JP3291998B2 - Distance measuring device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an external light triangulation type distance measuring device which can be employed in a vehicle rear-end collision prevention device and the like. Related to detection technology.

[0002]

2. Description of the Related Art Conventionally, an external light triangulation type (stereo type) distance measuring device mounted on an autofocus camera or the like, as shown in FIG. It comprises a compound-eye imaging optical system including a pair of left and right imaging lenses (positive lenses) 1 _R and 1 _L , and a distance measuring semiconductor integrated circuit 5. The integrated circuit 5 is arranged on a substantially focal plane of the imaging lenses 1 _R and 1 _L and converts a spatial one-dimensional illuminance distribution of the image into a photosensor array (photo image array) as a photoelectric conversion element group. Sensors) 2 _R , 2 _L , quantization circuits 3 _R , 3 _L for sequentially quantizing the analog output signals of the photosensor arrays 2 _R , 2 _L for each sensor (segment), and a pair of left and right collected digital values A logical unit 4 for performing a required logical operation on the basis of the image forming data sequence to calculate a distance signal
And

[0003] As shown in FIG. 5 (a), a pair of left and right imaging lenses 1 _R, 1 _L is the optical axis S _R, S _L are parallel focal length f
_e are the same, and are arranged on the same plane to form a binocular imaging optical system. The object T to be measured is a pair of left and right imaging lenses separated by a reference length (optical axis interval or interpupillary distance) B. 1 _R, are respectively the 1 _L imaged object image of each of the photosensor array 2 _R, on 2 _L inverted real image corresponding to the focal plane (space 1-dimensional intensity distributions) T _R, T _L is tied.

[0004] If the measured object T exists in far forward indefinitely, the imaging point is the optical axis S _R, the focal point on S _L (reference point) O _R,
To match the O _L. When the distance to the measured object T is finite length d, the distance is given by the following equation based on the principle of triangulation (similarity of a triangle). _{d = Bf e / (X R} + X L) = Bf e / X ... (1) where, X _R, X _L is the image point P _R of the representative object point P in the object T to be measured, P _L and an imaging lens 1 _R, 1 _L of the optical axis S _R, the distance between S _L (excursion), X is the sum of X _R and X _L, is the relative total deviation amount of an object image to be measured (phase difference) . Therefore, the distance d can be obtained by obtaining the spatial phase difference X.

However, since the measured object T is not scattered as the representative object point P but has a spatial spread,
As shown in FIG. 5A, the photo sensor arrays 2 _R and 2 _{R
On L} , object images (spatial one-dimensional illuminance distribution) T _R and T _L spanning a large number of sensor ranges in the array direction are formed, and are the same on the left and right photo sensor arrays 2 _R and 2 _L. The image points for the same object point on the measured object T cannot be easily specified. Therefore, the logic unit 4 of the integrated circuit 5, the illuminance of the same object image by examining the distribution correlation between image data string via imaging data sequence and the other photosensor array 2 _L by one photosensor array 2 _R pattern The phase difference X is obtained in the form of the number of segments of the photosensor array by finding the zone (window position) where.

FIG. 5B is an explanatory diagram showing an imaging position detecting method for examining the distribution correlation between left and right images. Right imaging data sequence D _R represents a digital value string corresponding to the space one-dimensional illumination intensity distribution obtained from the right side of the photosensor array 2 _R, shown in the rightmost segment data R ₀ ~ leftmost segment data R _n of the array It consists of n + 1 pieces of data. Also, the left image data sequence D _L represents the digital value string corresponding to the space one-dimensional illumination intensity distribution obtained from the left side of the photosensor array 2 _L, shown in the rightmost segment data L ₀ ~ leftmost segment data L _n It consists of n + 1 pieces of data. Now, since the photosensor array 2 _R, 2 _L measured object image on T _R, T _L is imaged, the detect these relative positional relationships, the hatching in FIGS. 5 (a) subjected shown was such an interval window for specifying or data extraction (window) W _R, sets the W _L, while shifting these windows located in the right and left alternately, imaging data sequence D shown in FIG. 5 (b) _R, respectively, from D _L window W _R, W _L within iNCLUDED partial data sequence D _Rk, correlating test whether coincide with each other in the left-right combination of D _Lj. Window W
_R, the width of W _L (total segment) number of data in the partial data string corresponding to w is the m + 1 (2m ≧ n> m). In the present application, for convenience, n is referred to as a value corresponding to the total number of segments, and m is referred to as a value corresponding to a window width.

[0007] Here, the shift method of the window W _R, W _L for creating a correlation function (evaluation function) are also known side shift method for shifting one by one the other window while fixing the side windows However, in this example, since the pair of left and right imaging systems are equal, a double-sided (left and right) alternate shift method is used. That is, in the first combination C (0,0) = C _0, and R _n to R _nm substrings D _R0 extracted from the leftmost of the right imaging data sequence D _R, the left image data string The correlation of the combination with the partial data string D _L0 of L _{0 to} L _m extracted from the rightmost side of D _L is tested. The correlation test method obtains, for example, the sum (correlation value) Σ of the absolute values of the differences between corresponding data between the right partial data strings R _{nm to} R _n and the left partial data strings L _{0 to} L _m . That is, Σ = | L ₀ −R _nm | + | L ₁ −R _{n−m + 1} | +... + | L _m −R _n | (2) The smaller the correlation value Σ, the smaller the partial data sequence The combination (distribution) is matched, which means that the correlation is strong.

When the measured object T is located at infinity in front of the center line L, the first combination C (0,0) = C ₀ should have the highest correlation, that is, the correlation value Σ should be the minimum value. . Followed next combination C (0,1) = C ₁ since this is the window W _R, W _L sequentially shifted right and left partial data sequence extracted position from the inside, one segment of the alternately left and right to the outside Be combined. That is, C (0,0) → C (0,
1) → C (1,1) → C (1,2)... C (nm−1,
nm-1) → C (nm-1, nm) → C (n−
m, nm). The total number of combinations is (2 (nm) +1). Generally, right and left partial data strings D _Lj (right end data L _j to left end data L _{m + j} ),
If the combination of D _Rk (right end data R _nmk to left end data R _nk ) is C (j, k) = C _i , the number of one-side shifts j,
Since k advances from 0 to nm, the total number of shifts i =
j + k takes a value of 0 to 2 (nm). Here, since i can be regarded as a combination variable (combination number), a pair of parts for all combinations C _i data sequence D _Lj,
As a result of performing a correlation test of D _Rk to obtain a correlation function (evaluation function) Σ (i), assuming that the σ-th combination Cσ has the highest correlation, the combination number σ indicating the highest correlation is an index of the phase difference X. . The phase difference X can be obtained by multiplying the number σ by the sensor pitch p, but the index σ is generally used as it is as the distance signal. The correlation test is based on the premise that most of the window fields overlap before and after the shift, and that the edges of the window fields overlap at the innermost and outermost positions. The number of times (nm) is set to m or less.

[0009]

By the way, when the total number of segments (total number of sensors) n + 1 of the photosensor arrays 2 _R and 2 _L is increased, it is advantageous for widening the azimuth of short-distance measurement or multi-directional ranging. However, if the total number of segments is increased, the integrated circuit becomes expensive. Therefore, at present, the total number of segments is 2 ⁸ =
About 256 is normal. Therefore, the maximum window width w _max (= m _max +1) is about half of the total number of segments n + 1. Here, the window width w (=
When m + 1) is wide, it means that an image on the line sensor is cut out at a wide viewing angle, and therefore the probability of extracting an illuminance pattern including a sharp contrast increases,
Since the probability of capturing an object image that is mixed in perspective is also high, for example, an approximate intermediate distance of a perspective object is detected, erroneous distance measurement is likely to occur, and the reliability of the measured value decreases. Conversely, when the window width w is narrow, it means that an image on the line sensor is cut out at a narrow viewing angle, and thus the probability of capturing a mixed object image is low, but the illuminance pattern including sharp contrast is reduced. The probability of extraction is low, making it difficult to detect correlation, making it difficult to measure distance.

In the case of a window width w (= m + 1),
The total number of shifts on one side is w-1 (= m), and the total number of combinations is 2
w−1 (= 2m + 1), the number of correlation value calculations is w
(2w-1), which is proportional to the square of the window width w. Therefore, if the window width w is large, the correlation function Σ
The number of calculations for obtaining (i) increases drastically, and the processing time increases. When a distance measuring device is applied to a collision prevention device, particularly high-speed arithmetic processing is required. However, if the window width w is excessively widened, it cannot be satisfied. Conversely, if the window width w is narrow, the processing time can be reduced, but the total number of shifts w-1 is also reduced, so that the window position cannot be adjusted to the outside, and a large phase amount X is captured. And short distance measurement becomes impossible.

In view of the above problems, the first aspect of the present invention
An object of the present invention is to provide a distance measuring apparatus capable of suppressing erroneous distance measurement regardless of distance by performing a shift operation of a window position by changing the length of a window width w. It is an object of the present invention to provide a distance measuring apparatus capable of reducing the number of correlation operations and realizing a faster processing time.

[0012]

In order to solve the above-mentioned problems, a measure taken by the present invention resides in that a variable window width is adopted so as to achieve optimization of the window width.
That is, the present invention converts a spatial one-dimensional illuminance distribution of an image formed by a pair of image forming optical systems that form a parallax between images facing a distance measurement target and an image forming optical system on one side into an electric signal sequence. A photoelectric conversion sensor array on one side, a photoelectric conversion sensor array on the other side for converting a spatial one-dimensional illuminance distribution of an image formed by the imaging optical system on the other side into an electric signal sequence, and an electric signal from the photoelectric conversion sensor array on the one side. One-side partial data extracting means for extracting one-side partial data sequence included in one-side window from the image data sequence obtained from the signal sequence, and an image obtained from the electric signal sequence of the photoelectric conversion sensor array on the other side Other-side partial data extraction means for extracting the other-side partial data string included in the other-side window of the data string; window movement control means for controlling the relative position between the one-side window and the other-side window to shift; One side partial data And a maximum correlation detecting means for detecting a combination of partial data strings having a maximum correlation based on a correlation function obtained from the correlation testing means. In the distance measuring apparatus, the distance measuring apparatus includes a test validity determination unit that determines the validity of the correlation test for the combination of the partial data strings, and a window width adjustment unit that increases or decreases the window width when the test is invalid. It is characterized by the following.

Here, the window width adjusting means is preferably a window width increasing means. Further, the partial data extracting means may be a thinning extracting means.

[Operation] When a correlation test is performed within a certain window width, and the test is invalid, as a result, the window width is reduced or increased by the window width adjusting means, and the correlation test is performed again. . For this reason, the correlation test can always be performed with an appropriate window width according to the measurement environment, and the measurement error can be suppressed. In particular, when the window width increasing means is used, the number of partial data can be reduced as much as possible, so that the correlation calculation processing can be sped up. In addition, when the thinning-out extraction unit is used, even if the window width is expanded by performing the correlation test again, the increase in the number of data is suppressed, so that the processing time can be shortened by suppressing the number of calculations. .

[0015]

BEST MODE FOR CARRYING OUT THE INVENTION

[0016]

Next, an embodiment of the present invention will be described with reference to the accompanying drawings.

[First Embodiment] FIG. 1 is a block diagram showing a configuration of a logic unit of a distance measuring apparatus according to a first embodiment of the present invention. As shown in FIG. 4, the distance measuring apparatus of the present embodiment also includes a pair of left and right imaging lenses (positive lenses) 1 _R and 1 _L that form a binocular parallax facing the measured object T. And a distance measuring semiconductor integrated circuit 5. This integrated circuit 5
Is a photosensor array (for example, a CCD) 2 _R as a photoelectric conversion element which is arranged on a substantially focal plane of the imaging lenses 1 _R and 1 _L and converts the image (one-dimensional illuminance distribution) into an electric signal sequence. , 2 _L and, photosensor array 2 _R, 2 _L quantization circuit 3 sequentially quantizes the analog output signal of each sensor (segment) of _R, 3 _L and a pair of left and right imaging data collected digital values A logic unit 4 for performing a required logical operation process based on the columns D _R and D _L to calculate a distance signal.

[0018] The example logic unit 4, as shown in Figure 1, the right side image data string from the right of the quantization circuit 3 _R D _R of (rightmost segment data R ₀ ~ leftmost segment data R _n) of the right portion data extraction section 10 for extracting right partial data string D _Rk (window rightmost data R _NMK ~ window leftmost data R _nk) included in the right side window W _R, the left of the left imaging from the quantization circuit 3 _L image data string D _L left partial data string D _Lj (window rightmost data L _j ~ contained in the left window W _L of the (rightmost segment data L ₀ ~ leftmost segment data L _n)
A left portion data extraction section 20 for extracting window leftmost data L _{m + j),} a window movement control means 30 for controlling so as to shift operation the position of the right window W _R and the left window W _L alternately left and right, the right partial data Row D _Rk
And a left side partial data sequence D _Lj , a correlation test means 40 for calculating a correlation value for the combination C _i, and a maximum correlation combination number σ is detected based on the correlation function Σ (i) obtained from the correlation test means 40. the maximum correlation detection means 50, the test enable determination means 60 for determining the validity of the correlation test for combination number i, window width for gradually increasing the right window W _R and the left window W _L of the window width w when invalid test Adjusting means (window width changing means) 70.

The right-side partial data extracting means 10 stores right-side image data D _R (right-most segment data R ₀ to left-most segment data R _n ) from the right quantization circuit 3 _R. (RAM) 12 and a right partial data string D _Rk (window right end data R
_nmk- address AD of window left end data R _nk )
(A right side data address counter) 14 that specifies (window right end address nmk to window left end address nk) in ascending order for each clock pulse CK applied to the clock terminal CL. Further, the left portion data extraction section 20 is also left image data storage unit for storing the left image data string from left quantization circuit 3 _L D _L (rightmost segment data L ₀ ~ leftmost segment data L _n) (RAM) 22 and the left partial data string D _Lj to be extracted (the right end data L
_j to the window left end data Lm _{+ j} ) address AD (window right end address _j to window left end address m +
j) is specified in ascending order for each clock pulse CK applied to the clock terminal CL, and a preset up counter (left side data / address counter) 24 is provided. Right partial data extracting means 10 and left partial data extracting means 20
The left and right partial data sequence D _Lj, and preset with the down counter (data butt end detecting circuit) 16 for detecting the end of the data butt for calculating the difference data between corresponding in combination C _i of D _Rk, then output Circuit 18 for generating a carry delay signal c _DL as a timing signal by slightly delaying a carry signal (matching end signal) c, and a carry delay signal C _DL and a test start signal ST as two inputs and an up counter 1 as an output.
OR gate 1 to supply to 4, 24 preset terminals PS
And 5. The down counter 16 with preset is used for the test start signal ST via the OR gate 19 or the delay circuit 1
Window width corresponding value m from the window width adjusting means 70 described later each time undergoing 8 carry delay signal c _DL of the preset terminal PS and is preset, a clock terminal C
Every time the clock pulse CK is added to L, the countdown is performed. The carry delay signal c _DL is supplied to a preset terminal PS of the up counters 14 and 24 with presets. Each time the carry delay signal c _DL is generated, the up counter with presets 14 moves the window movement control means 30.
Window rightmost data R _NMK window rightmost number in the right window W _R supplied from the (n-m-
with k) is preset, the preset with up counter 24 the window right edge number j of the window right edge data L _j in the left window W _L supplied from the window movement control means 30 is preset.

The window movement control means 30, the distance measuring start signal ST first combination C ₀ right window rightmost number (n-m) preset with the down counter to preset the supplied from the window width adjusting means 70 by the occurrence of ( Right window right end designating circuit) 31 and clear C
The first combination C is determined by the ranging start signal ST applied to the R terminal.
₀ up counter (left window right end designation circuit) 32 that is initialized to the left window rightmost number 0, the 1/2 frequency dividing circuit 33 which divided by 2 carry signal c, the 1/2 frequency divider Inverter 34 for inverting the output of 33
And

The down counter 31 has its clock terminal C
Each time the output of the frequency divider 33 applied to L goes high (when an even pulse of the carry signal c), the count-up is performed, and the up-counter 32 sets the inverted signal applied to its clock terminal CL to high. At the time of the odd pulse of the carry signal c, the countdown is performed.

The correlation test means 40 generates a partial data string D _Lj ,
A subtraction circuit 42 for taking the difference between the corresponding data in the combination C _i of D _Rk , an absolute value circuit 44 for taking the absolute value of the difference value, and an addition circuit 46 for successively adding in synchronization with the clock pulse CK. Have. Addition circuit 46 is partial data sequence D _Lj, the correlation value for the combination C _i of D _Rk (cumulative value) sigma
Is calculated. The carry delay signal c _DL via the OR gate 48 is applied to the clear terminal CR of the adder circuit 46.

The maximum correlation detecting means 50 outputs the carry signal c
And a comparison circuit 52 that generates a small value detection signal DT when the correlation value given from the adding circuit 46 is smaller than a storage value given from the correlation value register 54 when the correlation value given by the adding circuit 46 is smaller than the storage value given by the correlation value register 54. A combination of a correlation value register 54 which receives the small value detection signal DT at the latch terminal L and temporarily stores the correlation value と き at that time, and a partial data string D _Lj and D _Rk which receives the carry signal C at the clock terminal CL. An up counter 56 for counting the number i, and a maximum combination number register 58 for receiving the small value detection signal DT at the latch terminal L and temporarily storing the combination number i = σ of the partial data strings D _Lj and D _Rk at that time. Have.

The test validity determination means 60 outputs the small value detection signal D
This is a flip-flop that receives T at the set terminal S and receives the distance measurement start signal ST at the reset terminal R. When the small value detection signal DT is generated once, it is set and the Q output becomes high level, indicating that the test is valid. When the small value detection signal DT is never generated, the Q (bar) output becomes a high level, indicating that the test is invalid.

The window width adjusting means 70 adds the initial window width set value w ₀ supplied from the outside or created internally each time the window width extension control signal CW is applied to the clock terminal CL, and adds the window width w. _N (= N
w ₀ ) and a window width corresponding value m by taking the difference between the window width w _N and the numerical value 1
(= W _N −1).

A value n smaller than the total number of segments of the photosensor array by 1 (a value corresponding to the total number of segments) is given from the outside or created internally.
Takes the difference between the total number of segments corresponding value n and the window width corresponding value m, makes the window right edge to the right of the number window W _R of the first combination C ₀ between partial data (right partial data string) (n-m). The rightmost number of the right window (n-
m) is supplied to the preset down counter 31 and the multiple circuit 82. The value 2 obtained by the multiple circuit 82
(N-m) is equivalent to the total number of shifts of the left and right window W _R, W _L. Then, the comparison circuit 83 outputs the carry signal c
Every time is applied to the enable terminal EN.
A comparison is made as to whether the number of combinations (number of shifts) i supplied from 6 exceeds the total number of shifts 2 (nm) supplied from the multiple circuit 82, and if it exceeds, outputs a shift end signal ED.

In the window width adjusting means 70, the window width extension control signal CW is generated by an AND gate 73 which receives the Q (bar) output from the test validity determining means 60 and the shift end signal ED as two inputs. In the AND gate 74 which receives the Q output from the validity determination means 60 and the shift end signal ED as two inputs, a clear signal of the multiplying circuit 71 is created. The window width adjusting means 70 also has a window width extension confirmation circuit 75. Whether this window width extended check circuit 75, a multiple circuit 75a to the window width w _N to twice the time of occurrence of the window width extension control signal CW, the multiple value 2w _N exceeds the corresponding value n the total number of segments If it exceeds, the distance measurement impossible signal NG
And a comparison circuit 75b that outputs

FIG. 2 is a flowchart showing the operation of the present embodiment. Hereinafter, the operation of this example will be described with reference to FIG.

[0029] When the measurement operation is started, the right image data stream D _R (rightmost segment data R ₀ ~ leftmost segment data R _n) on the right image data storage unit from the right side of the quantization circuit 3 _R ( while being stored in the RAM) 12, the left side of the quantization circuit 3 the left image data string from _L D _L (rightmost segment data L ₀ ~ leftmost segment data L _n)
Are also stored in the left-side imaging data storage (RAM) 22.
The first, in the initial setting process in step S1, initial the multiplier circuit 71 of the logic unit 4 outside of the total number of segments corresponding value n, the initial window width w ₀ is given window width adjusting means 70 shown in FIG. 1 the value of the window width w ₀ is set. In this example, before shifting (reference position), which corresponds to infinity measurement, the initial window width w ₀ is
A narrow window width suitable for long distance measurement. For the sake of explanation, if n = 2 ⁸ = 256, the maximum possible window width is 2 ⁷ as described above. Here, it is assumed that w ₀ = 2 ⁵ = 32. Early output of the multiplier circuit 71, window width w _N = w ₀ = 2 ^5, and the output of the subtracting circuit 72 (window width corresponding value) m is w _N -1
= W ₀ -1 = 2 ⁵ -1. The output 2w _N of the multiple circuit 75a of the window width extension confirmation circuit 75 is 2w ₀
= 2 ⁶ , n> 2w ₀ , and the comparison circuit 75a of the window width extension confirmation circuit 75 outputs the distance measurement impossible signal NG
Does not occur, and the distance measurement with the window width w _N = w ₀ is allowed.

When the correlation test start signal ST is generated, the following correlation test is executed in step S2. First, a down counter (right window right end designating circuit) 31 of the window movement control means 30, an up counter 14 of the right partial data extracting means 10, an up counter 24 of the left partial data extracting means 20, and a down counter (data matching end detecting circuit). 16 is preset and the up counter (left window right end designation circuit)
32, the addition circuit 46 of the correlation test means 40 and the up counter 56 of the maximum correlation detection means 50 are cleared. Further, the correlation value register 54 is set to a predetermined value, and the test validity judging means 60 is reset. Up counter 56
Is clear, the count value (combination number) i = 0,
It refers to the first combination C _0. The down counter 31
Right window rightmost number of the first combination C ₀ supplied from the subtracting circuit 81 (n-m-k) is preset to (n-m). The right partial data extracting means 10
Is also preset at the right end address (nm) of the right window. On the other hand, up counter 3
Since the count value of 2 is cleared to zero, this corresponds to the fact that the right end number j of the left window of the first combination C ₀ has been preset to zero. Further, the up counter 24 of the left partial data extracting means 20 is also preset to the right end address 0 of the left window. The window counter value m is preset in the down counter 16. This means designation of the remaining number of times of data matching in the calculation of the correlation value of the first combination C ₀ .

When the correlation test start signal ST is added, (nm) is preset in the up counter 14 as an address counter, so that the data R _nm of the address ( _nm ) in the right side imaging data storage section 12 is stored. Since it is read out and 0 is preset in the up counter 24,
The data L _{0 at} address 0 in the left imaging data storage unit 22 is read. The two read data R _nm, L ₀ are subtracted by the subtraction circuit 42 of the correlation test means 40 to obtain a difference value (L ₀ −
R _nm ), and the absolute value | L ₀ −R _nm | obtained by the absolute value conversion circuit 44 is stored in the addition circuit 46 as it is.

When the first clock pulse CK ₁ is applied,
The count value of the down counter 16 (the number of remaining matches) becomes m-1, and the up counters 14 and 24 count up. Therefore, the data R _{n−m + 1} of the address (nm−1) of the right imaging data storage unit 12 is At the same time, the data L at the address 1 of the left
₁ is read. Both read data R _{n-m + 1,} L ₁
Is subtracted by the subtraction circuit 42 of the correlation test means 40 to obtain a difference value (L ₁ −R _{n−m + 1} ).
The absolute value | L ₁ −R _{n−m + 1} |
Is added to the existing value | L ₀ −R _nm |.

Thus, when the m-th clock pulse CK _m is applied, the count value of the down counter 16 (the number of remaining matches) becomes 0, the count value of the up counter 14 is n, and the count value of the up counter 24 is m. Therefore, the data R _n of the address n of the right imaging data storage unit 12 is read out, the left image data storage unit 2
Data L _m of second address m is read. The read data R _{n and} L _m are subtracted from the subtraction circuit 4 of the correlation test means 40.
Is subtracted by 2 difference value (L _m -R _n) is obtained, followed by
The absolute value | L _m −R _n | obtained by the absolute value converting circuit 44 is already added by the adding circuit 46 (| L ₀ −R _nm | +... + | L
_m−1 −R _n−1 |), and the total added value Σ (0) = (|
L ₀ −R _nm | +... + | L _m −R _n |).

Next, the (m + 1) th clock pulse C
When K _{m + 1} is added, the down counter 16 generates the first carry signal c ₁ and declares the end of the matching calculation of the first combination C ₀ . When generating the carry signal c _1, since applied to the comparison circuit 52 of the maximum correlation detection means 50 as an enable signal, the comparator circuit 52 is the total sum value Σ
It is compared whether (0) is smaller than the set value of the correlation value register 54, and when it is smaller, a small value detection signal DT is output. Here, when the small value detection signal DT is output, the correlation value register 54 temporarily stores the total added value Σ (0). The highest combination number register 58 stores the combination number i.
= 0 is temporarily stored. When the small value detection signal DT is not output, the set value remains in the correlation value register 54 as it is.
A carry signal c ₁ is generated, the count value is i = 2 next to the up-counter 56, the next combination C ₂ declared. Further, when the first carry signal c ₁ is generated, the output of the 1/2 frequency dividing circuit 33 is at low level, the down counter 31 is not incremented, only the up-counter 32 is incremented, the following combinations left window rightmost number j of C ₂ becomes 1. Furthermore, the down counter 16 is preset to m again. Thereafter, in the same manner as described above, the correlation calculation of the combined C ₂ is performed. When the correlation calculation up to the combination C _{2 (nm)} is performed in this manner, the combination number σ of the minimum correlation value Σ (σ) is output from the highest combination number register 58, and the combination number i = 2 (nm) ), The comparison circuit 83 outputs the combination end signal ED.

In the present example, the first combination C ₀ corresponds to the measurement at infinity, and the narrow window width w _N = w ₀ and the narrow viewing angle suitable for the measurement are set. Can be reduced, and the number of data in the partial data string is small, so that the correlation test can be sped up. In this example, in step S3 of the correlation test processing of the number of combinations 1 + 2 (nm), the test validity determination means 60 determines whether or not the small value detection signal DT has been generated. If the test is valid, step S4 is performed using the highest combination number σ.
Performs the distance calculation. Any correlation value Σ (i) is larger than the initial value of the correlation value register 54, and the highest combination number σ
Does not exist, meaning that a correlation test is not possible. This is convenient when data is captured in a narrow field of view with a narrow window width w ₀ , which is convenient when a wide-range object exists as described above, but is unsuitable for measurement of a short-range object and tends to be unable to be verified. .

Therefore, in this example, if the correlation test is not possible in step 5, the window width extension control signal CW is output after waiting for the output of the combination end signal ED, and the window width w _N is doubled by the window width adjusting means 70. The partial data string is extracted again with the window width w _N = 2w ₀ , and the correlation test is performed again.

That is, when the window width extension control signal CW is applied to the clock terminal CL of the multiplying circuit 71, the multiplying circuit 7
1 adds the value w ₀ of the initial window width to the already added value w _0.

As a result, the window width w _N becomes 2w ₀ = 2.
It becomes ⁶ . Then, the output m (the value corresponding to the window width) m of the subtraction circuit 72 becomes w _N -1 = 2w ₀ -1 = 2 ⁶ -1. Further, in step 6, the output 2w _N multiple circuit 75a of the window width extended check circuit 75, 4w ₀ = 2
^{Since 7} is satisfied, n> 4w ₀ , and the comparison circuit 75a of the window width extension confirmation circuit 75 does not generate the distance measurement impossible signal NG, and allows the distance measurement with the window width w _N = 2w ₀ . In the redo correlation test, the number of partial data is doubled. Although the number of times this time is increased compared to the number of times of the previous correlation calculation, the viewing angle is expanded, so that the probability of extracting an illuminance pattern including sharp contrast increases, and the correlation detection becomes suitable for short-range measurement. .

If the test validity discriminating means 60 declares impossible even in the correlation test with the window width w _N = 2w ₀ , the correlation test with the window width w _N = 3w ₀ = 3 × ²⁵ is performed again. Since the window width w _N = 5w ₀ is more than half 2 ⁷ of the total number of segments, in step 7, where outputs a distance disable signal NG measurement comparison circuit 75 performs a default process.

As described above, in the distance measuring apparatus of the present embodiment, the correlation test is performed with a narrow window width, and when the test cannot be performed, the window width is gradually increased and the correlation test is performed again. Therefore, the correlation test can always be performed with an appropriate window width according to the measurement environment. Since an image formed on the photosensor array can be cut out with a viewing angle as narrow as possible, the probability of capturing an object image that is mixed in a distance is reduced, and the measurement accuracy can be improved. In addition,
At a narrow viewing angle, the probability of extracting an illuminance pattern containing a sharp contrast is low, making it difficult to grasp the object image at a short distance as a whole. Does not matter. Further, since the number of partial data can be reduced as much as possible, the correlation calculation process can be sped up.

In the correlation test, a one-sided shift method in which one window is fixed and the other window is shifted may be used instead of alternately shifting the left and right windows. Further, instead of shifting the left and right windows from the inside to the outside, the left and right windows may be shifted from the outside to the inside. The first combination may correspond to a measurement at a close distance instead of a measurement at infinity. Further, in the window width adjusting method, instead of gradually increasing the window, a wider window width may be initially set, and the window may be gradually reduced in the redo correlation test.

[Second Embodiment] FIG. 3 is a block diagram showing a configuration of a logic unit of a distance measuring apparatus according to a second embodiment of the present invention. The logic unit 4 'in the distance measuring apparatus of this embodiment is different from the logic unit 4 of the first embodiment in that a frequency division ratio variable frequency dividing circuit 90 is added in order to reduce the number of calculations of the correlation test. The dividing ratio variable dividing circuit 90 increases the dividing ratio 1 / N every time the window width extension control signal CW is generated, and the down counter 1
The application of the carry signal c from 6 to the counters 14 and 24 is thinned out. When the window width expansion control signal CW is generated first, the frequency division ratio becomes 1/2 and the carry signal c is applied to the counters 14 and 24 every other pulse, so that the carry signal c is extracted even if the window width is doubled. The number of data in the partial data string remains w ₀ , and the number of calculations of the correlation test is the same. For this reason, although the window width is expanded when the correlation test is performed again, the increase in the number of data is suppressed, so that the processing time can be shortened by suppressing the number of operations.

The mounting directions of the pair of image forming lenses 1 _{R and} 1 _L and the photosensor arrays 2 _{R and} 2 _L are not limited to the left and right directions, but may be up and down or oblique directions.

[0044]

As described above, the distance measuring apparatus according to the present invention comprises a test validity determination means for determining the validity of a correlation test for a combination of partial data strings, and a window width when the test is invalid. A window width adjusting means for increasing or decreasing the window width, wherein the window width is variable so as to achieve optimization of the window width during measurement. Therefore, the following effects are obtained.

Even if the correlation test is invalid, the window width adjusting means reduces or increases the window width, and performs the correlation test again. For this reason,
The correlation test can always be performed effectively with an appropriate window width according to the measurement environment, and the measurement error can be suppressed.

In particular, when the window width increasing means is used, the number of partial data can be reduced as much as possible, so that the correlation calculation processing can be sped up.

When the thinning-out extraction means is used, the increase in the number of data is suppressed even if the window width is expanded by performing the correlation test again, so that the processing time is shortened by suppressing the number of calculations. be able to.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a logic unit in a distance measuring apparatus according to a first embodiment of the present invention.

FIG. 2 is a flowchart illustrating an operation of the distance measuring apparatus.

FIG. 3 is a block diagram illustrating a configuration of a logic unit in a distance measuring apparatus according to a second embodiment of the present invention.

FIG. 4 is a block diagram showing a schematic configuration of an external light triangulation type distance measuring apparatus.

FIG. 5A is an explanatory diagram showing a measurement principle of the external light triangular distance measuring device, and FIG. 5B is an explanatory diagram showing an imaging position detecting method of the distance measuring device.

[Explanation of symbols]

T: object to be measured 1 _R , 1 _L ... imaging lens 2 _R , 2 _L ... photosensor array 3 _R , 3 _L ... quantization circuit 4, 4 '... logic unit 5 ... distance measuring semiconductor integrated circuit 10 ... right part Data extraction means 12 Right image data storage unit (RAM) 14 Up counter with preset (right side data / address counter) 15 OR gate 16 Down counter with preset (data matching end detection circuit) 18 Delay circuit 19 OR gate 20 left-side partial data extraction means 22 left-side imaging data storage unit (RAM) 24 up-counter with preset (left-part data address counter) 30 window movement control means 31 down-counter with preset (right window right end designation Circuit) 32 ... Up counter (Left window right end designation circuit) 3 3 1/2 frequency divider 34 inverter 40 correlation test means 42 subtraction circuit 44 absolute value conversion circuit 46 addition circuit 48 OR gate 50 maximum correlation detection means 52 comparison circuit 54 correlation value register 56 Up counter 58 ... Highest combination number register 60 ... Test validity discriminating means 70 ... Window width adjusting means 71 ... Multiplying circuit (addition circuit) 72 ... Subtraction circuit 73,74 ... And gate 75 ... Window width expansion confirmation circuit 75a ... Multiplication circuit 75b ... comparison circuit 81 ... subtracting circuit 82 ... multiple circuits 83 ... comparator circuit 90 ... dividing ratio variable frequency dividing circuit D _R ... right image data sequence (the rightmost segment data R ₀
To the leftmost segment data R _n ) D _L ... Left-side imaging data string (rightmost segment data L ₀₎
~ Leftmost segment data L _n) W _R ... right window W _L ... left window D _Rk ... right partial data string (window rightmost data R
_nmk ~ window left end data R _nk ) D _Lj ... left partial data string (window right end data L _j ~
Window left end data L _{m + j} ) C _i ... combination of partial data i ... combination number Σ (i) ... correlation function σ ... combination number of maximum correlation c ... carry signal (matching end signal) c _DL ... carry delay signal DT ... small value detection signal CK ... clock pulse ST ... correlation test start signal ED ... combined end signal NG ... distance measuring impossible signal CW ... window width extended control signal w _n ... window width w ₀ ... initial window width m ... window width corresponding value n ... Value corresponding to the total number of segments.

Continuation of the front page (56) References JP-A-4-113211 (JP, A) JP-A-4-243349 (JP, A) JP-A-6-201379 (JP, A) JP-A-5-157558 (JP) , A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01C 3/06 G06T 1/00 G08G 1/16

Claims (3)

(57) [Claims] 1. A method for converting a spatial one-dimensional illuminance distribution of an image formed by a pair of image forming optical systems which form a parallax between images facing a distance measurement target into an electric signal sequence. Space 1 for imaging by the photoelectric conversion sensor array on the side and the imaging optical system on the other side
A photoelectric conversion sensor array on the other side that converts a two-dimensional illuminance distribution into an electric signal sequence, and one-side partial data included in one window of an imaging data sequence obtained from the electric signal sequence of the one-side photoelectric conversion sensor array One-side partial data extracting means for extracting a column, and another-side portion for extracting the other-side partial data sequence included in the other-side window from the imaging data sequence obtained from the electric signal sequence of the other-side photoelectric conversion sensor array Data extraction means, window movement control means for controlling a shift operation of the relative position between the one side window and the other side window, and correlation value calculation for a combination of the one side partial data string and the other side partial data string And a maximum correlation detecting means for detecting a combination of partial data strings of maximum correlation based on the correlation function obtained from the correlation testing means. A test validity determining means for determining the validity of a correlation test for the combination of the partial data strings; and a window width adjusting means for gradually increasing or decreasing the window width when the test is invalid. Distance measuring device. 2. A distance measuring apparatus according to claim 1, wherein said window width adjusting means is a window width increasing means for increasing said window width. 3. The distance measuring apparatus according to claim 1, wherein said partial data extracting means is a thinning-out extracting means for thinning out and extracting said image data.