JPH085369A - Distance-information measuring apparatus - Google Patents

Abstract

PURPOSE:To obtain a distance-information measuring apparatus by which distance information is computed at high speed and with high accuracy by a method wherein the operation of a normalization correlation is performed only when respective pixel output values of a reference field of view and a standard field of view satisfy a preset arithmetic value. CONSTITUTION:Images are formed on CCDs 3a, 3b by means of lenses 2a, 2b, their outputs are passed through A/D converters 4a, 4b, and signal outputs dsi, dri, in one scanning line each, which correspond to each other are stored in memories 5a, 5b. A product-sum arithmetic device 6, adders 7a, 7b and product-sum arithmetic devices 8a, 8b find respective SIGMA(ds.dr), SIGMAds, SIGMAdr, SIGMAds<2> and SIGMAdr<2> on the basis of the outputs dsi, dri. On the basis of respective outputs in the above, multipliers 9, 10 find respective m.SIGMA(ds.dr) [where (m) is an integer] and B=SIGMAds.SIGMAdr. An ALU 11 finds DELTA=A.B. When DELTA<=0, a computation is stopped. When 4>0, a normalization correlation value is computed by a subcorrelation arithmetic device 13 and a correlation arithmetic device 14. In addition, when this value is smaller than a threshold value, it is judged that the value is generated due to an irregularity in pixel sensitivity, and a negative value is given to a correlation value.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distance information measuring device, and more particularly to environmental recognition by a camera, an industrial robot, a measuring instrument, etc. that optically and electronically measures distance information to an object existing in a three-dimensional space. -It is suitable for understanding.

[0002]

2. Description of the Related Art Conventionally, for example, in JP-A-2-46984 and JP-A-4-67607, distance information to an object is optically and electronically measured by a so-called stereo method, and this distance information is measured. Has been proposed for obtaining distance information to an object existing in a plurality of directions by obtaining a plurality of portions on an image sensor. FIG. 2 shows an optical system for obtaining distance information by a conventional so-called stereo method. FIG. A known stereo method will be described based on this. 2A, 2B, and 2C are a plan view, a side view, and a front view of an image sensor, respectively, of the measurement optical system of the stereo method. In the figure, 2a and 2b are lenses having the same focal length, and 2A and 2B are their optical axes. The lenses 2a and 2b are arranged so that the optical axes 2A and 2B are parallel to each other and the base line 2C is orthogonal to the optical axes 2A and 2B. The distance between the optical axes 2A and 2B is D, which is referred to as the baseline length. 3a and 3b are lenses 2a and 2b, respectively
2C, which is a pixel value distribution measuring means corresponding to, and has two-dimensional array of M pixels in the X direction and N pixels in the Y direction, and is formed by the lenses 2a and 2b, respectively. It is a photoelectric conversion element that outputs an electric signal for each pixel according to the illuminance of an image, a so-called CCD area sensor (hereinafter referred to as CCD). The CCDs 3a and 3b are separated from the lenses 2a and 2b by the focal length f of the lenses 2a and 2b, and the optical axes 2
It is arranged so as to be orthogonal to A and 2B, and the arrangement of the CCD 3a with respect to the optical axis 2A and the CCD 3b with respect to the optical axis 2B.
Is the same as the arrangement.

In FIGS. 2A and 2B, I is an object to be measured, which has a certain area as described later. As shown in the figure, the object I exists in the direction of the angle θ in the X direction and the angle φ in the Y direction with respect to the optical axis 2A when viewed from the center of the lens 2a. CCD 3a by lens 2a, 2b,
The images of the object I on 3b are Ia 'and Ib', respectively.

As shown in FIG. 2 (c), the image Ia 'has the CCD 3
It is assumed that the pixel of a is formed by m ₀ pixels from the i ₀ th in the X direction in the j th line in the Y direction. This m
Each pixel is referred to as a reference visual field La. The object I is the conjugate range of the reference visual field La on the object side. Similarly, CCD3
Y including the portion where the image Ib ′ is located among the pixels of b
The n pixels from the p ₀ th position in the X direction on the j th line in the direction are selected as the reference visual field Lb. Here, m <n, that is, the number of strokes belonging to the reference visual field Lb is set larger than the number of pixels belonging to the standard visual field La (image Ia ′).

The reference field of view La (image I
a ′) and the output signals ds _i , d from the pixels of the reference visual field Lb
R _i is used to correlate both fields of view to obtain the relative positional deviation amount δ of the images Ia ′ and Ib ′ on the CCDs 3a and 3b.

First, i in the reference visual field La (image Ia ')
₀ th through i ₀ + m-1 th obtaining a normalized correlation value COR ₀ by the output dr _i of the output ds _i and at p ₀ th ~p reference field Lb ₀ + m-1 th pixel pixel. next,
In the same reference field La (image Ia ′) as before, i _0-
i ₀ + m-1 th p ₀ +1 th and one shift on the output ds _i and the reference field Lb of the pixel _{~ (p 0 + m-1} ) +1
The normalized correlation value COR ₁ based on the output dr _i of the th pixel is obtained. Hereinafter, the pixels taken up in the reference visual field Lb are shifted one by one to obtain COR _k (where k represents the number of shifts), and up to COR _mn .

Among the (m-n + 1) normalized correlation value COR _k values thus obtained, the number k indicating the highest correlation value is selected. Then, the portion having the highest correlation in the reference visual field with respect to the standard visual field La (image Ia ′), that is, the image Ib ′.
Is shifted from the reference visual field La (image Ia ′) by q = p _o −i ₀ + k (1) in terms of the number of pixels. Therefore, assuming that the width of one pixel is τ, the shift amount (shift amount) δ in the X direction can be obtained by δ = τ · q (2).

Next, the angles θ and φ are obtained as follows.
The optical axis intersection Oa 'on the CCD 3a is in the X direction at the i _AX address Y.
It is assumed to be on the pixel at address j _{AXE in the} direction. Reference visual field La
Corresponds to the image Ia ′, the angle θ in the X direction is tan θ = −τ · (i ₀ −i _AXE ) / f assuming that the starting address i _{0 of} the reference visual field La represents the image Ia ′ in this case. (3) On the other hand, the angle φ with respect to the Y direction is calculated by the following formula.

Tan φ = −τ · (j−j _AXE ) / f (4) Then, a = f · D / δ (5) is obtained from the relationship of δ / f = D / a. Next, b = a · sec θ, c = a · tan
From the relation of φ, the distance R to the object I is obtained by R = a · (sec ² θ + tan ² φ) ^1/2 (8), and the direction θ and φ of the object and the distance to the object And R are obtained.

As described above, the distance to the object corresponding to one reference visual field La in the j-th line in the Y direction among the pixels of the CCD was obtained, but in the same j-th line, a position different from the above position was obtained. By repeatedly setting the reference visual field La (image Ia ') and finding the distance to the object corresponding thereto, all distance information to the object corresponding to the entire j-th line can be obtained.

Further, by setting the jth column for setting the reference visual field La (image Ia ') in order from the first column to the Nth column, and repeating the above correlation calculation, the object visual field formed on the CCD 3a. Information on the distance to the object is obtained over the entire area.

This is the principle of measuring the distance information to the object by the stereo method.

[0013]

In the above stereo method, C is used.
To obtain distance information over the entire visual field of the object on the CD 3a, a huge amount of calculation is performed, so that it is difficult to obtain the distance information over the entire visual field in real time. Further, in such a method of measuring distance information by image correlation, a high correlation value cannot be obtained when there is no illuminance change, that is, image contrast in the portion of the image to be correlated, and therefore the measurement of distance information Will be difficult. Further, the sensitivities of the individual pixels forming the pixel value distribution measuring means (CCD) are not exactly the same and have slight variations. Therefore, if the correlation is simply calculated, if the image has uniform illuminance. Although the correlation should be zero, a non-zero correlation coefficient is obtained due to the variation in sensitivity, which gives a large error to the measurement result of the distance information. In the case of detection of distance information in a camera, that is, in the case of so-called autofocus, an error in measurement of distance information is caused by the user consciously selecting a subject whose illuminance changes in a portion where distance information is measured. It is possible to avoid it. However, it is generally difficult to form an image with a change in illuminance in all measurements. Therefore, the above problem always occurs when using such an arrangement.

According to the present invention, when the distance information to the object is obtained by the so-called stereo method, the standard field of view and the reference field of view are normalized even if there is some variation in the sensitivity of each pixel of the CCD as the pixel value distribution measuring means. It is an object of the present invention to provide a distance measuring device that can efficiently obtain a correlation value and can obtain distance information with high accuracy and at high speed.

[0015]

The distance information measuring device of the present invention has the following configuration.

(1-1) A reference system having a lens system and a large number of pixels, and a pixel value distribution measuring means for obtaining image information of an object image formed by the lens system by two-dimensional scanning, and the reference system. When the reference systems of the same structure are arranged such that their optical axes are parallel to each other, and the direction connecting the two optical axes on the measurement surface is the X direction and the direction perpendicular to this is the Y direction, the pixels of the reference system sets the reference field on the pixel column in the Y-coordinate y _j on the value distribution measuring means, wider area than the reference field on the pixel columns of the same y _j Y-coordinate on the pixel value distribution measuring means of the reference system is a reference field Of the reference field of view and the normalized correlation value between the pixel output value from the selected portion of the reference field of view having the same number of pixels as the standard field of view and the pixel output value from the reference field of view are sequentially By calculating, the standard field of view and the reference field of view In a distance information measuring device that obtains a deviation amount in the X direction from a portion having a high correlation and calculates distance information regarding the object from the deviation amount, the normalized correlation calculating means determines that pixel output values of the reference visual field and the standard visual field are The feature is that the normalized correlation is calculated only when a preset calculated value is satisfied.

In particular, (1-1-1) the normalized correlation calculation means calculates the product of each pixel value of the arbitrary standard image area and each pixel value of the reference image area obtained by the pixel value distribution measuring means. The difference between the sum value multiplied by the number of pixels in each image area and the sum value of the pixel values of the standard image area and the sum of the pixel values of the reference image area is equal to 0. If it is smaller than 0, the subsequent calculation of normalized correlation is interrupted.

(1-1-2) When it is determined that the normalized correlation value calculated by the normalized correlation calculation means is smaller than a preset threshold value, it is considered that the detection of the distance information is indefinite. , Records or displays information to that effect. It is characterized by such things.

Here, the normalized correlation value (COR _k ) _j calculated by the distance information measuring device according to the present invention is expressed by the following equation.
(Throughout this specification, () _j is the “j-th column ...
・ ”Is meant to be expressed)

[0020]

[Equation 1] Is. In the above, i ₀ and p ₀ are the starting addresses in the X direction of the standard visual field La and the reference visual field Lb, respectively,
i is the X-direction address of the pixel on the CCD, j is the Y-direction address, k is the number of shifts from p ₀ when m pixels are selected from the reference visual field Lb when calculating the correlation value, and k is 0 to 0. nm
Takes the value of. Next, A = m · Σ (ds · dr) (15) B = Σds · Σdr (16) C = m · Σds ² − (Σds) ² (17) D = m · Σdr ² − (Σdr) ² ‥‥‥ (18) E = (CD) ^1/2 ‥‥‥‥ (19) △ = AB ‥‥‥‥ (20), the normalized correlation value is ( COR _k ) _j = Δ / E (21)

[0021]

In the present invention, when calculating the normalized correlation value COR _k of the equation (21) in the distance information measuring device, first, the equations (15) A, (16) B and (20) are given. The value is calculated by calculating Δ, and if Δ is 0 or a negative number, it is determined that the distance is indefinite, and the subsequent calculation related to COR _k is stopped, thereby reducing a huge amount of calculation. , Reduce the distance information measurement time. At the same time, information reflecting the indefinite distance is set in the normalized correlation value.

Further, when the Δ value is positive, the normalized correlation value CO
The calculation of R _k is executed to the end, but when the normalized correlation value is obtained, it is determined whether the value is larger or smaller than a preset threshold value. It is judged that there is a high possibility that a positive value is obtained due to the variation of, and the information reflecting the indefinite distance is set as the normalized correlation value.

[0023]

Embodiment 1 FIG. 1 is a block diagram of the essential parts of Embodiment 1 of the present invention, showing a case where the invention is applied to a robot capable of autonomous traveling. In FIG. 1, I is an object to be measured in distance, 2a,
Reference numeral 2b denotes two identical lenses (focal length f), 3a and 3b denote pixel value distribution measuring means, which are composed of MxN pixels, and photoelectrically convert the illuminance of the image formed by the lens for each pixel. It is composed of a CCD area sensor (hereinafter referred to as CCD). 4a, 4b are A / D converter - a motor, CCD 3a, digital electric signals photoelectrically converted analog electrical signal from 3b pixel ds _i, dr
Converted to _i . Line buffer memories 5a and 5b temporarily store the output signals ds _i and dr _i of one scanning line from the A / D converters 4a and 4b.
6 is a sum-of-products calculator,
Sum of products Σ of pixel output values ds _i and dr _i output from b
(Ds · dr) is calculated. Reference numerals 7a and 7b _denote adders, which calculate addition Σds and Σdr of the pixel output values ds _i and dr _i output from the line buffer memories 5a and 5b. Numerals 8a and 8b are sum-of-products calculators, which calculate square sums Σds ² and Σdr ² of the pixel output values ds _i and dr _i output from the line buffer memories 5a and 5b. Reference numeral 9 denotes a multiplier, which multiplies the output data Σ (ds · dr) from the product-sum calculator 6 by an arbitrary integer m and A = m
・ Calculation for obtaining Σ (ds · dr) is performed. Reference numeral 10 denotes a multiplier, which outputs the output data Σds of the adder 7a and the adder 7a.
The output data Σdr of b is multiplied by B = Σds · Σdr. An ALU 11 calculates a difference Δ = AB between the result A of the multiplier 9 and the result B of the multiplier 10. 1
Reference numeral 2 is a comparator, which compares the result Δ of the ALU 11 with an arbitrary integer value. A sub-correlation calculator 13 calculates the denominator E = (C · D) ^1/2 of the equation for obtaining the normalized correlation value COR _k . That is, C = m · Σds ² − (Σds) ² D = m · Σdr ² − (Σdr) ² E = (C · D) ^1/2 is obtained.

Reference numeral 14 denotes a correlation calculator, which calculates a normalized correlation value COR _k of the reference visual field corresponding to the standard visual field.
Reference numeral 15 is a pixel shift amount detector, which searches for the maximum value of the normalized correlation value COR _k calculated by the correlation calculator 14, and obtains the pixel shift amount q = p ₀ −i ₀ + k from the shift number k in that case. ing. A distance measurement data memory 16 stores the pixel shift amount q obtained by the pixel shift amount detector 15 and the start address i ₀ of the reference visual field corresponding to the shift amount. Reference numeral 17 is a host computer, which controls the entire apparatus. Reference numeral 18 denotes a bus, which connects the distance measuring unit and the host computer 17. Reference numeral 19 is a robot, and 20 is a communication cable. Among the above blocks, the A / D converters-4a and 4b to the host computer 17 are the normalized correlation calculation means.

In this embodiment, the stereo optical system shown in FIG. 2 is adopted. 3 and 4 are flowcharts of the operation of this embodiment.

Using these figures, C in this embodiment
The process of obtaining distance information from the images captured by the CDs 3a and 3b and storing the result in the distance data memory 16 will be described.

The CCDs 3a and 3b detect the illuminance distribution of the image formed on the CCD 3a and 3b, and take in the illuminance distribution for each pixel. At this time, the number of horizontal scanning lines is equal to the number N of pixels in the Y direction of the CCD in this embodiment. That is, one pixel is scanned by one scanning line. The images captured by the CCDs 3a and 3b are output as video signals. First, the flow
In step S1 of the chart, the horizontal scanning line counter-j (Y-axis direction counter in FIG. 2C) of the CCD is set to 1.
Set to. Next, in step S2, it is determined whether the horizontal scanning line counter-j exceeds or does not exceed the limit value N. If j is N or less, the process proceeds to the correlation value calculation process from step S4. , J exceeds N CC
The process is terminated assuming that the calculation over the entire D is completed (step S3).

In step S4, the A / D converter is used.
The converters 4a and 4b convert each video signal on the j-th scanning line on the CCDs 3a and 3b from one scanning line analog signal to a digital signal at the same timing (step S4), and each converted digital signal is converted into a line buffer memory 5a. ，
It is temporarily stored in 5b (step S5). As a result, the CCD 3 is stored in the line buffer memories 5a and 5b.
The digitized video signal of the jth column on a and 3b, that is, the illuminance distribution of the jth pixel column is stored as an output value from each pixel. The pixel output value (pixel value) from the CCD 3a is d
s _i , the pixel output value (pixel value) from the CCD 3b is dr _i
And Next, in step S20, it is determined whether or not the calculation is completed for all the reference visual fields to be set in the jth column. In this case, since it is not over, the process proceeds to step S21, and the starting address i ₀ of the reference visual field La is set. Then, in step S22, the reference visual field L
The starting address p ₀ of b is set and the operation proceeds to step S6.

The processing flow after step S6 will be described. First, the pixel shift counter-k is set to 0 (step S6). Then line buffer memory 5
From the pixel output value ds _i stored in a, the reference visual field La
The m pixel outputs are extracted from a portion, that is, the address i ₀ , and Σds is calculated by the adder 7a (step S8-
2). In addition, the pixel output value dr _i of the reference visual field Lb portion stored in the line buffer memory 5b is from the p ₀ th to m
The pixel output is taken out, and Σdr is calculated by the adder 7b (step S8-1). Further, using the pixel output values ds _i and dr _i extracted here, the product-sum calculator 6 calculates Σ
(Ds · dr) is calculated (step S8-3). At the same time, Σds ² and Σdr ² are calculated by the product-sum calculators 8a and 8b (steps S8-4 and S8-5). Next, using the result Σ (ds · dr) of the product-sum calculator 6 and the number m of pixels in the reference visual field La, the multiplier 9 obtains A = m · Σ (ds · dr) (step S9-2). At the same time, B = Σds · Σdr is obtained by the multiplier 10 using the results of the adders 7a and 7b (step S9-1). Next, using the result A of the multiplier 9 and the result B of the multiplier 10, the difference Δ = by the ALU 11.
AB is obtained (step S10).

Next, the comparator 12 determines the sign of the difference Δ of the ALU 11 (step S11). If the difference Δ is negative or 0, it is determined that the distance data is indefinite, and the subsequent calculation of the normalized correlation value COR _k is stopped, and step S1
Go to 5 and set COR _k to any negative number (eg -1)
Is set, the process proceeds to step S23, the calculation of k shift is completed, and the correlation calculation of k + 1 shift is performed.

If the difference Δ is positive, the normalized correlation value CO
Find R _k to the end. That is, the adders 7a, 7b,
A sub-correlation calculator 13 using the results of the multipliers 8a and 8b
The value E is calculated by (step S12). Further, using this result and the difference Δ of the ALU 11, the correlation calculator 14 calculates a normalized correlation value COR _k (step S1).
3).

Next, the normalized correlation value CO obtained here
The host computer 17 determines whether R _k is larger or smaller than a preset threshold value. If the value of the normalized correlation value COR _k is larger than the threshold value, it is directly S23.
If the value of the normalized correlation value COR _k is smaller than the threshold value, it is determined that there is a high possibility that the result is due to the variation in the sensitivity of the CCD pixels, and the distance data is undefined.
Arbitrary negative number (eg -1) as the normalized correlation value COR _k
Is set (step S16), and the process proceeds to S23.

At the next step S23, the shift counter--
Store k and the normalized correlation value COR _k in memory.

With the above, one cycle of the calculation of the normalized correlation value COR _k is completed, and then the shift counter -k is set to k = k +.
Step forward by 1 (step S24), repeat the processing of steps S7 to S23 for this, and obtain the normalized correlation value COR.
Calculate _k . This is repeated from k = 0 to k = n-m, that is, within a range not exceeding the reference field of view. Shift No. is stored in the memory. And the corresponding normalized correlation value COR
_k is stored.

[0035] proceeds to step S17 when the shift counter -k exceeds n-m, the maximum value COR _k of COR _k stored in the memory by the pixel shift amount detector _{15, MAX}
And the shift number k _MAX in that case is calculated (step S1
7) The deviation amount q = p ₀ −i ₀ + k in the number of pixels and the starting address i ₀ of the reference visual field La (image Ia ′) in this case are stored in the distance measurement data memory 16 (step S18).

Next, returning to S20, it is judged whether or not all the reference visual fields to be set in the j-th column have been calculated. If the calculation is not completed, the next reference visual field La (image) on the j-th row is calculated in step S21. Ia ') is set, and the process proceeds to step S22 and beyond. If you judge that it is over, C
The correlation calculation of the jth column on the CD area sensors-3a and 3b is completed. After that, the j-column counter-j is incremented by 1 to obtain j =
The amount of misalignment q
ask for _j . By the above, the distance information of the entire screen on the CCD 3a is obtained.

Next, using the distance information of the entire screen and the pixel width τ stored in the distance measurement data memory 16, the host computer 17 can detect the object in a certain direction as necessary. The distance is calculated, and the robot 19 is transmitted via the communication cable 20.
Control.

In this embodiment, the CCD area sensor-
Although the number N of pixels in the Y direction is equal to the number of horizontal scanning lines, one pixel may be scanned by a plurality of horizontal scanning lines.

[0039]

As described above, according to the present invention, (2-1) When the distance to an object is obtained by the stereo method, the pixel output values of the standard visual field and the reference visual field do not satisfy the preset calculation values. In that case, it is possible to reduce the enormous amount of calculation and reduce the processing time for distance information measurement by determining that the distance is indefinite and omitting subsequent calculations, and it is possible to obtain distance information in real time for the obtained object field of view. Become.

(2-2) If the normalized correlation value COR _k is less than a certain threshold value, it is considered that the distance is indefinite, and the information that the distance of this portion is indefinite is given to give an erroneous result due to variation in pixel sensitivity. By providing a means for excluding the image portion to be measured from the measurement target or for distinguishing the distance data calculated from this image portion, erroneous distance information can be removed and the accuracy of the calculated distance information can be improved. .

(2-3) When obtaining the distance information to the object by the stereo method, C as the pixel value distribution measuring means
A distance information measuring device capable of efficiently obtaining a normalized correlation value between a standard field of view and a reference field of view even if there is some variation in the sensitivity of each pixel of a CD, and capable of obtaining distance information with high accuracy and at high speed. be able to.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a plan view of an optical system layout diagram for the stereo method (a). (B) is a side view. (C) is a front view of the pixel value distribution measuring means.

FIG. 3 is a flowchart of an embodiment of the present invention.

FIG. 4 is a flowchart of an embodiment of the present invention.

[Explanation of symbols]

I object Ia ', Ib' image 2a, 2b lens 3a, 3b pixel value distribution measuring means 4a, 4b A / D converters 5a, 5b line buffer memory 6 product-sum calculator 7a, 7b adder 8a, 8b product-sum Calculator 9 Multiplier 10 Multiplier 11 ALU 12 Comparator 13 Sub-correlation calculator 14 Correlation calculator 15 Pixel shift amount detector 16 Distance measurement data memory 17 Host computer 18 Bus 19 Robot 20 Communication cable La Standard Field of view i ₀ Starting address of standard field of view Lb Reference field of view p ₀ Starting address of reference field of view

Claims (3)

[Claims] 1. A reference system having a lens system and a large number of pixels, and a pixel value distribution measuring means for obtaining image information of an object image formed by the lens system by two-dimensional scanning, and a reference system having the same configuration. Pixel value distribution measurement of the reference system when the reference systems are arranged so that their optical axes are parallel to each other, and the direction connecting the two optical axes on the measurement surface is the X direction and the direction perpendicular to this is the Y direction. A standard field of view is set on the pixel column of Y coordinate y _j on the means, and a reference field of a region wider than the standard field of view on the pixel column of y _j whose Y coordinate is the same as the standard field of view on the pixel value distribution measuring means of the reference system. And sequentially calculating the normalized correlation value between the pixel output value from the selected portion and the pixel output value from the reference field of view in the reference field of view using the normalized correlation calculation means. Has the highest correlation with the standard visual field among the standard visual field and the reference visual field Determine the amount of deviation in the X direction of the minute,
In the distance information measuring device for calculating the distance information about the object from the deviation amount, the normalized correlation calculating means calculates the normalized correlation only when the pixel output values of the reference visual field and the standard visual field satisfy a preset calculation value. A distance information measuring device characterized in that calculation is performed. 2. The normalized correlation calculating means calculates the sum of products of each pixel value of an arbitrary standard image area and each pixel value of a reference image area obtained by the pixel value distribution measuring means in each image area. If the difference between the value obtained by multiplying the total number of pixels of the reference image area and the value obtained by multiplying the total value of the pixel values of the reference image area by 0 is less than or equal to 0, 2. The distance information measuring device according to claim 1, wherein the subsequent calculation of the normalized correlation is interrupted. 3. When it is determined that the normalized correlation value calculated by the normalized correlation calculation means is smaller than a preset threshold value, it is considered that the detection of the distance information is indefinite, and information to that effect is given. The distance information measuring device according to claim 1, wherein is recorded or displayed.