Distance measurement system and vehicle equipped with same

Abstract

An objective of the present invention is to provide technology for accurately detecting the accuracy of a distance measurement system without acquiring a binocular image as a basis for correction in advance. A representative distance measurement system according to the present invention comprises: a rail unevenness information generation unit for detecting the unevenness of a rail and the surroundings thereof; a first rail position detection unit for outputting a rail position on the basis of the output of the rail unevenness information generation unit; a rail width calculation unit for outputting rail width information on the basis of the output of the first rail position detection unit; a first rail distance calculation unit for outputting a first rail distance on the basis of the rail width information; a second rail distance calculation unit for outputting a second rail distance on the basis of the output of the distance calculation unit; a distance error calculation unit for comparing the first rail distance and the second rail distance and outputting an error; and a correction amount calculation unit for outputting the correction amount of the output of the distance calculation unit on the basis of the output of the distance error calculation unit.

Description

Distance measurement system and vehicle equipped with same

[0001] The present invention relates to a distance measurement system and a vehicle equipped with the same.

[0002] From the perspective of ensuring safety, mobile objects need to accurately confirm their own position and reliably detect obstacles. For this reason, technology that can accurately measure distances in the direction of travel of a mobile object is becoming increasingly important. Stereo cameras have traditionally been used as one of the distance measurement methods. However, it is known that misalignment of the optical axis of stereo cameras can occur due to various factors, such as deterioration over time due to vibrations during installation or while the object is moving, resulting in a decrease in the accuracy of distance measurements. To address this decrease in accuracy, technologies to correct distance measurement systems and improve accuracy are being investigated.

[0003] For example, Patent Document 1 discloses a technology including: "an image input unit having a stereo camera that captures an image in the direction of travel of a train, and inputting a pre-running binocular image captured by the stereo camera before the train runs and a running binocular image captured while the train is running; a landmark detection unit that detects a plurality of landmarks around the train from the pre-running binocular image and detects one landmark around the train from the running binocular image; and a calibration unit that performs a first calibration calculation on the pre-running binocular image based on position information of the plurality of landmarks around the train and performs a second calibration calculation on the running binocular image based on position information of the one landmark around the train, where the calibration unit compares estimated position information of the one landmark around the train calculated from the running binocular image corrected based on the first calibration calculation performed before the train runs with reference position information that is a reference position of the one landmark around the train, and if the deviation amount of the estimated position information from the reference position information is equal to or greater than a threshold, corrects the running binocular image based on the second calibration calculation."

[0004] International Publication No. 2021 / 044564

[0005] However, in Patent Document 1, in order to calibrate the forward monitoring device, it is necessary to acquire binocular images of landmarks that serve as a correction reference in advance, which results in a large burden of advance preparation when performing the calibration. Furthermore, in the calibration of Patent Document 1, since the correction is performed relatively based on information about landmarks acquired in advance, if the accuracy of the information about the landmarks acquired in advance is low, there is a problem that the accuracy of the corrected distance measurement result also remains low. Therefore, an object of the present invention is to provide a technology for accurately detecting the accuracy of a distance measurement system without acquiring binocular images that serve as a correction reference in advance.

[0006] In order to solve the above problems, one representative distance measurement system of the present invention is a distance measurement system equipped with a rail information acquisition unit that acquires rail information from multiple directions and a distance calculation unit that performs distance calculations by stereo matching, and is characterized by comprising: a rail unevenness information generation unit that detects unevenness on the rail and its surroundings; a first rail position detection unit that outputs the rail position based on the output of the rail unevenness information generation unit; a rail width calculation unit that outputs rail width information based on the output of the first rail position detection unit; a first rail distance calculation unit that outputs a first rail distance based on the rail width information; a second rail distance calculation unit that outputs a second rail distance based on the output of the distance calculation unit; a distance error calculation unit that compares the first rail distance with the second rail distance and outputs an error; and a correction amount calculation unit that outputs a correction amount for the output of the distance calculation unit based on the output of the distance error calculation unit.

[0007] According to the present invention, it is possible to provide a technology for accurately detecting the accuracy of a distance measurement system without acquiring binocular images that serve as a reference for correction in advance. Problems, configurations, and effects other than those described above will become apparent from the following description of the preferred embodiment of the invention.

[0008] FIG. 1 is a schematic diagram showing a distance measurement system according to a first embodiment. FIG. 2 is a diagram schematically showing rail unevenness information generated by a rail unevenness information generation unit. FIG. 3 is a graph in which four rows of FIG. 2 are extracted and the vertical axis is plotted as a disparity value. FIG. 4 is a diagram in which FIG. 2 is normalized and linearly approximated. FIG. 5 is a diagram in which the column addresses of each row of FIG. 4 are quantitatively shifted and rearranged. FIG. 6 is a graph in which region Y of FIG. 5 is extracted. FIG. 7 is a cross-sectional view of region X of FIG. 3 in the disparity value direction. FIG. 8 is a diagram showing matters taken into consideration by a reliability evaluation unit. FIG. 9 is a diagram showing the relationship between a distance measurement system mounted on a vehicle and a railroad or road surface. FIG. 10 is a flowchart showing processing in the distance measurement system. FIG. 11 is a schematic diagram showing a distance measurement system according to a second embodiment.

[0009] Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, the present invention is not limited to these embodiments. In the drawings, identical parts are denoted by the same reference numerals. When multiple components having the same or similar functions are present, they may be described using the same reference numerals with different subscripts. Furthermore, when it is not necessary to distinguish between these multiple components, the subscripts may be omitted. Furthermore, although terms such as "first," "second," and "third" may be used to describe various elements or components in this disclosure, it should be understood that these elements or components should not be limited by these terms. These terms are used only to distinguish one element or component from another. Therefore, a first element or component discussed below may also be referred to as a second element or component without departing from the teachings of the inventive concept. The position, size, shape, range, etc. of each component shown in the drawings may not represent the actual position, size, shape, range, etc., in order to facilitate understanding of the invention. Therefore, the present invention is not necessarily limited to the position, size, shape, range, etc. disclosed in the drawings.

[0010] [Example 1] In Example 1, an example of a distance measurement system equipped with a rail information acquisition unit will be described. In Example 1, a stereo camera will be used as the rail information acquisition unit. First, each functional block will be described.

[0011] (Configuration) FIG. 1 is a diagram showing a schematic configuration of a distance measurement system 100 equipped with a rail information acquisition unit. The distance measurement system 100 includes a right camera 200, a left camera 300, a first geometric correction unit 201, a second geometric correction unit 301, a stereo matching unit 400, a distance calculation unit 410, a three-dimensional object detection unit 420, a rail unevenness information generation unit 500, a first rail position detection unit 510, a second rail position detection unit 600, a reliability evaluation unit 700, a rail width calculation unit 710, a first rail distance calculation unit 720, a distance error calculation unit 730, a correction amount calculation unit 740, and a second rail distance calculation unit 810. Thus, the distance measurement system 100 measures the distance between the vehicle and the rails on its path. Although a stereo camera is used as the distance measurement sensor serving as the rail information acquisition unit in the first embodiment, other distance measurement sensors may also be used. For example, LiDAR (light detection and ranging) may be used.

[0012] The stereo camera includes two imaging units (right camera 200, left camera 300) and generates images captured synchronously from different viewpoints. In other words, it functions as a rail information acquisition unit that acquires rail information of rails acquired simultaneously from two different directions. The right camera 200 constitutes a right-side camera as one of the imaging units of the stereo camera, and similarly, the left camera 300 constitutes a left-side camera as one of the imaging units of the stereo camera. Note that, although the stereo camera includes two imaging units in the first embodiment, it may also be configured to include any other number of imaging units.

[0013] The right camera 200 connects and transmits the captured image to the first geometric correction unit 201. Similarly, the left camera 300 connects and transmits the captured image to the second geometric correction unit 301.

[0014] The left camera 300 and the right camera 200 constitute a stereo camera, but can also be configured to function as a monocular camera by switching a switch or the like. This configuration is an example, and the embodiment is not limited to this. A configuration in which a sensor for the stereo camera and a sensor for the monocular camera are provided separately may also be used.

[0015] The first geometric correction unit 201 corrects the image captured by the right camera 200 and transmits the corrected image to the stereo matching unit 400. Specifically, the first geometric correction unit 201 first performs distortion correction on the image captured by the right camera 200 to remove image distortion. This is because images captured by a camera are generally distorted radially from the center of the image due to lens distortion, and similar distortion occurs in the image captured by the right camera 200, so this distortion is intended to prevent the generation of parallax images by the stereo matching unit 400, which will be described later, from being affected.

[0016] Next, the first geometric correction unit 201 performs rectification so that corresponding points on the image captured by the right camera 200 and the image captured by the left camera 300 have the same row coordinates. This is because, as in this embodiment, slight misalignment occurs when images captured by two imaging units at different angles or distances are superimposed, and rectification in advance reduces the load on the stereo matching unit 400, which will be described later.

[0017] The second geometric correction unit 301 corrects the image captured by the left camera 300 in the same manner as the first geometric correction unit 201 , and transmits the corrected image to the stereo matching unit 400 .

[0018] The stereo matching unit 400 generates a parallax image from the images corrected by the first geometric correction unit 201 and the second geometric correction unit 301. Next, the stereo matching unit 400 generates a parallax image including a parallax value from the calculated parallax.

[0019] The distance calculation unit 410 generates a distance image from the parallax image generated by the stereo matching unit 400. Specifically, the distance calculation unit 410 converts the parallax value of each pixel included in the parallax image into a distance based on the 2D3D conversion parameters, generates a distance image, and transmits it to the three-dimensional object detection unit 420. Furthermore, the distance calculation unit 410 may generate the distance image taking into account the correction amount calculated by the correction amount calculation unit 740, as described below.

[0020] The three-dimensional object detection unit 420 detects three-dimensional objects, such as obstacles, from the distance image. Specifically, the three-dimensional object detection unit 420 performs noise removal and grouping from three-dimensional coordinates including distance based on the distance information contained in the distance image. In this way, the three-dimensional object detection unit 420 detects three-dimensional objects that are obstacles to a running train.

[0021] The rail unevenness information generator 500 extracts the rail and its surrounding area from the parallax image generated by the stereo matching unit 400, and generates rail unevenness information based on the parallax values. At this time, the generated rail unevenness information is sent to the first rail position detector 510.

[0022] The first rail position detector 510 detects the rail position based on the detected unevenness of the rail. At this time, the detected rail position is defined as the first rail position.

[0023] The second rail position detection unit 600 detects the rail position from the image captured by the left camera 300. At this time, the detected rail position is set as the second rail position. Specifically, the second rail position detection unit 600 may detect the position of the second rail from the image captured by the left camera 300 by machine learning. Furthermore, in the first embodiment, when the stereo camera generates a parallax image, the image captured by the left camera 300 is used as the reference image, and therefore the rail position is detected from the image captured by the left camera 300. However, if the reference image is the right camera 200, it is desirable to detect the rail position from the image captured by the right camera 200 on the same side.

[0024] Furthermore, the second rail position detection unit 600 may detect the rail position using the parallax image as a simple image, or may detect the second rail position by searching for changes in color and contrast contained in the image.

[0025] The reliability evaluation unit 700 calculates the accuracy of the detected rail position as a reliability evaluation value, and if the reliability evaluation value is equal to or greater than a predetermined value, it determines that the detected rail position is reliable.

[0026] The rail width calculation unit 710 calculates the rail width based on the reliable rail position information transmitted from the reliability evaluation unit 700. That is, the rail width calculation unit 710 calculates the rail width in terms of the number of pixels in the parallax image or the image based on the reliable rail position information transmitted from the reliability evaluation unit 700. Here, the rail width is determined for each line on which the train runs, and this value can be acquired in advance. Therefore, the distance to the target rail portion can be calculated using the difference between the column coordinates of the left rail and the column coordinates of the right rail, the focal length, and information on the normalized rail width. Furthermore, instead of acquiring the rail width as a value in advance, the type of track gauge of the railway may be acquired and the value calculated from the type of track gauge. For example, if the acquired type of track gauge is standard gauge, the rail width may be 1,435 mm, and if the acquired type is narrow gauge, the rail width may be 1,067 mm. Furthermore, the rail width may be acquired using information acquired during maintenance work on the line. In addition, the rail width calculation unit may set a weight on at least one of the rail position determined by the first rail position detection unit 510 and the rail position determined by the second rail position detection unit 600 output by the reliability evaluation unit 700, thereby calculating the rail width based on the rail position with the highest reliability.

[0027] The first rail distance calculation unit 720 calculates the rail distance from the stereo camera to the rail based on the rail width based on the reliable rail position and the focal length of the stereo camera. At this time, the calculated rail distance is set as the first rail distance. Furthermore, the rail width calculation unit 710 outputs the rail width only when a reliable rail position is output from the reliability evaluation unit 700. For this reason, the rail width calculation unit 710 does not always output the rail width, and the output rail width is intermittent data. Therefore, the output of the first rail distance calculation unit 720, which receives the rail width as input, is also intermittent and becomes discontinuous information.

[0028] The second rail distance calculation unit 810 calculates the rail distance from the stereo camera to the rail using the parallax information. Specifically, the second rail distance calculation unit 810 obtains timing information when the first rail position is output from the first rail position detection unit 510, and in accordance with this timing, obtains rail distance information based on the distance image from the distance calculation unit 410 and outputs it to the distance error calculation unit 730. In other words, the second rail distance calculation unit 810 continuously calculates the second rail distance and transmits it to the distance error calculation unit 730.

[0029] The distance error calculation unit 730 calculates the difference between the first rail distance and the second rail distance.

[0030] The correction amount calculation unit 740 calculates a distance correction value based on the distance error information calculated by the distance error calculation unit 730. The calculated distance correction value is transmitted to the distance calculation unit 410 and is used as a correction value when the distance calculation unit 410 calculates a distance image from the parallax value.

[0031] (Details of Rail Unevenness Information) Next, rail unevenness information will be described with reference to FIGS. 2 and 3. FIG. 2 is a diagram schematically illustrating rail unevenness information generated by the rail unevenness information generator 500. FIG. 3 is a graph in which four rows (row A-A', row B-B', row C-C', and row D-D') of FIG. 2 are extracted and plotted as disparity values ​​on the vertical axis. As rail unevenness information, the rail unevenness information generator 500 generates a diagram showing, in terms of disparity, the distance from a stereo camera mounted several meters above the road surface. For example, FIG. 2 is a diagram generated as rail unevenness information, showing a disparity image obtained when a stereo camera is mounted 0.5 meters above the ground on a vehicle. The diagram generated as rail unevenness information is a disparity image in which larger disparity values ​​are displayed in darker colors and smaller disparity values ​​are displayed in lighter colors. Because disparity is greater in areas closer to the front, areas closer to the bottom of the diagram are displayed in colors closer to black.

[0032] In FIG. 2, the upper surface of the rail 910 is located higher than the road surface 920, and in the AA' line indicated by the dashed line in FIG. 2, the rail 910 is depicted as standing out in a color closer to black than the road surface 920.

[0033] In addition, in FIG. 3, in which four rows (row A-A', row B-B', row C-C', and row D-D') in FIG. 2 are extracted with the number of horizontal pixels on the horizontal axis and the parallax value on the vertical axis, the part indicated by area X in FIG. 2 is expressed with a changing parallax value as shown by area X in FIG. 3.

[0034] (Details of Detection of First Rail Position) Next, detection of the first rail position will be described with reference to FIGS. 4 and 6. FIG. 4 is a diagram in which FIG. 2 is normalized and linearly approximated. FIG. 5 is a diagram in which the column addresses of each row in FIG. 4 are quantitatively shifted and rearranged. Furthermore, FIG. 6 is a graph in which region Y in FIG. 5 is enlarged and expressed as a disparity value for each pixel. The first rail position detection unit 510 detects the first rail position by detecting the outer and inner edges of the rail (the edges of the upper surface of the rail) based on rail unevenness information. In detection using parallax, the position where the disparity changes can be detected with extremely high accuracy compared to the detection accuracy of the parallax value itself.

[0035] Therefore, by removing noise from the rail unevenness information as shown in FIG. 2, it becomes possible to detect the first rail position with high accuracy.

[0036] Specifically, the rail unevenness information generator 500 generates rail unevenness information by using edge coordinates of different rows to normalize the disparity value of each pixel by the road surface disparity of each row and linearly approximating the normalized value, as shown in Figure 4. That is, the rail unevenness information generator 500 calculates the road surface disparity from the disparity image, subtracts the road surface disparity from the disparity value of each pixel in the disparity image, identifies the approximate position of the rail from a group of consecutive pixels having a disparity value that is positive or equal to or greater than a predetermined value, and detects the inner edge or outer edge of the rail from consecutive pixels around the approximate rail position where the disparity value is negative or equal to or less than the predetermined value. This makes it possible to maintain a constant edge threshold, thereby reducing the load on the first rail position detector 510 for detecting the first rail position and improving detection accuracy.

[0037] As another method for improving detection accuracy, the rail unevenness information generator 500 may rearrange the rails by quantitatively shifting the column addresses of each row so that the rails are aligned vertically on the parallax image, as shown in Fig. 5. This allows the parallax values ​​to be added or averaged in the column direction.

[0038] For example, even if the parallax value does not indicate unevenness in region Y where row E-E' in FIG. 4 intersects with rail 910, the rearranged rail unevenness information, for example, region Y in FIG. 5, stably indicates the average rail unevenness as shown in FIG. 6. As a result, if the standard deviation in FIG. 6 is at its minimum value, it can be determined that the quantitative value of the shifted column address is correct. On the other hand, since the rail width is based on the width of the inner side of the rail, the inner side of the rail can also be detected as the rail position. Note that the rail edge position in the parallax image may not match the rail edge position in the image after geometric correction. In such cases, the rail unevenness information generation unit 500 must take into consideration factors such as making corrections according to the block size when the stereo matching unit 400 performs stereo matching.

[0039] (Details of Calculation of Reliability Evaluation Value) Next, details of calculation of the reliability evaluation value will be described with reference to Fig. 7. Fig. 7 is a graph showing an overlapping display of the first rail position based on the unevenness of the rail detected by the first rail position detection unit 510 and the second rail position detected by the second rail position detection unit 600. In Fig. 7, the horizontal axis represents the number of pixels in the horizontal direction, and the vertical axis represents the parallax value. In Fig. 7, the solid line represents the first rail position based on the unevenness of the rail detected by the first rail position detection unit 510, and the dashed line represents the second rail position detected by the second rail position detection unit 600.

[0040] The first rail position is based on the unevenness of the rail, and is therefore shown in Fig. 7 as uneven portions where the parallax values ​​of the outer and inner edge portions of the rail are higher than those of the road surface portion. In contrast, the second rail position is detected based on the image captured by the left camera 300, and is shown in Fig. 7 as a dashed straight line.

[0041] The reliability evaluation unit 700 compares the first rail position with the second rail position, and if the difference between the first rail position and the second rail position (i.e., the relative position differences a, b in Figure 7) is less than a certain value, it determines that the reliability evaluation value is equal to or greater than a predetermined value and determines that the rail position is a reliable rail position. Then, the reliability evaluation unit 700 outputs at least one of the rail position determined by the first rail position detection unit 510 or the rail position determined by the second rail detection unit as a reliable rail position.

[0042] When the rail width calculation unit 710 calculates the rail width based on reliable rail position information, the correction amount calculation unit 740 (described later) can generate a correction value with enough accuracy to be used by the distance calculation unit 410.

[0043] 7, the second rail position is located less than a certain distance from the uneven portion of the first rail position. However, if the second rail position is located outside the outer and inner edges of the first rail position, rather than halfway between them, the detected rail position will be determined to be unreliable. Furthermore, the detected rail position may also be determined to be unreliable if the ratio of the second rail position's relative position a from the outer side of the first rail position to its relative position b from the inner side differs by more than a certain amount, or if the detected rail positions differ between the left and right rails.

[0044] In other words, if the second rail position is located at a position less than a certain value from the uneven portion of the first rail position, and the relative position a from the outside of the first rail position at the second rail position and the relative position b from the inside of the second rail position on the left and right rails do not differ by more than a certain value, the reliability evaluation unit 700 determines that the difference between the first rail position and the second rail position is less than a certain value, i.e., the reliability evaluation value is greater than or equal to a predetermined value.

[0045] Alternatively, the reliability evaluation unit 700 may detect the first rail position as the center of the positions of the inner and outer edges of the uneven portion of the first rail position. In this case, the difference between the first rail position and the second rail position is the difference between the center position and the second rail position. Note that when the rail position is determined to be the center of the edges in this manner, the rail width can be calculated from the relative distance between the left and right rail positions and the top surface width of the rail (a known value). Whether the rail width is the width between the inner sides of the rail or the width between the outer sides of the rail can be selected appropriately depending on the rail width definition information.

[0046] Furthermore, if the difference between the first rail position and the second rail position is greater than a certain value, i.e., if the reliability evaluation value is less than a predetermined value, the reliability evaluation unit 700 stops outputting the rail position information, preventing unreliable rail position information from being sent to the rail width calculation unit 710.

[0047] Next, with reference to FIG. 8 , matters that the reliability evaluation unit 700 takes into consideration when calculating a reliability evaluation value will be described. FIG. 8 is a diagram showing matters that the reliability evaluation unit 700 takes into consideration. In order to ensure the detection accuracy of the first rail position detection unit 510 and the second rail position detection unit 600, the reliability evaluation unit 700 may determine whether or not to calculate a reliability evaluation value by taking into consideration, for example, the weather, the time of day, the tilt of the sun, etc. In other words, as shown in FIG. 8 , the reliability evaluation unit 700 may be connected to components other than the first rail position detection unit 510 and the second rail position detection unit 600, such as the system controller 10, the right camera 200, and the stereo matching unit 400, and may calculate a reliability evaluation value by taking into consideration information acquired from these components.

[0048] For example, the reliability evaluation unit 700 may be connected to the system controller 10 of the distance measurement system 100, and obtain external information a such as the vehicle position and time from the system controller 10, and use the fact that the position and time are suitable for obtaining rail position information to determine whether or not it is possible to calculate a reliability evaluation value.

[0049] Similarly, the reliability evaluation unit 700 may obtain a condition b, such as day or night, backlight, etc., from the image obtained from the right camera 200, and determine whether or not the reliability evaluation value can be calculated.

[0050] Similarly, the reliability evaluation unit 700 may obtain the disparity value of the target area or the effective disparity rate c of the target area (the number of pixels for which the correct disparity is obtained relative to the number of target pixels) from the disparity image obtained from the stereo matching unit 400, thereby improving the accuracy of the reliability evaluation value.

[0051] Similarly, the reliability evaluation unit 700 may calculate the reliability of the input second rail position based on the distance information d regarding the rail obtained from the second rail position detection unit 600, and take this reliability into consideration.

[0052] Furthermore, the reliability evaluation unit 700 may not calculate the difference between the first rail position and the second rail position, and may decide to output either the first rail position or the second rail position as the reliable rail position, depending on the environment.

[0053] Furthermore, the reliability evaluation unit 700 may change the selection and generation method of the input information and output information.

[0054] For example, the reliability evaluation unit 700 may determine the linearity of the rails and the flatness of the road surface (presence or absence of undulations) from the rail positions detected by the first rail position detection unit 510 and the second rail position detection unit 600, and calculate the reliability evaluation unit 700 from the flat, straight road that is most suitable for obtaining the rail position.

[0055] In either case, the reliability evaluation unit 700 may stop outputting data based on the information input to the reliability evaluation unit 700. Furthermore, the reliability evaluation unit 700 may store the calculated reliability evaluation value in a storage unit (not shown) in association with the first rail position or the second rail position. This allows weighting that takes into account the associated value when calculating the reliability evaluation value again. Furthermore, the reliability evaluation value may be calculated by an external information processing device via a communication unit connected to the external information processing device. This reduces the load on the reliability evaluation unit 700. The information stored in the storage unit is not limited to the reliability evaluation value, and data included in the rail information acquisition unit, distance calculation unit 410, rail unevenness information generation unit 500, first rail position detection unit 510, reliability evaluation unit 700, rail width calculation unit 710, first rail distance calculation unit 720, second rail distance calculation unit 810, distance error calculation unit 730, and correction amount calculation unit 740 may be stored in association with at least one other data, and the data in the storage unit may be transmittable to an external device. Furthermore, the information processing performed by an external information processing device may be information processing performed by the distance calculation unit 410, rail unevenness information generation unit 500, first rail position detection unit 510, reliability evaluation unit 700, rail width calculation unit 710, first rail distance calculation unit 720, second rail distance calculation unit 810, distance error calculation unit 730, and correction amount calculation unit 740, and distance measurement may be performed based on the results performed by the external information processing device.

[0056] (Details of Correction of Second Rail Distance) Next, details of distance error calculation will be described with reference to FIG. 9 . FIG. 9 is a diagram showing the relationship between the distance measurement system 100 mounted on a vehicle and the track or road surface. The second rail distance calculation unit 810 finds it difficult to stably obtain the second rail distance because the area of ​​the rail top surface is small. Therefore, the second rail distance is calculated as the distance from the road surface between the target rails. This allows the second rail distance calculation unit 810 to stably calculate the second rail distance. However, since the first rail distance calculation unit 720 calculates the first rail distance based on the parallax value, it calculates the distance from the rail top surface closest to the distance measurement system 100. Therefore, if the second rail distance and the first rail distance are compared as is, a difference will occur between the distances because the measurement end points are different. To prevent this, the second rail distance calculation unit 810 needs to correct the second rail distance and calculate the distance from the rail top surface in the same way as the first rail distance.

[0057] In Fig. 9, the rail 910 is indicated by a solid line, and the road surface 920 is indicated by a dashed line. In this case, A shown in Fig. 9 indicates the distance between the distance measurement system 100 mounted on the train and the road surface 920, i.e., the second rail distance. Also, B shown in Fig. 9 indicates the distance between the distance measurement system 100 mounted on the train and the upper surface of the rail 910, i.e., the first rail distance.

[0058] Here, if the output of the second rail distance calculation unit 810 is Zb, the output of the first rail distance calculation unit 720 is Za, the camera height relative to the road surface is H, and the constant rail height is h, the corrected second rail distance Zb' is expressed by the following formula, and the final distance error is obtained by cutting high frequencies using an LPF or the like. Zb' = Zb / H (H - h) (1)

[0059] (Response to a Failure in the Monitoring System of the Present Disclosure) Next, the processing in the distance measurement system 100 will be described with reference to Fig. 10. Fig. 10 is a flowchart showing the processing in the distance measurement system 100.

[0060] In step S101 , the stereo matching unit 400 acquires an image captured by the right camera 200 and corrected by the first geometric correction unit 201 and an image captured by the left camera 300 and corrected by the second geometric correction unit 301 .

[0061] In step S102, the stereo matching unit 400 generates a parallax image from the acquired images.

[0062] In step S103, the rail unevenness information generator 500 generates rail unevenness information from the parallax image.

[0063] In step S104, the rail positions are detected. Specifically, the first rail position detection unit 510 detects the first rail position from the rail unevenness information, and the second rail position detection unit 600 detects the second rail position from the image captured by the left camera 300 and corrected by the second geometric correction unit 301.

[0064] In step S105, the reliability evaluation unit 700 compares the first rail position with the second rail position and determines the reliability evaluation value of the rail position. If the reliability evaluation value is equal to or greater than a certain value, the process proceeds to S106, and if it is less than the certain value, the process ends.

[0065] In step S106, the rail width is calculated. Specifically, the rail width calculation unit 710 calculates the rail width based on the first rail position or the second rail position transmitted from the reliability evaluation unit 700.

[0066] In step S107, the first rail distance is calculated. Specifically, the first rail distance calculation unit 720 calculates the first rail distance based on the rail width.

[0067] In step S108, the second rail distance is calculated. Specifically, second rail distance calculation unit 810 calculates the second rail distance based on the output of distance calculation unit 410.

[0068] In step S109, the distance error is calculated. Specifically, the distance error calculation unit 730 compares the first rail distance with the second rail distance to calculate the distance error.

[0069] In step S110, the correction amount is calculated. Specifically, the correction amount calculation unit 740 calculates the distance correction value based on the distance error information calculated by the distance error calculation unit 730.

[0070] In step S111, distance calculation unit 410 corrects the distance measurement result based on the calculated distance correction value.

[0071] The above has described the distance measurement system 100 according to Example 1. The distance measurement system 100 of the present disclosure mainly includes a first rail distance calculation unit 720 and a second rail distance calculation unit 810, and can accurately calculate a distance error by comparing a rail position based on unevenness information calculated from parallax, a first rail distance calculated based on a rail width that is accurately determined in advance by a standard, and a second rail distance calculated based on a parallax image, and can accurately correct the distance measurement of the distance measurement system 100 by calculating a correction amount based on the distance error.

[0072] Next, a distance measurement system 100 according to a second embodiment will be described with reference to Fig. 11. Fig. 11 is a schematic diagram showing the distance measurement system 100 according to the second embodiment. In the distance measurement system 100 according to the first embodiment, the difference between the first rail distance and the second rail distance is treated as the distance error of the distance measurement system 100. However, the distance measurement system 100 according to the second embodiment differs from the first embodiment in that the difference between the first rail distance and the second rail distance is treated as the parallax error of the distance measurement system 100. The distance measurement system 100 also includes a right camera 200, a left camera 300, a first geometric correction unit 201, a second geometric correction unit 301, a stereo matching unit 400, a distance calculation unit 410, a three-dimensional object detection unit 420, a rail unevenness information generation unit 500, a first rail position detection unit 510, a second rail position detection unit 600, a reliability evaluation unit 700, a rail width calculation unit 710, a first rail parallax calculation unit 721, a parallax error calculation unit 731, a correction amount calculation unit 741, and a second rail parallax calculation unit 811.

[0073] (Configuration Example) The first rail parallax calculation unit 721 calculates the target rail parallax based on the rail width information sent from the rail width calculation unit 710. At this time, the calculated rail parallax is referred to as the first rail parallax. Specifically, the first rail parallax calculation unit 721 calculates the first rail parallax according to the following formulas (2) and (3), and sends it to the parallax error calculation unit 731. Here, Rd is the first rail parallax, B is the baseline length of the stereo camera, fc is the focal length, Zb is the first rail distance, Rp is the pixel pitch, Rw is the rail width, and Rp is the number of pixels of the rail width on the image. Rd = B × fc / (Zb × Pp) (2) Zb = fc × Rw / Rp (3)

[0074] The second rail parallax calculation unit 811 uses the parallax information to calculate rail parallax corresponding to the rail distance from the distance measurement system 100 to the rail. Specifically, the second rail parallax calculation unit 811 determines a target area from the distance image generated by the distance calculation unit 410, and calculates the rail distance of the target area. At this time, the calculated rail parallax is set as the second rail parallax. Also, similar to the second rail distance calculation unit 810, the target area is acquired from the first rail position detection unit 510.

[0075] The parallax error calculation unit 731 calculates the parallax error of the second rail parallax calculation unit 811. Specifically, the distance error calculation unit 730 calculates the difference between the first rail parallax and the second rail parallax as the parallax error.

[0076] The correction amount calculation unit 741 calculates the amount of parallax correction for the stereo matching unit 400 based on the detection error. The calculated correction amount is transmitted to the stereo matching unit 400. The stereo matching unit 400 creates a distance image in which the parallax value has been corrected based on the transmitted correction amount.

[0077] The above has described the distance measurement system 100 according to Example 2. The distance measurement system 100 of the present disclosure mainly includes a first rail parallax calculation unit 721 and a second rail parallax calculation unit 811, and is capable of calculating a parallax detection error by comparing a first rail parallax calculated based on a rail width defined by a standard with a second rail parallax calculated based on a parallax image, and is capable of correcting the parallax of the distance measurement system 100 by calculating a correction amount based on the detection error.

[0078] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the gist of the present invention.

[0079] The present invention can also take the following forms: (Aspect 1) A distance measurement system equipped with a rail information acquisition unit that acquires rail information from multiple directions and a distance calculation unit that performs distance calculations by stereo matching, comprising: a rail unevenness information generation unit that detects unevenness on the rail and its surroundings, a first rail position detection unit that outputs a rail position based on the output of the rail unevenness information generation unit, a rail width calculation unit that outputs rail width information based on the output of the first rail position detection unit, a first rail distance calculation unit that outputs a first rail distance based on the rail width information, a second rail distance calculation unit that outputs a second rail distance based on the output of the distance calculation unit, a distance error calculation unit that compares the first rail distance with the second rail distance and outputs an error, and a correction amount calculation unit that outputs a correction amount for the output of the distance calculation unit based on the output of the distance error calculation unit. (Aspect 2) The distance measurement system according to Aspect 1, wherein the rail information acquisition unit is a stereo camera, and the rail unevenness information generation unit generates unevenness information based on a parallax image generated from images of the stereo camera. (Aspect 3) The distance measurement system according to Aspect 1 or 2, comprising: a reliability evaluation unit that determines the reliability of the rail position, and a second rail position detection unit that outputs the rail position from an image output by any of the rail information acquisition units, wherein the reliability evaluation unit compares the first rail position obtained by the first rail position detection unit with the second rail position obtained by the second rail position detection unit to calculate a reliability evaluation value of the rail position, and if the reliability evaluation value is equal to or greater than a predetermined value, outputs at least one of the first rail position or the second rail position as a reliable rail position.(Aspect 4) The distance measurement system according to Aspect 3, wherein the reliability evaluation unit calculates the reliability evaluation value based on at least one of the detected straightness or flatness of the rail, the position of a vehicle traveling on the rail, the weather, the time, the inclination of the sun, the detected distance of the rail, and the disparity value or effective disparity rate of the rail and its surroundings. (Aspect 5) The distance measurement system according to Aspect 3 or 4, wherein the first rail position detection unit detects the inner edge position and the outer edge position of the rail and outputs the midpoint between them as the rail position, and the rail width calculation unit outputs the rail width information based on information about the rail position and a normalized rail width. (Aspect 6) The distance measurement system according to any one of Aspects 2 to 5, wherein the rail unevenness information generation unit outputs road surface disparity from the disparity image, and performs linear approximation by normalizing the disparity value of each pixel of the disparity image by the road surface disparity of each row. (Aspect 7) The distance measurement system according to any one of Aspects 3 to 6, wherein the rail width calculation unit calculates the rail width based on a highly reliable rail position by setting a weight on at least one of the rail position determined by the first rail position detection unit or the rail position determined by the second rail position detection unit, both of which are output by the reliability evaluation unit. (Aspect 8) The distance measurement system according to any one of Aspects 3 to 7, comprising: the rail information acquisition unit and a second rail distance calculation unit that calculates the distance to the rail using the rail information, wherein the second rail distance calculation unit corrects the distance Zb obtained from the distance calculation unit to Zb' calculated based on the following equation (1): Zb' = Zb / H (H - h) (1) where H is the camera height from the road surface, h is the rail height, and Zb' is the rail distance obtained by correcting Zb.(Aspect 9) The distance measurement system according to any one of Aspects 3 to 8, further comprising: a storage unit that stores data included in the rail information acquisition unit, the distance calculation unit, the rail unevenness information generation unit, the first rail position detection unit, the reliability evaluation unit, the rail width calculation unit, the first rail distance calculation unit, the second rail distance calculation unit, the distance error calculation unit, and the correction amount calculation unit in association with at least one other data, wherein the data in the storage unit is transmittable to an external device. (Aspect 10) The distance measurement system according to any one of Aspects 3 to 9, further comprising: a communication unit that transmits and receives data included in the distance calculation unit, the rail unevenness information generation unit, the first rail position detection unit, the reliability evaluation unit, the rail width calculation unit, the first rail distance calculation unit, the second rail distance calculation unit, the distance error calculation unit, and the correction amount calculation unit to and from an external device, wherein the distance measurement system is capable of performing distance measurement based on a result of information processing performed externally. (Aspect 11) In the distance measurement system according to any one of Aspects 1 to 10, the first rail distance calculation unit is a first rail parallax calculation unit that calculates a parallax corresponding to the distance to the rail based on the rail width information, the distance error calculation unit is a parallax error calculation unit that compares a parallax output of the first rail parallax calculation unit with a parallax output obtained from a stereo matching unit and outputs a parallax error, and the correction amount calculation unit calculates a correction amount for correcting the parallax information of the stereo matching unit from the parallax error. (Aspect 12) A vehicle equipped with the distance measurement system according to any one of Aspects 1 to 11.(Aspect 13) A distance measurement system comprising a rail information acquisition unit that acquires rail information from a plurality of directions and a distance calculation unit that performs distance calculation by stereo matching, comprising: acquiring rail unevenness information that detects unevenness on the rail and its surroundings by the rail information acquisition unit, detecting a first rail position based on the rail unevenness information, calculating a rail width based on the first rail position, outputting a first rail distance based on the rail width, calculating a second rail distance based on the output of the distance calculation unit, comparing the first rail distance with the second rail distance to calculate an error, and correcting the output of the distance calculation unit based on the error. (Aspect 14) The distance measurement method of Aspect 13, wherein the rail information acquisition unit is a stereo camera, and generating unevenness information based on a parallax image generated from images of the stereo camera. (Aspect 15) In the distance measurement method described in Aspect 13 or 14, the distance measurement system comprises a reliability evaluation unit that calculates a reliability evaluation value that determines the reliability of the rail position, and a second rail position detection unit that outputs the rail position from an image output by any of the rail information acquisition units, wherein the reliability evaluation unit compares the first rail position with the second rail position to calculate a reliability evaluation value of the rail position, and when the reliability evaluation value is equal to or greater than a predetermined value, the reliability evaluation unit outputs at least one of the first rail position or the second rail position as a reliable rail position.

[0080] REFERENCE SIGNS LIST 10 System controller 100 Distance measurement system 200 Right camera 300 Left camera 201 First geometric correction unit 301 Second geometric correction unit 400 Stereo matching unit 410 Distance calculation unit 420 Three-dimensional object detection unit 500 Rail unevenness information generation unit 510 First rail position detection unit 600 Second rail position detection unit 700 Reliability evaluation unit 710 Rail width calculation unit 720 First rail distance calculation unit 721 First rail parallax calculation unit 810 Second rail distance calculation unit 811 Second rail parallax calculation unit 730 Distance error calculation unit 731 Parallax error calculation unit 740, 741 Correction amount calculation unit 910 Rail 920 Road surface

Claims

1. A distance measurement system equipped with a rail information acquisition unit that acquires rail information from multiple directions and a distance calculation unit that performs distance calculations by stereo matching, comprising: a rail unevenness information generation unit that detects unevenness on the rail and its surroundings; a first rail position detection unit that outputs the rail position based on the output of the rail unevenness information generation unit; a rail width calculation unit that outputs rail width information based on the output of the first rail position detection unit; a first rail distance calculation unit that outputs a first rail distance based on the rail width information; a second rail distance calculation unit that outputs a second rail distance based on the output of the distance calculation unit; a distance error calculation unit that compares the first rail distance with the second rail distance and outputs an error; and a correction amount calculation unit that outputs a correction amount for the output of the distance calculation unit based on the output of the distance error calculation unit.

2. A distance measuring system as described in claim 1, characterized in that the rail information acquisition unit is a stereo camera, and the rail unevenness information generation unit generates unevenness information based on a parallax image generated from an image of the stereo camera.

3. A distance measurement system as described in claim 1, comprising: a reliability evaluation unit which judges the reliability of the rail position; and a second rail position detection unit which outputs the rail position from an image output by any of the rail information acquisition units, wherein the reliability evaluation unit compares a first rail position determined by the first rail position detection unit with a second rail position determined by the second rail position detection unit to calculate a reliability evaluation value of the rail position, and when the reliability evaluation value is equal to or greater than a predetermined value, outputs at least one of the first rail position or the second rail position as a reliable rail position.

4. A distance measurement system as described in claim 3, characterized in that the reliability evaluation unit calculates the reliability evaluation value based on at least one of the following: straightness or flatness of the detected rail, or the position of a vehicle traveling on the rail, weather, time, inclination of the sun, detected distance of the rail, disparity value or effective disparity rate of the rail and its surroundings.

5. A distance measurement system as described in claim 3, characterized in that the first rail position detection unit detects the inner edge position and the outer edge position of the rail and outputs the center position between them as the rail position, and the rail width calculation unit outputs the rail width information based on the rail position and information on the standardized rail width.

6. A distance measuring system according to claim 2, characterized in that the rail unevenness information generating unit outputs a road surface parallax from the parallax image, and normalizes the parallax value of each pixel of the parallax image by the road surface parallax of each row to perform a linear approximation.

7. A distance measurement system as described in claim 3, characterized in that the rail width calculation unit calculates the rail width based on the rail position with the highest reliability by setting a weight on at least one of the rail position determined by the first rail position detection unit or the rail position determined by the second rail position detection unit output by the reliability evaluation unit.

8. A distance measurement system according to claim 3, comprising the rail information acquisition unit and a second rail distance calculation unit that calculates the distance to the rail using the rail information, wherein the second rail distance calculation unit corrects the distance Zb acquired from the distance calculation unit to Zb' calculated based on the following formula (1): Zb' = Zb / H (H-h) (1) where H is the camera height relative to the road surface, h is the rail height, and Zb' is the rail distance corrected from Zb.

9. A distance measurement system as described in claim 3, further comprising a memory unit that stores data contained in the rail information acquisition unit, the distance calculation unit, the rail unevenness information generation unit, the first rail position detection unit, the reliability evaluation unit, the rail width calculation unit, the first rail distance calculation unit, the second rail distance calculation unit, the distance error calculation unit and the correction amount calculation unit in association with at least one other data, and the data in the memory unit is capable of being transmitted to an external device.

10. A distance measurement system as described in claim 3, further comprising a communication unit capable of transmitting and receiving data contained in the distance calculation unit, the rail unevenness information generation unit, the first rail position detection unit, the reliability evaluation unit, the rail width calculation unit, the first rail distance calculation unit, the second rail distance calculation unit, the distance error calculation unit and the correction amount calculation unit to and from an external device, and characterized in that the distance measurement system is capable of performing distance measurements based on the results of information processing performed externally.

11. A distance measurement system as described in claim 1, wherein the first rail distance calculation unit is a first rail parallax calculation unit that calculates a parallax corresponding to the distance to the rail based on the rail width information, the distance error calculation unit is a parallax error calculation unit that compares the parallax output of the first rail parallax calculation unit with the parallax output obtained from a stereo matching unit to output a parallax error, and the correction amount calculation unit calculates a correction amount for correcting the parallax information of the stereo matching unit from the parallax error.

12. A vehicle equipped with a distance measuring system according to any one of claims 1 to 11.

13. A distance measurement system equipped with a rail information acquisition unit that acquires rail information from multiple directions and a distance calculation unit that performs distance calculation by stereo matching, comprising: acquiring rail unevenness information that detects unevenness of the rail and its surroundings by the rail information acquisition unit; detecting a first rail position based on the rail unevenness information; calculating a rail width based on the first rail position; outputting a first rail distance based on the rail width; calculating a second rail distance based on the output of the distance calculation unit; comparing the first rail distance and the second rail distance to calculate an error; and correcting the output of the distance calculation unit based on the error.

14. A distance measuring method according to claim 13, wherein the rail information acquisition unit is a stereo camera, and unevenness information is generated based on a parallax image generated from an image of the stereo camera.

15. A distance measuring method as described in claim 13, wherein the distance measuring system comprises a reliability evaluation unit which calculates a reliability evaluation value for judging the reliability of the rail position, and a second rail position detection unit which outputs the rail position from an image output by any of the rail information acquisition units, wherein the reliability evaluation unit compares the first rail position with the second rail position to calculate a reliability evaluation value for the rail position, and when the reliability evaluation value is equal to or greater than a predetermined value, the reliability evaluation unit outputs at least one of the first rail position or the second rail position as a reliable rail position.

Images