based on image data in the first embodiment, whereas information processing based on reference image data in the third embodiment. Since it is the same except performing point, explanation is omitted. Further, the description of the position output unit 804 has been made in Embodiment 2, and therefore will be omitted.
- this allows the three-dimensional voxel data to be calculated in advance even if the pitch angle of the camera with respect to the road surface fluctuates while the vehicle is running when there is bouncing or pitching of the host vehicle, road surface gradient, etc.
- the cycle of the pedestrian crossing sign can be calculated by a simple method even in such a case.
- FIG. 15 shows a block configuration of a pedestrian crossing sign detection apparatus 1300 according to the fourth embodiment.
- the difference from the first embodiment described above is that in the pedestrian crossing sign detection unit 1301, a cycle calculation unit 1302 is provided, and the cycle calculation unit 103 is not required.
- the pedestrian crossing sign detection apparatus 1300 applies the cycle calculation method shown in the first embodiment when the paint parts of the pedestrian crossing sign are not evenly arranged in the horizontal direction in the image data. is there.
- FIG. 16 shows a block configuration of the period calculation unit 1302.
- the period calculation unit 1302 includes a parameter acquisition unit 401, a memory 403, a turning detection unit 1401, an angle calculation unit 1402, and a calculation unit 1403.
- the memory 403 stores in advance the geometric parameters of the camera and the standard value of the pedestrian crossing sign.
- the parameter acquisition unit 401 receives the image data from the image data acquisition unit 102, selects the Y coordinate (yn) of the setting range, and from the memory 403, the geometric parameters of the camera and the pedestrian crossing markings. Read the standard value.
- the turning detection unit 1401 includes sensors such as a rudder angle sensor and a yaw rate sensor, and the pedestrian crossing marking device is mounted when the change value of the angular velocity acquired by the rudder angle sensor or the yaw rate sensor is larger than a predetermined threshold. Detects that the vehicle has turned.
- the angle calculation unit 1402 calculates the turning angle ⁇ based on the detected change value of the angular velocity.
- the turning angle ⁇ is the absolute value of the displacement from the straight traveling angle.
- the calculation unit 1403 calculates the cycle Ty using Equation 8 based on the read parameters.
- the width or interval of the paint part of the pedestrian crossing sign (that is, the interval is the width of the non-paint part) is W, but the width of the paint part of the pedestrian crossing sign and the width of the non-paint part. If the sum of and is W ′, Equation 9 is obtained.
- the width of the painted portion and the non-painted portion of the pedestrian crossing sign is different as in other countries, and the sum of the width of the painted portion and the non-painted portion is constant, Tyn can be calculated even in the case where the non-painted portion is alternately and marked.
- the geometric parameters of the camera and the standard value of the pedestrian crossing marking are stored in advance in a memory provided in the pedestrian crossing marking apparatus 1300.
- the period calculation unit 1301 can calculate the period of the paint part in the setting range when the paint part of the pedestrian crossing sign exists in the setting range. .
- the pedestrian crossing sign detection apparatus is useful as a device capable of detecting a pedestrian crossing sign from an imaging screen even if the pedestrian crossing sign has a weak luminance edge in the paint portion.
- Crosswalk sign detection apparatus 101 Crosswalk sign detection part 102 Image data acquisition part 103 Period calculation part 104 Frequency analysis part 105 Area
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
Description
図2は、本発明の実施の形態1における横断歩道標示検出装置100のブロック構成を示す。図2に示す横断歩道標示検出装置100は、横断歩道標示検出部101と、画像データ取得部102と、を含む構成である。横断歩道標示検出部101は、周期算出部103と、周波数分析部104と、領域検出部105と、位置出力部106と、を含む構成である。
図10は、実施の形態2における横断歩道標示検出装置800のブロック構成を示す。実施の形態1と異なるのは、距離算出部802、周期算出部803、及び位置出力部804を設け、周期算出部103、及び位置出力部106を不要とする点である。
図11は実施の形態3における横断歩道標示検出装置900のブロック構成を示す。前述の実施の形態1と異なるのは、車載カメラが二つの撮像系を搭載したステレオカメラであって、ステレオ画像データ取得部902、ステレオマッチング部903、距離算出部904、周期算出部905、及び位置出力部804を設け、周期算出部103、及び位置出力部106を不要とする点である。
図15は実施の形態4における横断歩道標示検出装置1300のブロック構成を示す。前述の実施の形態1と異なるのは、横断歩道標示検出部1301において、周期算出部1302を設け、周期算出部103を不要とする点である。
101 横断歩道標示検出部
102 画像データ取得部
103 周期算出部
104 周波数分析部
105 領域検出部
106 位置出力部
401 パラメータ取得部
402 算出部
403 メモリ
501 偶関数
502 奇関数
701 横断歩道標示領域
800 横断歩道標示検出装置
801 横断歩道標示検出部
802 距離算出部
803 周期算出部
804 位置出力部
900 横断歩道標示検出装置
901 横断歩道標示検出部
902 ステレオ画像データ取得部
903 ステレオマッチング部
904 距離算出部
905 周期算出部
1001 対象物
1002 基準カメラ
1003 参照カメラ
1300 横断歩道標示検出装置
1301 横断歩道標示検出部
1302 周期算出部
1401 旋回検出部
1402 角度算出部
1403 算出部
1500 横断歩道標示検出装置
1501 ステレオ画像データ取得部
1502 距離データ算出部
1503 横断歩道標示検出部
1504 輝度エッジ検出部
1505 幅判定部
1506 間隔判定部
1507 繰り返し数判定部
Claims (10)
- 第一の撮像系を有し、前記第一の撮像系を用いて撮像した路面の画像データを取得する画像データ取得部と、
前記画像データの設定範囲に横断歩道標示のペイント部分が含まれる場合の前記画像データ上の前記ペイント部分の周期を、前記設定範囲の座標と、前記第一の撮像系の幾何学的パラメータと、予め定められた横断歩道標示の規格値とに基づいて算出する周期算出部と、
前記ペイント部分の周期の整数倍を1周期とした基底関数に基づいて、前記設定範囲における画像輝度の周波数分析を行い、周波数パワー分布を算出する周波数分析部と、
前記設定範囲から、前記周波数パワー分布におけるパワーレベルが所定の閾値より大きい座標点を抽出し、前記座標点が存在する領域を横断歩道標示領域として検出する領域検出部と、
前記画像データにおける前記横断歩道標示領域の座標データを出力する位置出力部と、
を備える横断歩道標示検出装置。 - 前記周期算出部は、
前記第一の撮像系の幾何学的パラメータ、及び前記横断歩道標示の規格値を予め記憶するメモリ部と、
前記画像データ取得部から前記設定範囲の座標を取得し、前記第一の撮像系の幾何学的パラメータと前記横断歩道表示の規格値とを前記メモリ部から取得するパラメータ取得部と、
前記設定範囲の座標と、前記第一の撮像系の幾何学的パラメータと、前記横断歩道表示の規格値とに基づいて、前記ペイント部分の周期を算出する算出部と、
を備える請求項1に記載の横断歩道標示検出装置。 - 前記周期算出部は、
前記第一の撮像系の幾何学的パラメータ、及び前記横断歩道標示の規格値を予め記憶するメモリ部と、
前記画像データ取得部から前記設定範囲の座標を取得し、前記第一の撮像系の幾何学的パラメータと前記横断歩道表示の規格値とを前記メモリ部から取得するパラメータ取得部と、
角速度の変化を検出するセンサを有し、前記センサが検出した角速度の変化が所定の閾値よりも大きい場合に、前記横断歩道検出装置の旋回を検出する旋回検出部と、
前記横断歩道検出装置の旋回が検出された場合に、前記センサが検出した角速度の変化値に基づいて、旋回角度を算出する角度算出部と、
前記設定範囲の座標と、前記第一の撮像系の幾何学的パラメータと、前記横断歩道表示の規格値と、前記旋回角度とに基づいて、前記ペイント部分の周期を算出する算出部と、
を備える請求項1に記載の横断歩道標示検出装置。 - 前記画像データに含まれる設定範囲の座標と、前記第一の撮像系の幾何学的パラメータとに基づいて、前記第一の撮像系と、前記設定範囲が示す路面との実空間における距離を距離データとして算出する距離算出部とを、更に備え、
前記周期算出部は、前記設定範囲に横断歩道標示のペイント部分が含まれる場合の前記ペイント部分の周期を、前記距離データと、横断歩道標示の規格値と、に基づいて算出し、
前記位置出力部は、前記画像データにおける前記横断歩道標示領域の座標データ、及び前記距離データを出力する、
請求項1に記載の横断歩道標示検出装置。 - ステレオマッチング部と、距離算出部とを、更に備え、
前記画像データ取得部は、更に、第二の撮像系を有し、前記第一の撮像系を用いて路面を撮像した基準画像データと、前記第二の撮像系を用いて前記路面を撮像した参照画像データと、を取得し、
前記ステレオマッチング部は、前記基準画像データが有する基準点と、前記参照画像データが有する複数の探索点との、画像輝度の相違度を示す評価値分布を算出し、前記評価値分布に基づいて算出される視差と、前記第一の撮像系の焦点距離と、前記第一の撮像系と前記第二の撮像系との距離と、に基づいて、前記第一の撮像系と前記基準画像データに撮像された路面との実空間における距離を第一の距離データとして算出し、
前記距離算出部は、前記第一の距離データに基づいて前記基準画像データの3次元ボクセルデータを算出し、前記3次元ボクセルデータに基づいて、前記第一の撮像系と、前記基準画像データに含まれる設定範囲が示す路面との実空間における距離を第二の距離データとして算出し、
前記周期算出部は、前記設定範囲に横断歩道標示のペイント部分が含まれる場合の前記ペイント部分の周期を、前記第二の距離データと、横断歩道標示の規格値と、に基づいて算出し、
前記位置出力部は、前記基準画像のデータにおける前記横断歩道標示領域の座標データと、及び前記第二の距離データを出力する、
請求項1に記載の横断歩道標示検出装置。 - 前記第一の撮像系および前記第二の撮像系の幾何学的パラメータは、撮像系の設置ピッチ角と、設置高と、焦点距離と、画素ピッチと、を含むパラメータである、
請求項1、4及び5のいずれかに記載の横断歩道標示検出装置。 - 前記周波数分析部は、1周期が前記ペイント部分の周期の2倍以上である基底関数に基づいて、周波数分析を行う、
請求項1、4及び5のいずれかに記載の横断歩道標示検出装置。 - 前記領域検出部は、前記車両の周辺の照度に基づいて、前記所定の閾値を制御する、
請求項1、4及び5のいずれかに記載の横断歩道標示検出装置。 - 撮像系を用いて撮像した路面の画像データを取得し、
前記画像データの設定範囲に横断歩道標示のペイント部分が含まれる場合の前記ペイント部分の周期を、前記設定範囲の座標と、前記撮像系の幾何学的パラメータと、横断歩道標示の規格値と、に基づいて算出し、
前記ペイント部分の周期の整数倍を1周期とした基底関数に基づいて、前記設定範囲における画像輝度の周波数分析を行い、周波数パワー分布を算出し、
前記設定範囲から、前記周波数パワー分布におけるパワーレベルが所定の閾値より大きい座標点を抽出し、前記座標点が存在する領域を横断歩道標示領域として検出し、
前記画像データにおける前記横断歩道標示領域の座標データを出力する横断歩道標示検出方法。 - 前記撮像系の幾何学的パラメータは、前記撮像系の設置ピッチ角と、設置高と、焦点距離と、画素ピッチと、を含むパラメータである、
請求項9に記載の横断歩道標示検出方法。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/392,148 US8744131B2 (en) | 2009-09-29 | 2010-09-22 | Pedestrian-crossing marking detecting method and pedestrian-crossing marking detecting device |
JP2011534060A JP5548212B2 (ja) | 2009-09-29 | 2010-09-22 | 横断歩道標示検出方法および横断歩道標示検出装置 |
CN201080038501.9A CN102483881B (zh) | 2009-09-29 | 2010-09-22 | 人行横道线检测方法及人行横道线检测装置 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2009223976 | 2009-09-29 | ||
JP2009-223976 | 2009-09-29 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2011039977A1 true WO2011039977A1 (ja) | 2011-04-07 |
Family
ID=43825829
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2010/005741 WO2011039977A1 (ja) | 2009-09-29 | 2010-09-22 | 横断歩道標示検出方法および横断歩道標示検出装置 |
Country Status (4)
Country | Link |
---|---|
US (1) | US8744131B2 (ja) |
JP (1) | JP5548212B2 (ja) |
CN (1) | CN102483881B (ja) |
WO (1) | WO2011039977A1 (ja) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20130086819A (ko) * | 2012-01-26 | 2013-08-05 | 현대엠엔소프트 주식회사 | 차량 mms를 이용한 과속방지턱 정보 취득방법 |
JP2017207973A (ja) * | 2016-05-19 | 2017-11-24 | パナソニックIpマネジメント株式会社 | 検出装置および検出方法 |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103679120B (zh) * | 2012-09-11 | 2016-12-28 | 株式会社理光 | 不平道路的检测方法和系统 |
JP6544445B2 (ja) * | 2016-01-29 | 2019-07-17 | 日産自動車株式会社 | 車両の走行制御方法および車両の走行制御装置 |
CN108319895B (zh) * | 2017-12-29 | 2021-09-17 | 百度在线网络技术(北京)有限公司 | 用于识别电子地图中的路口的方法和装置 |
JP2022112897A (ja) * | 2021-01-22 | 2022-08-03 | トヨタ自動車株式会社 | 情報処理装置及び情報処理方法 |
JP7480752B2 (ja) * | 2021-06-11 | 2024-05-10 | トヨタ自動車株式会社 | 方法、標示装置、及びシステム |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0855221A (ja) * | 1994-08-13 | 1996-02-27 | Tsunehiko Nakanishi | 移動物体認識方法及びそれを用いた移動物体検出装置 |
WO2008065707A1 (fr) * | 2006-11-28 | 2008-06-05 | Fujitsu Limited | Procédé de reconnaissance de données d'image, dispositif de traitement d'image et programme de reconnaissance de données d'image |
JP2009180631A (ja) * | 2008-01-31 | 2009-08-13 | Denso It Laboratory Inc | ナビゲーション装置、ナビゲーション方法およびプログラム |
Family Cites Families (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6370260B1 (en) * | 1999-09-03 | 2002-04-09 | Honeywell International Inc. | Near-IR human detector |
JP2005078528A (ja) * | 2003-09-02 | 2005-03-24 | Honda Motor Co Ltd | 物体検出装置及び方法 |
EP1876411A4 (en) * | 2005-04-25 | 2011-06-29 | Geo Technical Lab Co Ltd | METHOD FOR ANALYZING THE PICTURE POSITION |
JP4659631B2 (ja) * | 2005-04-26 | 2011-03-30 | 富士重工業株式会社 | 車線認識装置 |
JP4568637B2 (ja) | 2005-04-26 | 2010-10-27 | 富士重工業株式会社 | 道路認識装置 |
AT502551B1 (de) * | 2005-06-15 | 2010-11-15 | Arc Seibersdorf Res Gmbh | Verfahren und bildauswertungseinheit zur szenenanalyse |
JP4755012B2 (ja) * | 2005-06-29 | 2011-08-24 | 株式会社エヌ・ティ・ティ・ドコモ | 映像評価装置、時空間周波数解析装置、映像評価方法、時空間周波数解析方法、映像評価プログラム、および時空間周波数解析プログラム |
JP4822766B2 (ja) * | 2005-08-18 | 2011-11-24 | 富士通株式会社 | 路面標示認識装置及びシステム |
JP5022609B2 (ja) * | 2006-02-27 | 2012-09-12 | 日立オートモティブシステムズ株式会社 | 撮像環境認識装置 |
JP4724043B2 (ja) * | 2006-05-17 | 2011-07-13 | トヨタ自動車株式会社 | 対象物認識装置 |
JP4654163B2 (ja) * | 2006-07-14 | 2011-03-16 | 日立オートモティブシステムズ株式会社 | 車両の周囲環境認識装置及びシステム |
US8204278B2 (en) * | 2006-11-28 | 2012-06-19 | Fujitsu Limited | Image recognition method |
JP4569837B2 (ja) * | 2007-03-30 | 2010-10-27 | アイシン・エィ・ダブリュ株式会社 | 地物情報収集装置及び地物情報収集方法 |
JP4437556B2 (ja) * | 2007-03-30 | 2010-03-24 | アイシン・エィ・ダブリュ株式会社 | 地物情報収集装置及び地物情報収集方法 |
JP4446201B2 (ja) * | 2007-03-30 | 2010-04-07 | アイシン・エィ・ダブリュ株式会社 | 画像認識装置及び画像認識方法 |
JP4623135B2 (ja) * | 2008-05-08 | 2011-02-02 | 株式会社デンソー | 画像認識装置 |
US8953838B2 (en) * | 2009-06-24 | 2015-02-10 | Here Global B.V. | Detecting ground geographic features in images based on invariant components |
-
2010
- 2010-09-22 WO PCT/JP2010/005741 patent/WO2011039977A1/ja active Application Filing
- 2010-09-22 US US13/392,148 patent/US8744131B2/en not_active Expired - Fee Related
- 2010-09-22 CN CN201080038501.9A patent/CN102483881B/zh not_active Expired - Fee Related
- 2010-09-22 JP JP2011534060A patent/JP5548212B2/ja not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0855221A (ja) * | 1994-08-13 | 1996-02-27 | Tsunehiko Nakanishi | 移動物体認識方法及びそれを用いた移動物体検出装置 |
WO2008065707A1 (fr) * | 2006-11-28 | 2008-06-05 | Fujitsu Limited | Procédé de reconnaissance de données d'image, dispositif de traitement d'image et programme de reconnaissance de données d'image |
JP2009180631A (ja) * | 2008-01-31 | 2009-08-13 | Denso It Laboratory Inc | ナビゲーション装置、ナビゲーション方法およびプログラム |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20130086819A (ko) * | 2012-01-26 | 2013-08-05 | 현대엠엔소프트 주식회사 | 차량 mms를 이용한 과속방지턱 정보 취득방법 |
KR101933580B1 (ko) * | 2012-01-26 | 2018-12-28 | 현대엠엔소프트 주식회사 | 차량 mms를 이용한 과속방지턱 정보 취득방법 |
JP2017207973A (ja) * | 2016-05-19 | 2017-11-24 | パナソニックIpマネジメント株式会社 | 検出装置および検出方法 |
Also Published As
Publication number | Publication date |
---|---|
US8744131B2 (en) | 2014-06-03 |
JP5548212B2 (ja) | 2014-07-16 |
CN102483881A (zh) | 2012-05-30 |
CN102483881B (zh) | 2014-05-14 |
JPWO2011039977A1 (ja) | 2013-02-21 |
US20120148104A1 (en) | 2012-06-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11854272B2 (en) | Hazard detection from a camera in a scene with moving shadows | |
EP2650639B1 (en) | Disparity calculation apparatus | |
JP5548212B2 (ja) | 横断歩道標示検出方法および横断歩道標示検出装置 | |
US9697421B2 (en) | Stereoscopic camera apparatus | |
KR100816377B1 (ko) | 호프 변환을 이용한 주차구획 인식 방법, 장치 및 그를이용한 주차 보조 시스템 | |
JP6015257B2 (ja) | 偏光画像による車両位置検知方法及びシステム | |
JP2012159469A (ja) | 車両用画像認識装置 | |
JP2007234019A (ja) | 車両画像領域特定装置およびその方法 | |
JP2013186663A (ja) | 横断歩道認識装置および横断歩道認識方法 | |
JP2013186664A (ja) | 横断歩道認識装置および横断歩道認識方法 | |
US20200193184A1 (en) | Image processing device and image processing method | |
JP2011170568A (ja) | 障害物検出装置、及び障害物検出方法 | |
JP5091897B2 (ja) | 停止線検出装置 | |
JP2012159470A (ja) | 車両用画像認識装置 | |
JP2016162130A (ja) | 横断歩道検出装置、横断歩道検出方法及び横断歩道検出用コンピュータプログラム | |
WO2022009537A1 (ja) | 画像処理装置 | |
JP2011076248A (ja) | 走行区分線検出装置 | |
TWI588444B (zh) | 路面檢測方法、路面檢測裝置與路面檢測系統 | |
JP5579297B2 (ja) | 視差算出方法、および視差算出装置 | |
JP2012089005A (ja) | 画像処理装置、及び画像処理方法 | |
JP4381394B2 (ja) | 障害物検知装置及びその方法 | |
JP2008234178A (ja) | 表示装置 | |
EP2958054B1 (en) | Hazard detection in a scene with moving shadows | |
JP2020095625A (ja) | 画像処理装置および画像処理方法 | |
KR20130094535A (ko) | 횡단 보도를 검출하는 장치 및 방법 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 201080038501.9 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 10820102 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2011534060 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13392148 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 10820102 Country of ref document: EP Kind code of ref document: A1 |