JPH0875454A - Ranging device - Google Patents

Abstract

(57) [Summary] [Purpose] By not performing corresponding point search, the speed and accuracy of the measurement of the distance from the distance measuring device to the target object can be increased. An image pickup means 10 which uses a plurality of cameras (cameras 1 to N) and picks up an image of an object such that the parallax of the adjacent cameras in the horizontal direction or the vertical direction is several pixels.
1 and a plurality of video signals picked up by a plurality of cameras of the image pickup means 101 at the same time, and the video signals of the same coordinate line in the horizontal direction or the vertical direction of the plurality of video signals are combined to create a parallax detection image. Using the parallax detection image creating unit 102 and the parallax detection image created by the parallax detection image creating unit 102, the parallax is obtained from the inclination of the straight line formed at the point corresponding to the edge portion of the object. The parallax calculating means 103 and the distance calculating means 104 for calculating the distance to the object based on the parallax calculated by the parallax calculating means 103 are provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distance measuring device for imaging a target object with a plurality of cameras and calculating the distance to the target object, and more particularly to a distance measuring device for speeding up the distance calculation. .

[0002]

2. Description of the Related Art As a conventional distance measuring device, for example, FIG.
There is a configuration shown in. This distance measuring device is composed of a camera 1001 that captures an image of an object and a camera 1002 that has the same focal length as the camera 1001 and that captures an object. Here, the optical axes of the cameras 1001 and 1002 are parallel, and the cameras 1001 and 1
The image pickup surface of 002 is arranged in such a position and orientation that the y-axis lies on the same line. In general, the condition of arranging in such a position and posture is called an epipolar condition, and hereinafter, this condition is referred to as an epipolar condition.

FIG. 11 shows a camera 1001 and a camera 100.
2 shows an image 1101 and an image 1102 captured, and in the figure, 1103 indicates two captured images 1101,1.
In FIG. 102, a certain point of the object projected on the same y coordinate is shown as a corresponding point. A conventional distance measuring device normally performs image processing such as edge detection processing and pattern matching processing to detect similar patterns on two images, and the corresponding point 1103 of each y coordinate is detected depending on the position of the pattern. Is determined, and the calculated coordinate value of the corresponding point 1103 is geometrically calculated to obtain the distance to the object.

Here, an example of calculating the distance to the object by the conventional distance measuring device will be specifically described with reference to FIG. FIG. 12 shows the relationship between two corresponding points in the cameras 1001 and 1002 and points on the object corresponding to the corresponding points. In the figure, 1204 is an image pickup surface 1201 of the camera 1001.
The above-mentioned corresponding points are shown, 1205 shows the corresponding points on the imaging surface 1202 of the camera 1002, and 1203 shows the points on the object.

Edge search is performed from two images (images of the image pickup surfaces 1201 and 1202) imaged under the epipolar condition.
By performing processing such as pattern matching, similar patterns are detected on the two images, and corresponding points 1204 and 1205 for each y coordinate are determined from the detected pattern positions. Based on the determined corresponding points 1204 and 1205, the camera 1
The distance l from the center of each camera lens of 001 and 1002 to the object is obtained. The coordinate value of the corresponding point 1204 captured by the camera 1001 is (x1, ye), and the coordinate value of the corresponding point 1205 captured by the camera 1002 is (x
2, ye), the distance l to the object is obtained by the equation 1. Further, here, (x2-x1) corresponds to the parallax.

L: (l + f) = d: (d + k (x2-x1)) l = df / (k (x2-x1)) (1) where f: focal length of camera lens d: center distance between lenses of the camera 1001 and the camera 1002 k: size of one element of the imaging surface of the camera (however, one element corresponds to one pixel of the image)

[0007]

However, according to the above-described conventional distance measuring device, the object is imaged by using two cameras, and the corresponding points are searched for on the same y coordinate on the two imaged images. Then, although the distance to the object is calculated,
Since the parallax between the two cameras is large, it takes a long time to find the corresponding points, and it is difficult to speed up the whole processing time. Also, the result of the corresponding point search is not always correct. There was a problem that the accuracy was poor because the distance to could not be obtained accurately.

The present invention has been made by paying attention to such a conventional problem, and by performing no corresponding point search,
An object of the present invention is to increase the speed and accuracy of the measurement of the distance from the distance measuring device to the object.

[0009]

In order to achieve the above object, the distance measuring device according to claim 1 uses a plurality of cameras (cameras 1 to N) as shown in the claim correspondence diagram of FIG. In addition, the image pickup means 101 for picking up an object so that the parallax in the horizontal direction or the vertical direction of the adjacent cameras becomes several pixels, and a plurality of video signals picked up by a plurality of cameras of the image pickup means 101 at the same time. Parallax detection image creating means 102 for creating a parallax detection image by capturing and synthesizing the video signals of the same coordinate line in the horizontal direction or the vertical direction of the plurality of video signals, and the parallax detection image creating means 10
Using the image for parallax detection created in 2, the parallax calculation means 103 for obtaining parallax from the inclination of the straight line formed at the point corresponding to the edge portion of the object, and the parallax calculation means 10
The distance calculation means 104 for calculating the distance to the object based on the parallax obtained in 3 above.

In the distance measuring apparatus according to the second aspect, the parallax calculating means 103 obtains the direction of the gradient vector on the parallax detecting image as the inclination of the straight line.

Further, in the distance measuring device according to the third aspect, the parallax calculating means 103 obtains the gradient vector by referring to a table of gradient vectors stored in advance.

Further, in the distance measuring apparatus according to a fourth aspect, the parallax calculating means 103 has a predetermined threshold with a horizontal differential component and a vertical differential component of a straight line formed at a point corresponding to an edge portion of the object. By comparing the values and removing a portion below a predetermined threshold as a noise portion from the parallax detection image, the slope of a straight line formed at a point corresponding to the edge portion of the object is obtained. .

[0013]

According to the distance measuring device of the present invention (claim 1), a plurality of cameras (cameras 1 to N) are used, and the parallax in the horizontal or vertical direction of the adjacent cameras is several pixels. An image pickup means 101 for picking up an object and a plurality of video signals picked up by a plurality of cameras of the image pickup means 101 are simultaneously taken in, and video signals of the same coordinate line in the horizontal direction or the vertical direction of the plurality of video signals are combined. By using the parallax detection image creating unit 102 that creates the parallax detection image by using the parallax detection image created by the parallax detection image creating unit 102, the image is formed at a point corresponding to the edge portion of the object. Parallax calculating means 10 for obtaining parallax from the inclination of a straight line
3 and the distance calculation means 104 for calculating the distance to the object based on the parallax obtained by the parallax calculation means 103, so that the corresponding point is not searched by the distance measuring device. Fast and accurate measurement of the distance to is required.

A distance measuring device according to the present invention (claim 2)
The parallax calculating means 103 can convert the obtained inclination of the straight line into a distance by obtaining the direction of the gradient vector on the parallax detection image as the inclination of the straight line.

Further, the distance measuring device of the present invention (claim 3)
The parallax calculation means 103 can obtain the gradient vector by referring to a gradient vector table stored in advance, and the gradient tan θ of a straight line formed at the corresponding points can be obtained quickly and accurately.

Further, the distance measuring device of the present invention (claim 4)
Means that the parallax calculation means 103 compares a horizontal differential component and a vertical differential component of a straight line formed at a point corresponding to the edge portion of the object with a predetermined threshold value, After removing the portion as a noise portion from the parallax detection image, the inclination of the straight line formed at the point corresponding to the edge portion of the object is obtained to obtain the camera N.
Even if some of the units have noise, the slope of the straight line can be accurately determined.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The distance measuring device of the present invention will be described in detail below in the order of [Embodiment 1], [Embodiment 2], [Embodiment 3] with reference to the drawings.

[Embodiment 1] FIG. 2A shows a schematic configuration of a distance measuring device according to Embodiment 1, in which cameras 202 to 204 for picking up an image of an object 201 and a half mirror for changing an optical axis of the camera 202 are shown. 205, a half mirror 206 for changing the optical axis of the camera 203, and images (video signals) from the cameras 202 to 204 at the same time, and video signals of a plurality of video signals on the same coordinate line in the horizontal direction (or the vertical direction). Is generated to generate a parallax detection image, the parallax detection image is used to obtain the parallax from the inclination of a straight line formed at a point corresponding to the edge portion of the object 201, and based on the obtained parallax , An image processing device 207 that calculates the distance to the object 201.

The cameras 202 to 204 are of the same model and have the same focal length, and the object 2
01, the parallax corresponds to several pixels of the image captured under the epipolar condition, and the optical axes of the cameras 202 to 204 are set such that the intervals between adjacent optical axes are equidistant. .

Further, in the above structure, the camera 202
To 204 and half mirrors 205 and 206 correspond to the image pickup means of the present invention, and the image processing device 207 corresponds to the parallax detection image creating means, the parallax calculation means, and the distance calculation means of the present invention.

Further, the cameras 202 to 204 are the same as those shown in FIG.
As shown in (b), it is necessary to make the center distance between the lenses extremely close in order to make the parallax several pixels, but in reality,
It is expected that it is impossible to bring the cameras 202 to 204 into close proximity to each other because of the physical size of the cameras. Therefore, in the first embodiment, as shown in FIG. 2A, the half mirrors 205 and 206 are arranged and the optical axes of the cameras 202 and 204 are bent so that the respective cameras 202 to 2 can be bent.
No. 04 is installed without being very close to each other so that an image of the object can be obtained within a parallax range of several pixels.
In addition, at this time, the cameras 202 to 204 use the target object 20.
The position and orientation are adjusted so that the images in the same direction 1 can be obtained.

The operation of the above configuration will be described with reference to the flowchart of FIG. First, the object 201 is imaged using the three cameras 202 to 204 (S301). As described above, in the first embodiment, since the cameras 202 to 204 are arranged so that the parallax in the horizontal direction is several pixels, the images captured by the cameras 202 to 204 are horizontal in the x-axis and vertical in the vertical direction. In the coordinate system whose direction is the y-axis, the image is shifted in the x-axis direction by the parallax (several pixels).

4A to 4C show cameras 202 to 2
3 shows three images captured in 04. Here, (a) of the figure is an image captured by the camera 202, and the coordinates of one point 401 corresponding to the edge portion of the object are represented by (x
1, ye). Further, FIG.
The coordinates of one point 402 corresponding to the edge portion of the object are (x2, ye). Further, FIG. 6C is an image captured by the camera 204, and the coordinates of one point 403 corresponding to the edge portion of the target object are (x3, ye).

Next, the image processing device 207 causes the camera 20 to
Simultaneously capture the 3 images captured in 2 to 204,
The parallax detection image is created by synthesizing the video signals of the same coordinate line in the vertical direction (y-axis direction) of the images.
2). Specifically, as shown in FIGS. 4 (a) to 4 (c),
The three video signals on the coordinate line whose y coordinate is ye are shown in FIG.
As shown in (d), three lines are arranged side by side to create a parallax detection image. Since the parallaxes of the images captured by the adjacent cameras are set to several pixels, the edges (straight lines) that can be connected to three pixels in the y-axis direction when the image of FIG. It is composed of corresponding points of the image.

Subsequently, using the image for parallax detection shown in FIG. 4D, the parallax is obtained from the inclination of the straight line 404 formed at the points corresponding to the edge portion of the object (S303). Specifically, if the slope of the straight line 404 is A, then A = 2 /
(X3-x1), where camera 202 and camera 204
Since the parallax of the pixel is (x3-x1), the parallax (x3-x1) can be obtained from 2 / A.

Finally, the distance to the object is calculated based on the obtained parallax (S304). Specifically, the distance l to the object is calculated by the following equation (2). l = 2 · d · f / (k (x3−x1)) (2) Where, f: focal length of camera lens 2 · d: center distance between lens of camera 202 and camera 204 k: camera The size of one element of the image pickup surface (However, one element corresponds to one pixel of the image)

Since the parallax (x3−x1) = 2 / A, the equation (2) can be rewritten as the equation (3). l = A · f · d / k (3)

As described above, the parallax can be obtained if the inclination of the straight line formed by the corresponding points of the image captured under the epipolar condition is known, and therefore the distance 1 to the object can be obtained from the equation (3).

As described above, in the first embodiment, an example in which an image of an object is picked up by using three cameras has been shown, but the same applies when the number of cameras is different. For example, when the calculation method when N cameras of camera 1 to camera N are used is described by a mathematical expression, the coordinates of the corresponding point of the camera 1 are (x1 ′, y
e) and the coordinates of the corresponding points of the camera N are (xn, ye), the distance 1 to the object is as shown in equation (4). l = (N−1) · d · f / (k (xn−x1 ′)) (4)

On the other hand, an image for parallax detection created from images taken by N cameras can be made up of N lines. The inclination A of the straight line on the parallax detection image is A = (N-1)
/ (Xn-x1 '), and the parallax is (xn-x1') =
Since (N-1) / A, the equation (4) can be rewritten as the equation (5). l = A / f / d / k (5)

In other words, even when the number of cameras is different, the distance 1 can be obtained as in the first embodiment.

Next, referring to FIG. 5, step S in FIG.
A method in which the image processing apparatus 207 obtains the parallax from the inclination of the straight line 404 in 304 will be specifically described. The inclination of the straight line 404 representing the parallax can be considered as the direction of the density gradient vector. Here, as shown in FIG. 5, FIG. 5A is an original image showing the density, and a straight line 40 is drawn at the boundary where the density changes.
There are four. The slope tan θ of the straight line 404 is calculated by the following equation. Assuming that the luminance I is the variation amount dI of the luminance I, tan θ = (dI / dx) / (dI / dy) = horizontal differential component / vertical differential component (6) where dx is the horizontal change Amount, dy is the amount of change in the vertical direction

In equation (6), tan θ is the slope of the straight line A = y
Is the amount of increase / the amount of increase in x. The horizontal derivative of equation (6) is
It is obtained by applying the horizontal differential Sobel operator shown in FIG. In addition, the vertical differential of equation (6) is shown in FIG.
It is obtained by applying the Sobel operator for vertical differentiation shown in (c).

In this way, the distance to the object can be obtained at high speed by obtaining the parallax from the gradient vector (tilt) of the parallax detection image and converting the parallax into the distance.

As described above, according to the first embodiment, three cameras with respect to the object 201 under epipolar conditions have a parallax of several pixels, and each camera has an interval between adjacent optical axes. Set so that they are equidistant, and one line image of the same y coordinate of the images captured by each camera is sequentially copied to create a composite image that forms three continuous y coordinate images, and synthesizes them. 3 in the y-axis direction when the image is differentiated
If the inclination of an edge (straight line) formed by corresponding points of images captured by each camera so as to be continuous with each other is obtained, the distance l to the object can be obtained, so that the conventional corresponding point search becomes unnecessary and parallax is reduced. The speed of the whole process can be increased by obtaining the speed with a simple algorithm. Also, hardware can be easily implemented. Furthermore, since the parallax is obtained from the tilt without being affected by the result of the corresponding point search, the distance to the object can be accurately obtained.

[Second Embodiment] In the second embodiment, a gradient vector table (hereinafter referred to as a gradient table) stored in advance is referred to in place of the method of calculating the gradient vector to obtain the slope of the straight line. , It is designed to obtain the slope of a straight line.

FIG. 6 shows a schematic structure of the distance measuring apparatus of the second embodiment, using a plurality of cameras 1 to N so that the parallax of adjacent cameras in the horizontal or vertical direction is several pixels. An image pickup unit 600 for picking up an object (not shown) and a plurality of video signals picked up by a plurality of cameras 1 to N of the image pickup unit 600 at the same time, and the same coordinate line in the horizontal direction or the vertical direction of the plurality of video signals. A parallax detection image creation unit 601 that creates a parallax detection image by synthesizing the video signals of
Using the parallax detection image created by the parallax detection image creation unit 601, a vertical differential component dI / created by the inclination of the straight line formed at the point corresponding to the edge portion of the object.
Using the vertical differential unit 602 for obtaining dy and the parallax detection image created by the parallax detection image creation unit 601, a horizontal differential created by the inclination of the straight line formed at the point corresponding to the edge part of the object. A horizontal differentiating unit 603 that obtains the component dI / dx, a gradient table 604 that stores the slope (direction of the gradient vector) tan θ of the straight line in advance in association with the vertical differential component dI / dy and the horizontal differential component dI / dx, and the straight line And a distance calculation unit 605 that calculates the distance to the object based on the calculated parallax.

In the above structure, the vertical differentiator 6
02, the horizontal differentiator 603, the gradient table 604, and part of the distance calculator 605 form the parallax calculator of the present invention. Further, the distance calculation unit 605 corresponds to the distance calculation means.

FIG. 7 is an explanatory diagram showing a method of creating the gradient table 604. The value of the inclination (direction of the gradient vector) tan θ of the straight line is set in advance on the coordinates of the gradient table 604 so that the value of tan θ can be obtained by looking up the coordinates. Specifically, the horizontal differential component dI /
tan θ when dx = a and vertical differential component dI / dy = b
The value of is stored in the position of the coordinates (a, b) in the gradient table 604. As a result, the horizontal differential component dI / dx =
a, by referring to the contents of the coordinates (a, b) of the vertical differential component dI / dy = b, tan θ = (dI / dx) /
The value of (dI / dy) can be derived.

The operation of the above configuration will be described. First, the parallax detection image creation unit 601 inputs the video signals from the N cameras 1 to N while synthesizing them line by line, and creates a parallax detection image. At the same time, the vertical differentiator 602 and the horizontal differentiator 603 determine the horizontal differential component dI / dx = a and the vertical differential component dI / dy = b. Next, using the obtained values of a and b, the slope tan θ of the straight line is obtained from the gradient table 604, and the distance calculation unit 60
In 5, the parallax is calculated using the value of the slope tan θ of the straight line,
Further, the distance to the object is calculated based on the obtained parallax.

As described above, according to the second embodiment, by utilizing the gradient table, the horizontal differential component dI / d
The slope tan θ of the straight line can be obtained without calculation simply by obtaining x and the vertical differential component dI / dy.
It is possible to easily obtain the distance from the obtained tan θ to the object.

[Third Embodiment] A third embodiment shows another example of the method of calculating the distance when N cameras for picking up an object are used, and a straight line formed at a point corresponding to the edge portion of the object. The horizontal differential component and the vertical differential component of are compared with a predetermined threshold value, and after removing a portion below the predetermined threshold value as a noise portion from the parallax detection image, at a point corresponding to the edge portion of the object. The inclination of the formed straight line is obtained. The basic configuration and operation are common to those of the second embodiment, and only different parts will be described here.

FIG. 8 shows a gradient table 801 according to the third embodiment. Here, the size of the gradient table 801 is set to 256.
X256. In the second embodiment, the horizontal differential component dI / d
The value of tan θ when x = a and the vertical differential component dI / dy = b is the coordinate (a, b) of the gradient table 604 (see FIG. 7).
However, in the third embodiment, the gradient table 8 is stored.
Instead of storing the value of tan θ in 01, the coordinate value on the gradient table 801 is set to dI /
A flag indicating that an edge does not exist is stored in a portion of dx <threshold and dI / dy <threshold (hereinafter, referred to as a flag portion 802).

Here, FIG. 9 shows a gradient image 901 created using the table of the gradient table 801 and a camera N.
An image 902 for parallax detection is shown, in which images of one line each having the same y-coordinate of the image captured by the table are sequentially copied and configured by N lines.

First, video signals from N cameras are input while being synthesized line by line, and an N line parallax detection image 902 is created. At the same time, the horizontal differential component dI
/ Dx = a and the vertical differential component dI / dy = b are obtained. Next, using the values of a and b thus obtained, the gradient table 8
01 to dI / dx and dI / dy are referred to whether or not they correspond to the flag portion 802, and if they correspond to the flag portion 802, the pixel (video signal) of that portion is deleted and the flag portion 802 is applied. If not, a gradient image 901 in which the pixels of that portion are stored is created.

Next, for this gradient image 901, Ho
The straight line inclination is obtained by performing the Ugh conversion, the parallax is calculated using the value of the linear inclination, and the distance to the object is calculated based on the obtained parallax.

By obtaining the inclination of the straight line in this way, the inclination of the straight line can be accurately obtained even if, for example, some of the N cameras have noise. Further, by using the N cameras, the inclination of the straight line can be obtained with high accuracy, so that the measurement accuracy of the distance to the object can be improved. In addition, when N cameras are used, the larger the value of N, the better the inclination detection accuracy.

[0048]

As described above, the distance measuring device of the present invention (claim 1) uses a plurality of cameras (cameras 1 to N) and the horizontal or vertical direction of the adjacent cameras. An image pickup means for picking up an object so that the parallax becomes several pixels, and a plurality of video signals picked up by a plurality of cameras of the image pickup means at the same time, and the same coordinate line in the horizontal direction or the vertical direction of the plurality of video signals. Using the parallax detection image creating means for creating the parallax detection image by synthesizing the video signals of the above, and the parallax detection image created by the parallax detection image creating means, a point corresponding to the edge part of the object Since the disparity calculating means for calculating the disparity from the inclination of the straight line formed in 1) and the distance calculating means for calculating the distance to the object based on the disparity calculated by the disparity calculating means are provided, the conventional corresponding inspection There will unnecessary, parallax can be made to speed up the whole process by obtaining a fast and simple algorithm, also hardened also easily.
Further, since the parallax is obtained from the tilt without being affected by the result of the corresponding point search, the distance from the distance measuring device to the object can be measured quickly and accurately.

The distance measuring device according to the present invention (claim 2)
Since the parallax calculation means obtains the orientation of the gradient vector on the parallax detection image as the inclination of the straight line, the orientation of the obtained gradient vector is converted into a distance to quickly determine the distance to the object. You can ask.

The distance measuring device according to the present invention (claim 3)
In the above, since the parallax calculating means is configured to refer to a table of gradient vectors stored in advance, the gradient vector can be obtained, and the slope of the straight line can be obtained quickly and accurately without calculation. The distance from the tilt to the object can be easily obtained.

Further, the distance measuring device of the present invention (claim 4)
Means that the parallax calculation means compares a horizontal differential component and a vertical differential component of a straight line formed at a point corresponding to an edge portion of the target object with a predetermined threshold value, and a portion below the predetermined threshold value Is removed from the parallax detection image as a noise portion, and then the inclination of the straight line formed at the point corresponding to the edge portion of the object is obtained. The slope of a straight line can be accurately calculated even when the
When N cameras are used, the larger the value of N, the higher the accuracy of inclination detection.

[Brief description of drawings]

FIG. 1 is a diagram corresponding to the claims of the present invention.

FIG. 2 is an explanatory diagram showing a schematic configuration of a distance measuring device according to a first embodiment.

FIG. 3 is a schematic flowchart of Example 1.

FIG. 4 is an explanatory diagram showing the operation of the first embodiment.

FIG. 5 is an explanatory diagram illustrating a method for the image processing apparatus according to the first exemplary embodiment to obtain parallax from the inclination of a straight line.

FIG. 6 is an explanatory diagram showing a schematic configuration of a distance measuring device according to a second embodiment.

FIG. 7 is an explanatory diagram showing a gradient table according to the second embodiment.

FIG. 8 is an explanatory diagram showing a gradient table according to the third embodiment.

FIG. 9 is an explanatory diagram illustrating a gradient image and a parallax detection image according to the third embodiment.

FIG. 10 is a configuration diagram showing a conventional distance measuring device.

FIG. 11 is an explanatory diagram showing an image captured by a camera of a conventional distance measuring device.

FIG. 12 is an explanatory diagram showing a relationship between an image pickup surface picked up by a camera of a conventional distance measuring device and corresponding points.

[Explanation of symbols]

101 image pickup means 102 parallax detection image creation means 103 parallax calculation means 104 distance calculation means 201 objects 202 to 204
Cameras 205, 206 Half mirror 207 Image processing device 401-403 One point corresponding to the edge portion of the object 404 Straight line 600 Imaging unit 601 Parallax detection image creation unit 602 Vertical differentiation unit 603 Horizontal differentiation unit 604 Gradient table 605 Distance calculation Part 801 Gradient table 802 Flag part 901 Gradient image 902 Parallax detection image 1001, 10
02 camera 1101, 1102 image 1103, 1204, 1205 corresponding point 1201, 1202 imaging surface

Claims (4)

[Claims] 1. An imaging means for imaging an object using a plurality of cameras so that the parallax in the horizontal direction or the vertical direction of the adjacent cameras is several pixels, and a plurality of cameras of the imaging means. A parallax detection image creating means for simultaneously taking in a plurality of picked-up image signals and synthesizing the image signals of the same coordinate line in the horizontal direction or the vertical direction of the plurality of video signals to create a parallax detection image; Using the parallax detection image created by the parallax detection image creating means, the parallax calculating means for finding the parallax from the inclination of the straight line formed at the point corresponding to the edge portion of the object, and the parallax found by the parallax calculating means A distance measuring device comprising: a distance calculating means for calculating a distance to an object based on the distance measuring means. 2. The distance measuring apparatus according to claim 1, wherein the parallax calculating unit obtains a direction of a gradient vector on the parallax detection image as the inclination of the straight line. 3. The distance measuring apparatus according to claim 2, wherein the parallax calculating unit obtains the gradient vector by referring to a gradient vector table stored in advance. 4. The parallax calculation means compares a horizontal differential component and a vertical differential component of a straight line formed at a point corresponding to an edge portion of the target object with a predetermined threshold value, and determines a predetermined threshold value. 2. The distance measuring apparatus according to claim 1, wherein after the following portion is removed as a noise portion from the parallax detection image, the inclination of a straight line formed at a point corresponding to the edge portion of the object is obtained.