JP2006349607A - Distance measuring device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a distance measuring device which does not need other moving distance measuring means or the like as an essential requirement of a device structure by using a monocular image pick-up means as an external information acquisition means. <P>SOLUTION: This distance measuring device 100 essentially comprises: the monocular image pick-up means taking an external image; a space coordinate temporarily determination means 2 temporarily determining, based on a plurality of images with different pick-up times, a space coordinate Q on a three-dimensional space of an external fixed point taken on the images in an optional scale; a road surface equation temporary determination means 3 temporarily determining, based on the temporarily determined space coordinate Q of the fixed point, a plane equation of an approximate plane approximately showing a traveling road surface; a scale determination means 4 determining the scale, based on an attachment height h from the road surface of the image pick-up means 1 in a mobile body and the plane equation; and a space coordinate correction means 6 correcting the space coordinate Q based on this scale. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

The present invention relates to a distance measuring device that is mounted on a moving body that travels on a road surface and measures a distance to an object that is located outside the moving body.
INDUSTRIAL APPLICABILITY The present invention is very useful for robustly measuring the distance from a moving body equipped with a monocular imaging device to an object located in the outside world against disturbances during traveling.

  As a conventional technique that makes it possible to measure the amount of movement of a moving body using a monocular imaging device mounted on the moving body, for example, there is one disclosed in Patent Document 1 below. In the technology, no consideration is given to the dynamic fluctuation of the pitch angle of the moving body while traveling, and the case where the pitch angle of the moving body dynamically varies during traveling, or the gradient or curvature of the road surface. When is large, the amount of movement and the distance to the object in the outside world cannot be obtained accurately.

  As another conventional apparatus proposed for solving such a problem, for example, there is one disclosed in Patent Document 2 below. This conventional apparatus includes a monocular imaging unit mounted on a vehicle, a vehicle speed sensor that detects the vehicle speed of the host vehicle, and a structure that detects a structure fixed in the vicinity of the road in an image obtained by the imaging unit. Detection means, structure distance estimation means for estimating the distance from the imaging means to the structure based on the state of image change of the structure and the vehicle speed of the host vehicle, and measurement existing on the road on the image It is characterized by comprising object distance estimating means for estimating the distance from the imaging means to the object to be measured based on the positional relationship on the image between the target object and the structure.

In this conventional apparatus, the relative position and orientation of the moving body between these two times are obtained using two images captured at different positions and orientations while the moving body is moving. Since it is difficult to obtain the absolute value of the translational component in the relationship, that is, the distance between the two points from these two images, the distance between the two points is calculated by the distance calculation process using the vehicle speed sensor and the timer. It supplements.
JP2003-178309 JP2003-247824

However, since the vehicle speed sensor is an indispensable constituent element in the conventional device described in Patent Document 2 described above, such a conventional device cannot be effectively operated unless an interface with the vehicle speed sensor is provided. Can not.
For this reason, the application of this conventional apparatus to various moving bodies such as robots that do not include a vehicle speed sensor cannot be expected.

  The present invention has been made to solve the above-mentioned problems, and its purpose is to use a monocular image capturing means as external information acquisition means, and use an interface with the vehicle speed sensor, the vehicle speed sensor itself, or GPS. It is to realize a distance measuring device that does not make the positioning device etc. an essential requirement of the device configuration.

In order to solve the above problems, the following means are effective.
That is, the first means of the present invention is a monocular that captures an image of the outside world in a distance measurement device that is mounted on a mobile body that travels on a road surface and measures the distance to an object located in the outside world of the mobile body. An image imaging means and a spatial coordinate provisional means for temporarily obtaining a spatial coordinate Q on a three-dimensional space of a fixed point of the outside world imaged on the images based on a plurality of images having different imaging times on an arbitrary scale A road surface equation provisional means for provisionally obtaining a plane equation of an approximate plane that approximately represents the road surface being traveled based on the provisional fixed point spatial coordinates Q, and a road surface of the image pickup means in the moving body A scale determining unit for obtaining the scale based on the mounting height h from the plane and the plane equation, and a spatial coordinate correcting unit for correcting the spatial coordinate Q based on the scale. That.

  However, although the spatial coordinate provisional means obtains the spatial coordinates Q in the three-dimensional space of the fixed points, various kinds of handling for the fixed points can be performed in substantially the same manner simultaneously or in parallel. It does not interfere with the handling of other moving points. Embodiments involving such moving points may be additional or incidental and may be arbitrary.

Further, the second means of the present invention is highly likely to be located on the road surface traveling from the set of spatial coordinates Q of the fixed point in the road surface equation provisional means in the first means. A road surface candidate point selection means for selecting a set of spatial coordinates Q _G of the road surface candidate points is provided, and the road surface candidate point selection means selects a fixed point imaged on a predetermined road surface candidate area defined on the image as a road surface candidate. It is to select as a point.

  Further, a third means of the present invention is the above-mentioned road surface in such a manner that in the second means, the road surface candidate points selected by the road surface candidate point selection means are concentrated on the approximate plane or in the vicinity thereof. It is to provide candidate area optimization means for dynamically redefining the range of the candidate area as needed.

  However, as such an optimization method, for example, a method of redefining (optimizing) the road surface candidate area based on the density of each part of the captured image, a white line drawn on the road surface, or the like is detected. A method of redefining the road surface candidate region by using them, a method of redefining the road surface candidate region while excluding a partial region on the image that continuously gives an exceptional value, and the like can be considered. Of course, these methods may be arbitrarily combined appropriately.

  According to a fourth means of the present invention, in any one of the first to third means described above, at least eight feature point sets consisting of a set of two points whose features match between two different images are provided. Relative position of the above moving body between two different imaging times based on the corresponding point detection means to be obtained in pairs or more and the plane coordinates (x1, x2) of each feature point of each set on those images The position and orientation detection means for obtaining the orientation is provided in the spatial coordinate provisional means.

  The fifth means of the present invention is a road surface plane estimating means for obtaining the above plane equation using the well-known LedS method (The Least Median Squares method) in any one of the first to fourth means. Is provided in the road surface equation provisional means.

However, the robust estimation method applicable to the present invention is not limited to the LMedS method, and a known or arbitrary robust estimation method can be applied. As such a robust estimation method, for example, there are various estimation methods such as an M-estimator that appropriately weights errors.
By the above means of the present invention, the above-mentioned problem can be effectively or rationally solved.

The effects obtained by the above-described means of the present invention are as follows.
That is, according to the first means of the present invention, the inclination of the approximate plane with respect to the road surface on which the moving body is traveling can be obtained dynamically and accurately at any time. Is a known constant, and the scale of the spatial coordinate Q can be calculated correctly. Therefore, according to the first means of the present invention, even when the pitch angle of the moving body fluctuates dynamically during traveling or when the road surface has a large gradient or curvature, It is possible to accurately measure the distance to an object located outside the moving body without using a vehicle speed sensor.

  Further, according to the second means of the present invention, the coordinates of the fixed point used for estimating the inclination of the road surface are the coordinates of the fixed point (road surface candidate point) that is highly likely to be located on the road surface during traveling. You can easily narrow down to data only. For this reason, according to the 2nd means of this invention, the overhead of the process which estimates the inclination of a road surface can be reduced effectively, At the same time, the estimation precision can be raised effectively.

  According to the third means of the present invention, since the second means is dynamically optimized as needed, the coordinate data of the fixed point used for estimating the slope of the road surface is used as the road surface during traveling. It is possible to more effectively narrow down only to the coordinate data of a fixed point (road surface candidate point) that is more likely to be located above. For this reason, according to the third means of the present invention, it is possible to further effectively reduce the overhead of the process of estimating the slope of the road surface, and at the same time, it is possible to further increase the estimation accuracy.

  Further, according to the fourth means of the present invention, the above spatial coordinate provisional means can be configured most simply and rationally in accordance with a well-known mathematical method using a calibration matrix or a basic matrix. it can. Therefore, according to the fourth means of the present invention, it is possible to simply configure the spatial coordinate provisional means with less overhead.

  Further, according to the fifth means of the present invention, it is easy to robustly estimate the slope of the road surface. That is, for example, when the least square method is used, when there is a lot of data that gives an exceptional value, or when the deviation of the exceptional value is large, those exceptional values with respect to the desired estimated value (road slope) However, since the fifth means of the present invention complies with the robust estimation method, according to the fifth means of the present invention, the desired estimated value is obtained from the exceptional value. There is no significant influence.

Hereinafter, the present invention will be described based on specific examples.
However, the embodiments of the present invention are not limited to the following examples.

  FIG. 1 shows a logical basic structure of the distance measuring apparatus 100 according to the first embodiment. The distance measuring device 100 is a device that is mounted on a moving body that travels on a road surface and measures the distance to an object that is located outside the moving body. The main configuration of the distance measuring apparatus 100 includes a monocular image capturing unit 1 that captures an image of the outside world, and a three-dimensional fixed point of the outside world that is captured on the images based on a plurality of images having different imaging times. A spatial coordinate provisional means 2 that provisionally obtains a spatial coordinate Q on an arbitrary scale, and a plane equation of an approximate plane that approximately represents the road surface being traveled based on the provisional spatial coordinate Q of the fixed point. Tentatively obtaining road surface equation 3, scale determining means 4 for obtaining the scale based on the mounting height h of the image pickup means 1 from the road surface on the moving body and the plane equation, and the scale The spatial coordinate correcting means 5 corrects the spatial coordinate Q based on the above.

  In addition, the spatial coordinate provisional means 2 described above includes corresponding point detection means 21 that obtains at least eight sets of feature points each consisting of a set of two points whose features match between two different images, and on the images. Position / orientation detection means 22 for obtaining the relative position / orientation of the moving body between two different imaging times based on the plane coordinates (x1, x2) of each feature point of each feature point set is provided. . Hereinafter, the two-point set of plane coordinates (x1, x2) is referred to as a corresponding point. Furthermore, the spatial coordinate provisional means 2 calculates the spatial coordinates Q in the three-dimensional space of these corresponding points based on the output results of these respective means (corresponding point detecting means 21 and position / orientation detecting means 22). Spatial coordinate calculation means 23 is provided.

Also, the road surface equation provisional means 3 selects a set of spatial coordinates Q _G of road surface candidate points that are likely to be located on the traveling road surface from a set of detected spatial coordinates Q of fixed points. Candidate point selection means 31 is provided. Further, the road surface equation provisional means 3 is provided with road surface plane estimation means 32 for obtaining the above plane equation using the well-known LMedS method (The Least Median Squares method).

The hardware configuration diagram of FIG. 2 is a diagram illustrating the system configuration of the distance measuring apparatus 100 described above corresponding to a specific physical configuration. The distance measuring apparatus 100 is physically composed of an electronic control unit 100A in the figure and a monocular camera 100B corresponding to the image capturing means 1. On the other hand, each arithmetic processing means (each means such as 2, 3, 4, 5) is realized by an electronic control unit 100A in the figure.
For example, the ROM 102 stores a control program for realizing these means (2, 3, 4, 5). The CPU 101 uses the storage area provided by the RAM 103 as a load area or work area for the control program. The interface unit 104 includes input / output channels and the like, and is interposed in input / output processing with the camera 100B, for example.

Hereinafter, the control procedure of the distance measuring apparatus 100 will be described with reference to the general flowchart of FIG. The distance measuring apparatus 100 is an apparatus that repeatedly executes steps 310 to 350 described below periodically in accordance with a predetermined imaging cycle.
First, in step 310, a plurality of images captured at different positions by the image capturing unit 1 of FIG. 1, the corresponding point detection unit 21 constituting the first stage of the spatial coordinate provisional unit 2, and the road surface equation provisional It outputs to the road surface candidate point selection means 31 which comprises the first stage of the means 3.

  Next, at step 320, the spatial coordinate provisional means 2 in FIG. 1 executes a feature point extraction process such as a differentiation process or a tracking process for the feature point, and responds between two times when each image is captured. A pair of feature points (that is, the corresponding points (x1, x2) described above) is obtained, and spatial coordinates Q in the three-dimensional space of each corresponding point are provisionally calculated with an arbitrary scale S.

  At this time, first, the corresponding point detection means 21 extracts feature points from a plurality of inputted images. However, the feature point here refers to an easily traceable point that can be distinguished from other surrounding points. For these feature points, the above-described extraction process is executed using a method (for example, the Tomasi-Kanade method or the Harris operator) that detects pixels in which the gradient value of the gradation change is two-dimensionally large. At this time, for each feature point, using the assumption that the brightness of a point corresponding to a plurality of images having different imaging times and the brightness of surrounding points does not change, the feature point to be identified between images, that is, the same It is possible to easily detect the correspondence between images that are expected to be fixed points. When the corresponding points between the images (that is, the corresponding points (x1, x2)) are obtained, the respective plane coordinates (x1, x2) are output to the position / orientation detection means 22.

Hereinafter, the relationship between the extracted feature points (fixed points) and the motion of the moving body will be described with reference to FIG. The position / orientation detection unit 22 calculates the relative position and orientation of the imaging apparatus when an image is captured from the corresponding points detected by the corresponding point detection unit 21. As shown in FIG. 4, the position and orientation between the images are composed of a rotation matrix R and a translation vector t. For example, the plane coordinate of the feature point on the second image corresponding to the plane coordinate x1 of the feature point on the first image is x2. These are points indicating the same fixed point. At this time, if a calibration matrix K for correcting image distortion due to the imaging characteristics of the camera is used, the relationship between them can be expressed as the following equation. Here, the matrix F represents a known basic matrix.

According to the above configuration, if eight or more pairs of feature points corresponding to each other (that is, the same fixed point) between the two times are obtained, the basic matrix F is calculated based on that. Furthermore, if the camera calibration matrix K is known at this time, as can be seen from the above formula, the movement of the moving body between the two times (= the rotation of the moving body between the two times) The matrix R and the translation vector t) can be determined.
In other words, if there are 8 or more corresponding points (fixed points) between two images, the position and orientation detection means 22 can calculate the rotation matrix R and the translation vector t.

However, in reality, there may be a case where the obtained corresponding point does not represent a fixed point. Therefore, it is desirable that the corresponding point detecting means 21 obtains about 20 to 200 corresponding points. It is more preferable to detect corresponding points of around 100 points. If this value is appropriately determined, it is possible to achieve both high-speed processing and high estimation accuracy.
Further, at the stage of executing the spatial coordinate provisional means 2, the actual size of the translation vector t (that is, the moving distance of the moving body between two times of execution of imaging) and the coordinate system in which the spatial coordinates Q are to be defined. The scale S may remain unknown.

  Thereafter, the spatial coordinate calculation means 23 calculates the spatial coordinates Q of the corresponding points based on the principle of triangulation from the corresponding points detected by the corresponding point detection means 21 and the position and orientation between the images detected by the position and orientation detection means 22. Calculate At this time, since the moving distance of the moving body between two times at the time of execution of imaging is unknown, the spatial coordinate Q of the corresponding point is calculated by provisionally setting the translation vector t to a certain appropriate constant.

Next, in step 330 of FIG. 3, the road surface equation provisional means 3 of FIG. 1 approximates the running road surface in a plane and tentatively obtains an equation of the approximate plane. In other words, parameters (a, b, c) representing the inclination in the three-dimensional space of the traveling road surface are obtained.
At this time, first, in the road surface candidate point selection means 31 described above, the spatial coordinate Q of the fixed point imaged on the predetermined road surface candidate area defined on the image is used as the spatial coordinate Q _G of the road surface candidate point. Select. FIG. 5 illustrates a definition form of the road surface candidate area defined on the image. For example, in this way, the road surface candidate area on the image is defined in advance with an appropriate size below the image plane area. However, this road surface candidate area may be dynamically redefined at any time (third means of the present invention). As a method for realizing such redefinition, for example, a method for redefining (optimizing) a road surface candidate region based on image density information, or a method for redefining a road surface candidate region by detecting a white line drawn on the road surface. Possible methods.

Next, the road surface plane estimation means 32 in the next stage uses the spatial coordinates Q _G of the road surface candidate points obtained from the road surface candidate point selection means 31 to estimate an approximate plane for the road surface on which the moving body is traveling. FIG. 6 is a flowchart showing a control procedure executed at this time. The equation of the approximate plane to be estimated here can be provisionally given by the following equation (1) in the orthogonal coordinate system (X, Y, Z).
(Equation of approximate plane)
aX + bY + cZ = 1 (1)

That is, the road surface plane estimating means 32 in FIG. 1 obtains parameters a, b, and c indicating the inclination of the approximate plane. The set of spatial coordinates Q _G of the road surface candidate points may include spatial coordinates that do not exist on the road surface. For example, by introducing LMedS estimation (robust estimation method) as shown in FIG. The parameters a, b and c can be determined accurately.

Hereinafter, the control procedure for realizing the road plane estimation means 32 will be described in detail by way of example with reference to FIG.
First, in step 505 and step 508, initialization of the road surface plane estimation process, that is, initial setting of variables n, e, and i is performed. For example, the real number E (> 0) is the maximum allowable error as a median value eM described later. The natural number I is the number of road surface candidate points obtained by the above processing.

Next, in step 510, three points are randomly selected by random sampling from the set of spatial coordinates Q _G of the road surface candidate points obtained by the road surface candidate point selection means 31. Then, by these three points, in the next step 515, a plane to be evaluated as to whether or not it is more suitable as a desired approximate plane is specified. That is, in the next step 515, the inclination of the plane passing through these three points, that is, the coefficients a, b, and c satisfying the equation (1) are obtained.
For example, in this way, particularly when using the LmedS method or the like, the estimation process for the parameter to be finally obtained results in a non-linear optimization problem. Therefore, in order to speed up the optimization operation, a sub-optimal solution is used. For example, the Random Sample Consensus method for searching by random sampling as in step 510 is effective.

Next, in steps 520 to 530, for each spatial coordinate Q _G = (x _i , y _i , z _i ) (1 ≦ i ≦ I) of all road surface candidate points, an error evaluation function that satisfies the following equation (2): Each e (i) is obtained. However, at this time, these calculation results are arranged and saved in the ascending order of the values.
(Error evaluation function e (i))
e (i) = (ax _i + by _i + cz _i −1) ² (2)

Next, in step 540, the median of the above calculation results (: e (i)) saved in the ascending order is stored in the variable eM. However, the median here refers to the I / 2th or (I + 1) / 2th error evaluation function e (i) arranged in ascending order.
Next, in step 545, the magnitude relationship between the variable e and the variable eM is determined. If eM <e, the process proceeds to step 550, and if not, the process proceeds to step 570.

In step 550, the contents of each variable are updated by the following substitution operation.
(Assignment operation)
e = eM, A = a, B = b, C = c (3)
However, each of the save areas A, B, and C represents an argument area for transferring each value of the aforementioned parameter (a, b, c) indicating the inclination of a desired approximate plane to the caller of the subroutine. ing.

  In step 560, the end condition of this subroutine is determined. That is, if the value of the variable e is smaller than the predetermined positive constant ε, the process returns to the calling source of this subroutine, and if not, the process proceeds to step 570. Further, the end condition is also confirmed in steps 570 and 575. That is, the constant N is a natural number that defines the maximum number of repetitions of the above process. If this end condition is not satisfied, the process returns to step 508 again, and the process below the next plane is specified again. repeat. Then, by repeating these steps, the optimum parameters (a, b, c) are finally selected. That is, the desired slope (a, b, c) of the approximate plane is written to the output area (A, B, C) in step 550 by the above road plane estimation process (road plane estimation means 32).

However, if any median eM does not fall below the maximum allowable error E due to various unforeseen circumstances such as the imaging situation or imaging status, the above output areas (A, B, C ) Is not updated. Therefore, in such a case, the values of the parameters (a, b, c) obtained in the previous imaging cycle are continuously adopted.
For example, a robust estimation operation can be executed as described above. In addition, by optimizing the parameters I, N, E, and ε and the execution period (that is, the imaging period) of the subroutine, the calculation processing overhead in the estimation process of the subroutine is reduced and the estimation accuracy is improved. Can be rationalized at the same time.

Next, in step 340 of FIG. 3, the scale S of the spatial coordinate Q calculated by the spatial coordinate calculation means 23 is obtained from the road plane estimated by the road plane estimation means 32 by the scale determination means 4 of FIG. At this time, for example, as illustrated in the graph of FIG. 7, the inclination of the approximate plane that approximates the running road surface is illustrated by the camera viewpoint of the image pickup means 1 as the origin, the optical axis direction as Z, and the upward direction of the image as Y. Is the absolute value of the Y-axis intercept of the approximate plane, that is, the height of the camera viewpoint from the approximate plane in the coordinate system of the provisional spatial coordinate Q.
At this time, if the actual installation height of the imaging device is h, the scale S of the spatial coordinates Q calculated by the spatial coordinate calculating means 23, that is, provisional by the spatial coordinate provisioning means 2, is S = H / h. Can be calculated as

Next, in step 350 of FIG. 3, using the scale S obtained as described above by the spatial coordinate correcting unit 5 of FIG. Correct each coordinate value. That is, when the provisional scale S of the spatial coordinates Q is used, the spatial coordinates Q: (X, Y, Z) = (x _i , y _i , z _i ) are corrected based on the scale S. As a result, the desired orthogonal coordinates (x, y, z) corresponding to the actual three-dimensional space in the external world of the moving object with the camera viewpoint of the image pickup means 1 as the origin are rewritten as It can be obtained as in 4).
(Spatial coordinate Q 'after correction)
Q ′: (x, y, z) = (x _i / S, y _i / S, z _i / S) (4)

By using the above-described means, if the distance measuring device 100 is mounted on a moving body, even when the slope of the road surface changes dynamically during running or the pitch angle of the moving body fluctuates, The distance r from the camera viewpoint of the image pickup means 1 to an object such as a utility pole, guardrail, or road sign in the outside of the moving body can be accurately obtained according to the following equation (5).
(Distance r to object)
r = (x ² + y ² + z ² ) ^1/2 = (x _i ² + y _i ² + z _i ² ) ^1/2 / S (5)

[Other variations]
The embodiment of the present invention is not limited to the above-described embodiment, and other modifications as exemplified below may be made. Even with such modifications and applications, the effects of the present invention can be obtained based on the functions of the present invention.

(Modification 1)
For example, in the first embodiment, the road surface candidate area is a fixed area on the image plane. However, it is more desirable to dynamically redefine the road surface candidate area as needed. According to the configuration (third means of the present invention) using such an optimization technique, it is possible to effectively reduce the calculation time and improve the estimation accuracy of the approximate plane at the same time. As such an optimization method, for example, a method for redefining (optimizing) the road surface candidate area based on the shade of each part of the captured image, a white line drawn on the road surface, and the like are detected. Can be used to redefine the road surface candidate area, or from among the road surface candidate areas already defined on the image, the road surface candidates while excluding the partial areas that continue to give exceptional values continuously A method of redefining the area can be considered. Of course, these methods may be arbitrarily combined appropriately.

  The application target of the present invention is not limited to a moving body that travels on a road, and can be applied to, for example, a robot that moves on a floor surface. Moreover, it can be applied to ships and the like by considering the water surface instead of the road surface.

Control block diagram showing logical basic structure of distance measuring apparatus 100 Hardware configuration diagram showing the physical basic structure of the distance measuring apparatus 100 General flowchart illustrating the control procedure of the distance measuring apparatus 100 Conceptual diagram explaining the relationship between extracted feature points (fixed points) and movement of moving objects Screen plan view illustrating the definition form of the road surface candidate area defined on the image Flowchart illustrating the control procedure for realizing the road surface plane estimation means 32 A graph illustrating the slope of an approximate plane that approximates a running road surface

Explanation of symbols

100: Distance measuring device 100A: Electronic control unit 100B: Camera R: Rotation matrix t: Translation vector

Claims (5)

In a distance measuring device that is mounted on a moving body that travels on a road surface and measures the distance to an object located in the outside of the moving body,
Monocular image capturing means for capturing an image of the outside world;
Spatial coordinate provisional means for tentatively obtaining a spatial coordinate Q on a three-dimensional space of a fixed point of the outside world imaged on the image based on the plurality of images having different imaging times;
Road surface equation provisional means for provisionally obtaining a plane equation of an approximate plane that approximately represents the road surface being traveled based on the spatial coordinates Q of the fixed point provisionally,
Scale determining means for obtaining the scale based on the mounting height h from the road surface of the image pickup means in the moving body and the plane equation;
A distance measuring device comprising: a spatial coordinate correcting unit that corrects the spatial coordinate Q based on the scale. The road surface equation provisional means is:
Road surface candidate point selection means for selecting a set of spatial coordinates Q _G of road surface candidate points that are likely to be located on the road surface being traveled from the set of spatial coordinates Q of the fixed points;
The road surface candidate point selection means is:
The distance measuring apparatus according to claim 1, wherein a fixed point imaged on a predetermined road surface candidate area defined on the image is selected as the road surface candidate point. Candidate area optimization that dynamically redefines the range of the road surface candidate area as needed so that the road surface candidate points selected by the road surface candidate point selection means are concentrated on the approximate plane or in the vicinity of the approximate plane. The distance measuring device according to claim 2, further comprising: a converting unit. The spatial coordinate provisional means is:
Corresponding point detection means for obtaining at least eight sets of feature points each consisting of a set of two points whose features match between two different images;
Position and orientation detection means for obtaining a relative position and orientation of the moving body between two different imaging times based on the plane coordinates (x1, x2) of each feature point of each set on the image; The distance measuring device according to any one of claims 1 to 3, further comprising: The road surface equation provisional means is:
5. The distance measuring device according to claim 1, further comprising a road surface plane estimation unit that obtains the plane equation using an LMedS method. 6.