JPH09138125A - Distance measuring apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a distance measuring apparatus, which is not greatly affected by noises. SOLUTION: This apparatus has the following parts performing respective functions. Data input terminals receive the first and second sensor data outputted from the first and second optical sensors, which are constituted of a plurality of pixels. In sensor-data adding parts 5R and 5L, the first sensor data are received from the data input terminal, the sensor data of a plurality of neighboring pixels are added and the first add data are formed, the second sensor data are received from the input terminal, the sensor data of the same number of the neighboring pixels as in the case of the first added data are added and the second data are formed. A correlation operating means 7 operates the correlation degree of the first added data and the second added data. A measured- distance-value operating means 8 operates the measured distance value in correspondence with the result of the operation by the correlation operating means.

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,
In particular, it relates to a distance measuring device using a triangulation method.

[0002]

2. Description of the Related Art FIG. 16A is a schematic view for explaining an external light triangular optical system of a phase difference detection type distance measuring device.
Light beams 134B, 1 emitted from the distance measurement object 133
34R passes through two lenses 131B and 131R,
It is projected on the two sets of optical sensors 132B and 132R.

Light beam 134 passing through reference lens 131B
B is imaged on the reference optical sensor 132B, and the image of the distance measurement object 133 is projected. The light beam 134R passing through the reference lens 131R is imaged on the reference light sensor 132R, and the image of the distance measurement object 133 is projected. The image of the distance measurement object 133 is projected on the reference light sensor 132B and the reference light sensor 132R, respectively.

If the distance measuring object 133 is located at infinity from the lens 131, the distance between the image formed on the standard light sensor 132B and the image formed on the reference light sensor 132R is equal to that of the lens 131B. The base line length B is the distance between the optical axis and 131R. As shown in the figure, when the distance measurement object 133 is separated from the lens 131 by the distance L, the image formed on the reference light sensor 132B and the reference light sensor 132 are formed.
The distance between the images formed on R is B + x. That is, an image is formed on the optical sensor 132 with a distance of the phase difference x in addition to the baseline length B.

The distance f between the lens and the sensor is determined by the lens 131.
To the surface of the optical sensor 132 on which the optical image of the distance measurement object 133 is projected. The distance measuring distance L is the distance from the distance measuring object 133 to the lens 131, and this distance is measured as the distance from the distance measuring device to the distance measuring object.

As shown in the figure, it is assumed that the distance measuring object 133 is on the optical axis of the reference lens 131B. At this time, the optical axis of the reference lens 131R, the reference lens 1
The two triangles formed by the ray passing through the center of 31R, the object plane including the distance measurement object 133, and the image plane on the optical sensor 132 are similar, and the relationship of L / B = f / x (1) holds. .

That is, as shown in FIG. 16 (B), L = B · f / x (2) holds.

When x is represented by the number n of sensor pitches p, L = B · f / (n · p) (3)

The sensor pitch p is an interval between a plurality of light receiving elements forming an optical sensor, and takes a value of, for example, about 20 [μm]. At this time, the denominator is represented with an accuracy of an integral multiple of the sensor pitch p.

The sensor pitch p is further converted into k by using an interpolation method.
When it is divided and when x is represented by the subdivision, it corresponds to i, x = i (p / k), and L = B · f / (i / k) p (4) Become. In other words, the denominator can be represented by the interpolation method with an accuracy that is an integral multiple of p / k, and the distance measurement distance L that is more accurate than Expression (3) is obtained.

Next, the correlation calculation performed to obtain the phase difference x between the image on the standard light sensor 132B and the image on the reference light sensor 132R will be described. FIG. 17 is a conceptual diagram for explaining the phase difference detection by the correlation calculation.

FIG. 17A shows an image formed on the photosensor. The optical sensors 132B and 132R are line sensors in which a plurality of photodiodes are arranged one-dimensionally.
The number of photodiodes forming the photosensor 132 corresponds to the number of pixels of the image formed on the photosensor 132.
The number of pixels of the reference light sensor 132R is the standard light sensor 132R.
The number of pixels of B is equal to or more than that.

The reference optical sensor 132B includes a reference lens 1
An image of the object for distance measurement is formed via 31B. An image of the object to be measured is formed on the reference light sensor 132R separated from the reference light sensor 132B in the horizontal direction of the base line via the reference lens 131R.

When the object to be measured is at the infinity position, the same image is formed on the light receiving elements of the photodiodes corresponding to the standard light sensor 132B and the reference light sensor 132R. If the object to be measured is not at infinity, the optical sensor 13
The images on 2B and 132R are displaced in the horizontal direction. That is, the distance between the images increases as the object to be measured approaches, and the distance between the images decreases as the object to measure distance increases. In order to detect the variation in the distance between the images, the reference light sensor 132R
In many cases, the number of pixels is set to be larger than that of the reference light sensor 132B.

In order to detect the variation in the distance between the image on the standard light sensor 132B and the image on the reference light sensor 132R, a phase difference detection method by a correlation calculation is used. In the phase difference detection by the correlation calculation, the correlation value S (n) of the pair of image formations on the optical sensors 132B and 132R is obtained by the calculation based on the following equation (5), and these image formations are performed until the correlation value becomes the minimum. The relative movement value (phase difference) of is calculated.

S (m) = Σ (k = 1 to i) | B (k) −R (k + m) | (5) where Σ (k = 1 to i) is k from 1 to i Represents the sum of functions. k designates a pixel in the reference light sensor 132B. Further, m is, for example, an integer from −6 to 6 and indicates the relative movement amount in the reference light sensor.

B (k) is an electric signal output from each pixel of the reference photosensor 132B in time series, and R (k +)
m) is an electric signal output from the pixel of the reference light sensor 132R in time series.

FIG. 17B shows the relationship between the pixel shift amount and the correlation value. If the calculation of the above equation (5) is performed every time the pixel shift amount m is changed from −6 to 6, the correlation values S (−6), S (−5), ... (6) is obtained. For example, when the correlation value S (0) is the minimum value, the distance to the object to be measured is set in advance to a predetermined value. When the correlation value at the position deviated from this is the minimum value, the deviation of the object to be measured from the predetermined position, that is, the distance to the object to be measured can be detected by the amount of deviation.

By the way, the light receiving elements of the standard light sensor 132B and the reference light sensor 132R are arranged at a pitch of 20 [μm], for example. The correlation value is 20 on the image plane.
It is calculated for each shift distance in units of [μm]. When the distance to the object to be measured corresponds to the middle position of the pitch of the light receiving element, make sure that the correlation value on the right side of the extreme value of the correlation value and the correlation value on the left side are different as shown by the broken line in the figure. Become. In such a case, it is possible to obtain a resolution higher than the resolution determined by the pitch interval by performing the interpolation calculation.

FIG. 17C is a schematic diagram for explaining the three-point interpolation method. The position where the minimum correlation value is obtained is defined as x2, and the sample positions on both sides thereof are defined as x1 and x3. The correlation value actually obtained by calculation is shown by a black circle. As shown in the figure, when the correlation value y3 at x3 is lower than the correlation value y1 at x1, it is considered that the true minimum value exists somewhere in advance from x2 to x3.

If the minimum value is exactly at the position of x2, the correlation value curve is bent at x2 as shown by the broken line g1 and rises symmetrically, the correlation value y3a at x3 is the correlation value y1 at x1. Is equal to

On the other hand, if the midpoint of x2 and x3 is the position of the true minimum correlation value, the correlation value curve is bent at the midpoint of x2 and x3 as shown by the broken line g2, and the correlation value y2 at x2 becomes The correlation value y3b at x3 becomes equal. As shown in the figure, the difference (y
3a-y3b) is the difference in correlation value between x1 and x2 (y1-
equal to y2). That is, the correlation value of one unit changes as the pitch advances by half a pitch. Therefore, the true correlation value minimum position can be obtained by checking which intermediate position of the above-mentioned two cases the correlation value actually obtained by the calculation is. The shift amount d from x2 is given by d = (y1-y3) / 2 (y1-y2) (6) when the distance between adjacent sample points is 1.

[0023]

To obtain the distance measurement value,
The correlation calculation of Expression (5) is performed. The correlation value is a value obtained by obtaining the absolute value of the difference between the standard sensor data and the reference sensor data in each pixel and integrating the absolute value for each pixel. When the sensor data contains noise, the noise is also integrated.

Since the correlation calculation takes the difference between the standard sensor data and the reference sensor data, when the noise is white noise, the noise is canceled and the correlation value is not affected by the noise.

However, when the noise is clock noise, the correlation value may be affected by the noise. The clock noise is due to the sensor data transfer clock.
This is the noise that occurs. A transfer clock is used when capturing sensor data from the optical sensor.

It is considered that the clock noise is generated by, for example, the signal line of the sensor data and the signal line of the transfer clock affecting each other on the cable or the circuit. The clock noise is stationary noise that is present in the sensor data depending on the clock frequency. When the clock noise has a cycle that is a multiple of 2 with respect to the sensor data, the phase of the noise may match due to the pixel shift when the correlation calculation is performed.

That is, depending on the amount of pixel shift, noise may be reflected largely or hardly in the correlation value. This causes the correlation curve to collapse,
The correct phase difference cannot be obtained.

An object of the present invention is to provide a distance measuring device that is resistant to noise.

[0029]

A distance measuring device of the present invention includes a data input terminal for receiving first and second sensor data output from first and second photosensors each of which is composed of a plurality of pixels. The first sensor data is received from the data input terminal, the sensor data of a plurality of adjacent pixels are added to generate first addition data, the second sensor data is received from the data input terminal, and the first addition is performed. A sensor data adding unit that adds sensor data of the same number of adjacent pixels as in the case of data to generate second added data;
Correlation calculation means for calculating the correlation degree between the addition data and the second addition data, and distance measurement value calculation means for calculating the distance measurement value according to the calculation result by the correlation calculation means.

By performing a correlation operation on the added data obtained by adding the sensor data of the adjacent pixels, the function of a low-pass filter can be achieved and the influence of noise can be removed. It is possible to remove the influence of noise and obtain an appropriate distance measurement value.

[0031]

1 is a diagram for explaining the function of a distance measuring device according to an embodiment of the present invention. The sensor unit includes a right sensor unit 1R and a left sensor unit 1L. The right sensor unit 1R includes right lenses 2R and 2R.
It has a three-dimensional optical sensor 3R. The left sensor unit 1L is
It has a left lens 2L and a two-dimensional optical sensor 3L. The optical sensors 3R and 3L convert the amount of light incident through the lenses 2R and 2L into electric signals for each pixel.

The right sensor data fetching section 4R fetches an electric signal generated by the optical sensor 3R and outputs it as right sensor data DR. The left sensor data capturing unit 4L captures the electric signal generated by the optical sensor 3L and outputs it as the left sensor data DL.

For example, the sensor data DR and DL are fetched in synchronization with the transfer clock. At that time, due to the influence of the transfer clock, clock noise is likely to occur in the sensor data DR and DL.

The clock noise is generated not only when the sensor data DR and DL are fetched. If the sensor data signal line and the clock signal signal line are adjacent to each other, clock noise may occur. More specifically, the circuit configuration will be described later.

The horizontal addition processing units 5R and 5L perform horizontal addition processing on the sensor data of the two-dimensional pixels on a line-by-line basis. The horizontal addition processing unit 5R adds four pixels adjacent to the right sensor data DR in the horizontal direction of the line sensor to generate horizontal addition data DR ′. Next, shift by one pixel and add the data DR of the next horizontal 4 pixels,
Horizontal addition data DR ′ of the next pixel is generated.

FIG. 2 shows a method of generating horizontal addition data DR '(j) based on the sensor data DR (j). The sensor data DR (j) indicates the sensor data of the j-th pixel. The horizontal addition data DR ′ (1) and DR ′ (2) are data obtained by adding sensor data of four adjacent pixels,
Each is calculated by the following equation.

DR '(1) = DR (1) + DR (2) +
DR (3) + DR (4) DR '(2) = DR (2) + DR (3) + DR (4) +
DR (5) Horizontal addition data DR '(j) is obtained by the following equation.

DR ′ (j) = DR (j) + DR (j + 1) + DR (j + 2) + DR (j + 3) (7) Similarly, the horizontal addition processing unit 5L determines the horizontal of the line sensor for the left sensor data DL. Four pixels adjacent in the direction are added to generate horizontal addition data DL '.

The horizontal addition data DR 'and DL' are horizontal 4
Since pixels are added, noise having a cycle of 2 pixels or 4 pixels is softened. The horizontal addition processing units 5R and 5L function as low pass filters. Horizontal addition is not limited to 4 pixels. It suffices to add pixels that are integer multiples of the noise period.

In FIG. 1, the correlation calculator 7 obtains the correlation value between the horizontal addition data DR 'and DL'. The correlation value S (m) can be obtained by S (m) = Σ (k = 1 to i) | DR ′ (k) −DL ′ (m + k) | (8) That is, the correlation value S (m)
Is calculated by changing k from 1 to i with respect to the shift number m and adding the absolute values of the differences between DR ′ and DL ′ at each k.

After that, the minimum correlation value S (m) among the correlation values S (m) is calculated, and the shift number m at that time is calculated.
After that, three-point interpolation according to the equation (6) is performed to obtain a highly accurate shift number (phase difference).

The distance measurement value calculation unit 8 calculates the distance measurement value L by the equation (4) based on the phase difference calculated by the correlation calculation unit 7, and outputs it. FIG. 3 shows a specific example of the sensor data DR, DL.
The horizontal axis is the pixel position, and the vertical axis is the sensor data. The sensor data is originally two-dimensional data, but since the processing is performed for each line data, the line data will be described below.

The sensor data DR is right sensor data output from the right sensor data fetching section 4R. The sensor data DL is left sensor data output from the left sensor data acquisition unit 4L. Sensor data DR, D
L includes high frequency clock noise. The shift amount (phase difference) between the right sensor data DR and the left sensor data DL is d. The data a1 of the pixel with the right sensor is shifted by d in the left sensor and is captured as a2. However, clock noise is superimposed.

FIG. 4 shows a specific example of the horizontal addition data DR ', DL'. The horizontal axis is the pixel position, and the vertical axis is the horizontal addition data. The horizontal addition data DR ′ is data obtained by adding the horizontal 4 pixels of the right sensor data DR of FIG. The horizontal addition data DL ′ is data obtained by adding the horizontal 4 pixels of the left sensor data DL in FIG. Horizontal addition data D
Clock noises of R ′ and DL ′ are relaxed as compared with sensor data. The shift amount (phase difference) between the horizontal addition data DR and DL is d, as in the case of the sensor data DR and DL in FIG. The horizontal addition data a1 ′ corresponding to the pixel of the right sensor data a1 is deviated from the horizontal addition data a2 ′ corresponding to the pixel of the left sensor data a2 by d.

Next, the principle of addition of sensor data will be described. As shown in FIGS. 17A to 17C, the correct phase difference d can be obtained by performing the correlation calculation between the right sensor data DR and the left sensor data DL. In the present embodiment, instead of this, horizontal addition data DR ′ and D
It was shown that the correlation calculation with L'is performed. It will be described that the correct phase difference d can be obtained even in this case. That is, the data D obtained by adding the sensor data
The phase difference d is preserved between R ′ and DL ′.

The case where horizontal addition is performed on four pixels has been described above, but a case where horizontal addition is performed on two pixels will be described for simplicity of explanation. In that case, the horizontal addition data DR ′ (j) and DL ′ (j) are represented by DR ′ (j) = DR (j) + DR (j + 1) DL ′ (j) = DL (j) + DL (j + 1), respectively. To be done.

FIG. 5 shows general basic functions f (x), f
(X-d), g (x), g (x-d) are shown. Function f
(X) and the function g (x) are independent of each other. Function f
(X-d) is a function deviated from the function f (x) by d. The function g (x-d) is d with respect to the function g (x).
It is a function that is deviated.

Here, the function f (x) is the data DR (j).
And the function f (x−d) corresponds to the data DL (j). The function g (x) corresponds to the data DR (j + 1),
The function g (x-d) corresponds to the data DL (j + 1).

FIG. 6 shows a function H obtained by adding basic functions.
(X) and H (x-d) are shown. When the function H (x) is represented by H (x) = f (x) + g (x), the function H (x-d) is H (x-d) = f (x-d) + g (x- d) can be represented.

Here, the function H (x) is the horizontal addition data D
It corresponds to R ′ (j), and the function H (x−d) corresponds to horizontal addition data DL ′ (j). At this time, it is proved that the phase difference d is also preserved between the function H (x) and the function H (x-d). Therefore, even if the correlation calculation is performed between the function H (x) and the function H (x-d), the correct phase difference d can be detected.

As described above, even if the sensor data DR and DL are horizontally added, horizontal addition data DR 'and DL' are obtained, and the correlation calculation is performed between the data DR 'and the data DL', the correct positions are obtained. The phase difference d can be detected.

Since the data DR 'and DL' are obtained by adding the sensor data of the objects having different distances, there is a problem in that respect.

However, originally, the two-dimensional image projected on the sensor is a mixture of objects in perspective and perspective, and the mixture of perspective is approved from the beginning. Therefore, regarding the sensor data, no special problem occurs even if data with different distance information is added.

FIG. 7 shows the overall structure of the distance measuring device according to this embodiment. The sensor unit includes a right sensor unit 1R and a left sensor unit 1L. Right sensor unit 1R
Has a right lens 2R and a two-dimensional optical sensor 3R, and the left sensor unit 1L has a left lens 2L and a two-dimensional optical sensor 3L. The optical sensors 3R and 3L are, for example, a plurality of 100-pixel one-dimensional line sensors arranged in parallel, and convert the amount of light incident through the lenses 2R and 2L into electric signals.

The system controller 14 is, for example, a microcomputer, receives a timing pulse generated by the sensor control system 12, and returns a sensor control signal to the sensor control system 12. The sensor control system 12 supplies a drive pulse to the drive circuits 11R and 11L based on the sensor control signal. The drive pulse is about 10 MHz.

The drive circuits 11R and 11L drive the sensors 3R and 3L according to the drive pulse. The sensor data is supplied from the output terminals of the sensors 3R and 3L to the sensor data quantization unit 13. At that time, the sensor data is affected by the drive pulse and includes clock noise. The sensor data quantizer 13 quantizes sensor data.

The system controller 14 supplies a write control signal and a write address to the image memory 15,
Write control is performed. The image memory 15 is, for example, SRA.
M, and stores the quantized data quantized by the sensor data quantization unit 13 according to the write control signal.

The arithmetic unit 16 is, for example, a RISC microcomputer, and sends a transfer control signal to a parallel input / output circuit (parallel I
O) supply to 18. The parallel input / output circuit 18 takes in conditions such as an area for which distance measurement is desired from an external device 19 and transfers it to the data transfer memory 17. The data transfer memory 17 is, for example, SRAM.

Further, the arithmetic unit 16 supplies the read / write control signal and the address to the image memory 15 or the data transfer memory 17. The calculation unit 16 reads the sensor data DR and DL from the image memory 15 and performs horizontal addition processing on, for example, four adjacent pixels of the sensor data to generate horizontal addition data DR ′ and DL ′. Store in the buffer.

After that, the correlation calculation between the data DR 'and the data DL' is performed. After the correlation calculation, three-point interpolation is performed to obtain a highly accurate phase difference d. After obtaining the phase difference d, the distance measurement value is calculated.

The arithmetic section 16 writes the obtained distance measurement value in the data transfer memory 17. The parallel input / output circuit 18 receives the transfer control signal from the arithmetic unit 16 and transfers the distance measurement value stored in the data transfer memory 17 to the external device 19.

FIG. 8 is a flow chart showing the processing procedure of the distance measuring apparatus according to this embodiment. The distance measuring device can detect the distance measuring value in each area in a plurality of areas in the two-dimensional area.

In step S1, the sensor data obtained from the optical sensors 3R and 3L is written in the image memory 15 as two-dimensional image data. The system controller 14 performs the process of step S1. The calculation unit 16 performs the subsequent processing.

In step S2, various conditions obtained from the external device 19 are loaded into the data transfer memory 17. The various conditions include, for example, designation of an area for which distance measurement is desired, designation of the size of the area, and the like, and are used when calculating a distance measurement value later.

In step S3, it is determined whether or not the processing of all lines has been completed. The two-dimensional image data consists of a plurality of lines, and processing is performed for each line. If the arithmetic processing has not been completed for all lines, the process proceeds to step S4.

In step S4, the image memory 15
Capture line data. In step S5, horizontal addition processing of the captured line data is performed. The horizontal addition process is an important process in this embodiment. Arithmetic unit 16
Performs horizontal addition processing of, for example, 4 pixels based on the sensor data DR, DL fetched from the image memory 15 according to the equation (7) to generate horizontal addition data DR ', DL'.

In step S6, it is determined whether or not the processing for all areas has been completed. The line data includes a plurality of areas. When the arithmetic processing is not performed for all areas in the line data, the process proceeds to step S7.

In step S7, the correlation calculation for one area is performed. The correlation calculation is performed between the data DR ′ and the data DL ′ using the equation (8). After that, three-point interpolation according to the equation (6) is performed to obtain a highly accurate phase difference d. Then, the distance calculation is performed using the equation (4) to obtain the distance measurement value.

In step S8, the data of the obtained distance measurement value is written in the transfer memory 17. After that, the process returns to step S6 to obtain the distance measurement value for the next area. When it is determined in step S6 that the calculation has been completed for all areas in the line data, step S3
Return to and repeat the process for the next line.

If it is determined in step S3 that the calculation has been completed for all lines, the process is ended.

[0071]

[Embodiment] Next, a simulation result of correlation calculation using the distance measuring apparatus of this embodiment will be shown.

FIG. 9 shows sensor data DR (n) and DL (n) when there is no clock noise. n indicates a pixel. Each sensor data DR in which the pixel n is from 0 to 31
(N) and DL (n) are shown.

The right sensor data DR (n) has 106, 109, 112, 115, 116,
118, ... The left sensor data DL (n) is
103, 106, 109, 11 according to the order of the pixel n
2, 115, 116, ... FIG. 10 is a graph showing the right sensor data DR (n) and the left sensor data DL (n). The sensor data DR (n) and DL (n) include
There is no clock noise.

The correlation value S (m) is a value obtained according to the equation (5). m is the pixel shift amount. The shift amount m is −
The correlation value S (m) from 2 to 5 is shown. FIG. 15 (A)
In the graph, the relationship between the correlation value S (m) and the shift amount m is shown. Since there is no clock noise on the sensor data DR (n) and DL (n), this correlation value curve has a clean V
I draw a glyph.

FIG. 11 shows sensor data DR (n) and DL (n) when there is clock noise. Similar to FIG. 9, each sensor data DR in which the pixel n is 0 to 31
(N) and DL (n) are shown.

FIG. 12 is a graph showing the right sensor data DR (n) and the left sensor data DL (n). Clock noise is present on the sensor data DR (n) and DL (n).

NS is clock noise on the sensor data. Its amplitude is 4. When the pixel n is an odd number, noise of size 4 is added to both the right sensor data and the left sensor data. When the pixel n is an even number, noise is not added to either the right sensor data or the left sensor data.

The correlation value S (m) is a value obtained according to the equation (5), and shows the correlation value S (m) when the pixel shift amount m is from -2 to 5. FIG. 15B is a graph showing the relationship between the correlation value S (m) and the shift amount m. Sensor data DR
Since the clock noise NS is present on (n) and DL (n), the correlation value curve is broken and a W-shape is drawn.

This correlation value S (m) has two local minimum values, and an appropriate phase difference cannot be obtained. Therefore, in this embodiment, sensor data DR (n) with clock noise,
Horizontal addition processing is performed based on DL (n).

FIG. 13 shows horizontal addition data DR '(n) and DL' (n). Horizontal addition data DR '(n), D
L ′ (n) is data obtained by adding four horizontal pixels based on the sensor data DR (n) and DL (n) including the clock noise shown in FIG.

The correlation value S (m) is a value obtained according to the equation (8), and shows the correlation value S (m) when the pixel shift amount m is from -2 to 5. FIG. 15C is a graph showing the relationship between the correlation value S (m) and the shift amount m. An appropriate phase difference can be obtained by drawing a V-shape on the correlation value curve.

FIG. 14 is a table comparing the correlation values S (m), and FIG. 15 is a graph of the correlation values S (m).
9A shows the correlation value S (m) shown in FIG. 9, which is obtained when there is no clock noise. The phase difference obtained from this correlation value is normal. The minimum value of the correlation value S (m) is 0, and the shift amount m at that time is 1. Therefore, the phase difference d is about 1.

(B) is the correlation value S (m) shown in FIG. 11, and is when there is clock noise. Due to the influence of clock noise, the correlation value S (m) becomes an abnormal value when the shift amount m is an odd number. Originally, when the shift amount m is 1, the correlation value S (m) should be the minimum value, but when the shift amount m is 0 and 2, it is a smaller value.

(C) is the correlation value S (m) shown in FIG. 13, which is obtained when the horizontal addition processing according to this embodiment is performed. Even when the shift amount m is an odd number, the correlation value S (m) close to the normal value is obtained. The minimum value of the correlation value S (m) is 24, and the shift amount m at that time is 1.
Therefore, the phase difference d is about 1 as in (A).

Correlation value S (m) when horizontal addition is performed
Is slightly different from that when there is no clock noise, but the finally obtained phase difference d is almost the same value, and an appropriate phase difference can be detected. The horizontal addition process serves as a low-pass filter and eliminates the influence of clock noise.

In the above, the case where the period of the clock noise is 2 pixels has been shown, but the present invention can be applied to noise having a longer period. In that case, it is necessary to increase the number of pixels to be horizontally added, if necessary. It is also possible to remove the influence of noise other than clock noise.

The horizontal addition processing is not limited to horizontal addition for four pixels, but horizontal addition may be performed for pixels having an integral multiple of the noise period. For example, if the noise period is 2 pixels, horizontal addition is performed for 2, 4, 6, or 8 pixels. It should be noted that if the number of pixels is too large, the processing time becomes long, so 8 pixels or less is preferable.

If the number of pixels for horizontal addition is increased, the number of pixels is not limited to an integer multiple of the noise cycle, but may be any integer multiple or more. For example, when the noise period is 2 pixels, the effect of clock noise can be softened by performing horizontal addition for 5 pixels or 7 pixels.

The distance measuring apparatus according to this embodiment can be used for general cameras and on-vehicle multi-point distance measuring.
Although the present invention has been described with reference to the embodiments, the present invention is not limited thereto. For example, it will be apparent to those skilled in the art that various modifications, improvements, combinations, and the like can be made.

[0090]

As described above, according to the present invention,
By performing a correlation operation on the added data obtained by adding the sensor data of the adjacent pixels, it is possible to remove the influence of noise and obtain an appropriate distance measurement value.

[Brief description of the drawings]

FIG. 1 is a block diagram for explaining a function of a distance measuring device according to an embodiment of the present invention.

FIG. 2 is a schematic diagram showing a configuration of a horizontal addition processing unit.

FIG. 3 is a graph showing a specific example of sensor data DR and DL.

FIG. 4 is a graph showing a specific example of horizontal addition data DR ′, DL ′.

FIG. 5 shows general basic functions f (x), f (x−d), g
It is a graph which shows (x) and g (x-d).

FIG. 6 shows functions H (x) and H (x−, which are obtained by adding basic functions.
It is a graph which shows d).

FIG. 7 is a block diagram showing an overall configuration of a distance measuring device according to the present embodiment.

FIG. 8 is a flowchart showing a processing procedure of the distance measuring apparatus according to the present embodiment.

FIG. 9: Sensor data DR when there is no clock noise
It is a chart which shows (n) and DL (n).

FIG. 10 is the sensor data DR (n), DL shown in FIG.
It is a graph which shows (n).

FIG. 11: Sensor data D when there is clock noise
It is a chart showing R (n) and DL (n).

FIG. 12 is the sensor data DR (n), DL shown in FIG.
It is a graph which shows (n).

13 is a diagram showing sensor data DR (n) and DL shown in FIG.
Horizontal addition data DR ′ obtained by horizontal addition processing based on (n)
It is a chart which shows (n) and DL '(n).

14 is a correlation value S (m) shown in FIGS. 9, 11 and 13. FIG.
It is the chart which compared.

15 is a graph showing the correlation value S (m) of FIG.

FIG. 16 is a schematic diagram for explaining a method for measuring a distance measurement distance. FIG. 16A is a schematic diagram of an external light triangulation optical system of the phase difference detection type distance measuring device, and FIG. 16B is an arithmetic expression for calculating the distance measuring distance.

FIG. 17 is a diagram for explaining phase difference detection by correlation calculation. 17A is a graph showing image signals obtained in the reference part and the reference part, FIG. 17B is a graph showing obtained correlation value curves, and FIG. 17C is for explaining a three-point interpolation method. FIG.

[Explanation of symbols]

 1 Sensor Unit 2 Lens 3 Optical Sensor 4 Sensor Data Importer 5 Sensor Data Horizontal Addition Processor 7 Correlation Calculator 8 Distance Measurement Calculator 11 Drive Circuit 12 Sensor Control System 13 Sensor Data Quantizer 14 System Controller 15 Image Memory 16 Arithmetic unit 17 Data transfer memory 18 Parallel input / output circuit 19 External device

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Akio Izumi 1-1, Tanabe-Nitta, Kawasaki-ku, Kawasaki-shi, Kanagawa Prefecture Inside Fuji Electric Co., Ltd.

Claims (2)

[Claims] 1. A data input terminal for receiving first and second sensor data output from first and second photosensors each having a plurality of pixels, and a first sensor data for receiving the first sensor data from the data input terminal. , Add sensor data of a plurality of adjacent pixels to generate first added data, receive second sensor data from the data input terminal, and receive the same number of adjacent pixels as in the case of the first added data. Sensor data addition unit (5R, 5L) that adds the sensor data of (1) to generate second addition data, and a correlation calculation unit that calculates the degree of correlation between the first addition data and the second addition data ( A distance measuring device having 7) and a distance measuring value calculating means (8) for calculating a distance measuring value according to a calculation result by the correlation calculating means. 2. A step of receiving first and second sensor data output from first and second photosensors each of which is composed of a plurality of pixels, and a step of receiving a plurality of adjacent pixels with respect to the first sensor data. Add the sensor data to generate the first addition data,
Generating the second addition data by adding sensor data of the same number of adjacent pixels to the second sensor data as in the case of the first addition data; and the first addition data and the first addition data. 2. A distance measuring method including a step of calculating a degree of correlation with the addition data of 2, and a step of calculating a distance measuring value according to a calculation result of the correlation calculation.