JP3847390B2 - Ranging device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a distance measuring device used for a camera, a video, and the like, and more particularly to a technique for providing a device capable of more accurate distance measurement and focusing.
[0002]
[Prior art]
In general, distance measuring devices used for cameras and the like have many methods that use light. For example, an “active type” method that projects signal light from the distance measuring device side and a “passive” method that uses a luminance distribution image of an object. It is roughly divided into “type”. Both methods use triangulation as the basic principle. In the active type, the distance between the projection and reception positions is the basic length (that is, the baseline length), and in the passive type, the object is based on the parallax with reference to the two reception positions. The distance to the object is obtained according to the relative position difference of the object image. In order to observe the image of the object, it is necessary to measure the amount of light at each light receiving position. Therefore, a line sensor in which a plurality of optical sensors are arranged is used.
[0003]
First, the distance measuring method of this passive type distance measuring device will be described with reference to FIG. Reference numerals 1a and 1b denote two light receiving lenses, which are separated by a base line length B. Light incident from the subject 6 through these lenses forms light patterns as shown by the curved patterns 3a and 3b on the line sensors 2a and 2b, respectively. The relative positional difference x between the two patterns 3a and 3b obtained depends on B and the lens, the distance between sensors f and the subject distance L. That is, the following relationship is established.
x = B × f / L (formula a)
Japanese Patent Publication No. 7-54371 discloses a method for calculating a relative position difference by calculation from the output of such a line sensor (hereinafter referred to as correlation calculation).
[0004]
In addition, it is an important technique to use which part of the light pattern to measure the distance. For example, as shown in FIG. 5, if a person 6a is to be measured, the Rn2 part of the sensor array 2a If the incident light is taken into account, the brightness of the distant background mountain 6b is mixed (this is referred to as “mixing of perspective”), and correct distance measurement cannot be performed. That is, in order to correctly measure the distance, a technique of using only the light incident on the sensors R1 to Rn1 in the figure but not using the sensor signal of Rn2 is necessary. Japanese Patent Publication No. 5-88445 is known as a proposal of such a technique.
[0005]
[Problems to be solved by the invention]
However, the technique disclosed in Japanese Patent Publication No. 7-54371 does not touch on the above-mentioned problem of “mixing of perspective”. On the other hand, in the technique disclosed in Japanese Patent Publication No. 5-88445, this result is obtained after the correlation calculation. When it is inappropriate, the area of the sensor is switched. It takes extra time and reprocesses the data over and over, so a microcomputer (hereinafter abbreviated as “microcomputer”) is used. There are also problems such as an increase in memory capacity and program capacity for judgments and calculations. That is, even if these conventional techniques are simply combined, it is difficult to provide a high-speed and high-precision distance measuring device.
[0006]
The present invention has been made in view of the above points, and appropriately switches the distance measurement range of the passive type distance measurement device, does not measure unnecessary measurement areas, and enables high-speed calculation. An object of the present invention is to provide a distance device. For example, when the present invention is applied for focusing on a camera or a video, it is an invention for accurately focusing an object.
[0007]
[Means for Solving the Problems]
Therefore, the present invention takes the following means in order to solve the above-described problems and achieve the object. That is,
[1] A light-receiving unit that receives a subject image by a pair of light-receiving element rows having parallax and outputs a pair of subject image signal rows; and an output of the light-receiving unit; Among the correlation means for outputting the correlation value of the first portion, the entire subject image signal sequence of one of the pair of light receiving element sequences, and the subject image signal sequence of the other light receiving element sequence. Provided is a distance measuring device including a data input unit that selectively inputs only partial data of a second part that is narrower than the one part, and an interpolation unit that performs a predetermined interpolation operation based on the input data. To do.
[2] Light receiving means for receiving a subject image by a pair of light receiving element rows having parallax and outputting a pair of subject image signal rows, and a plurality of subject image signal rows having a first number of pixels. A means for dividing the block into a block; a correlation means for receiving the output of the light receiving means; and outputting a correlation value of the pair of subject image signal sequences; and if the correlation value satisfies a predetermined condition, A second number of pixels smaller than the number of pixels, a subject image signal sequence of one of the pair of light receiving element rows, and a subject image signal sequence of the other light receiving element row A distance measuring apparatus is provided that includes data input means for selectively inputting only data of the second number of pixels and interpolation means for performing a predetermined interpolation calculation based on the inputted data.
[3] Based on the parallax of the subject image by comparing the two light pattern signals with two line sensors that output two light pattern signals according to the luminance distribution of the subject image observed from different fields of view having parallax Based on the output result of the pitch difference calculation means, the pitch difference calculation means for calculating the relative position difference in units of the arrangement pitch of the line sensor, Used for comparison in the pitch difference calculation means Output of the line sensor Part of the light pattern signal Select this Selected light pattern signal Based on the above Based on parallax of subject image The relative position difference Array of An arithmetic control means for calculating with a finer precision than the pitch, the pitch difference calculating means is constituted by a hard logic circuit, and the arithmetic control means is constituted by a microcomputer and software. Providing equipment.
[0008]
DETAILED DESCRIPTION OF THE INVENTION
A plurality of embodiments according to the present invention will be described with reference to the accompanying drawings.
(First embodiment)
First, a method for calculating the relative position difference of the light pattern in the passive type distance measuring apparatus according to the present invention will be described with reference to FIGS.
In FIG. 1A, the relative position difference x of the light distribution incident on the sensor arrays 2a and 2b varies depending on the subject distance L due to the difference B (base line length) between the positions of the light receiving lenses 1a and 1b. . If the focal length of each light receiving lens is f, the subject distance L is obtained by the following equation.
L = B × f / x (Formula b)
Since each sensor of the sensor array outputs a current signal according to the amount of incident light, if these are converted into digital signals by the A / D converter 4, the above calculation is performed by the correlation calculation by the calculation unit 5 that calculates the image shift amount. The relative position difference x can be detected. The subject distance L is obtained by inputting this result into a calculation control means (CPU) 10 comprising a one-chip microcomputer or the like and calculating based on the above equation b. The above is the basic principle and general apparatus configuration of the passive type triangulation system.
[0009]
The deviation amount calculation function is generally composed of two processes as will be described later, but these functions may be incorporated in the CUP 10 as a control program. When the camera is focused using such a technique, the CPU 10 controls the operation of the camera, and an automatic focus (AF) function can be achieved by appropriately controlling the photographing focus lens and the like via an actuator such as a motor. Can be provided with a camera.
[0010]
In order to calculate the image shift amount, a calculation step (that is, a correlation calculation) for checking how much the image shifts in units of sensor pitch in both line sensors is required. Then, a calculation step (hereinafter referred to as interpolation calculation) for calculating the shift amount more accurately with a finer resolution is required.
When light is incident on the sensor array 2a in a pattern like the waveform 3a shown in FIG. 1 (a), the output magnitudes of the sensors R1 to R6 are as shown by bar graphs in FIG. 2 (b). Distribution 3a is obtained. Here, “R” indicates the right sensor, “L” indicates the left sensor, and subscripts 1 to 6 attached to these indicate the absolute position of the sensor position with respect to the optical axis of the light receiving lens, for example. If the same signals as the outputs R1 to R6 are output from the outputs L1 to L6 of the left sensor, the relative position difference x is 0, so that the subject distance L to be obtained is “infinity”.
If the subject is present at a "finite distance", the left sensor L shifted by the number S of sensors determined from the x and the sensor pitch SP has the outputs R1 to R6 as shown in FIG. A similar value of the output signal is obtained.
[0011]
The value FF (i) on the vertical axis in the graph of FIG.
FF (i) = Σ | R (i) −L (i) | (Formula c)
In other words, the result FF obtained by subtracting the output of the corresponding L sensor from the output of a certain R sensor and adding the absolute value for each sensor may be used. That is, first, Li is subtracted from Ri to take its absolute value, i is changed within a certain width, and these are added.
[0012]
Next, if one sensor of Ri or Li is shifted by one unit and the difference is taken in the same manner as the adjacent sensor which has previously taken the difference, FF (i + 1) can be expressed by the following equation.
FF (i + 1) = Σ | R (i + 1) −L (i) |
Thus, the FF can be obtained while sequentially changing the shift amount (hereinafter referred to as SIFT amount), but the SIFT amount where the FF, which is the sum of the difference between R and L, becomes the minimum value (Fmin) is the most. Since it is considered that the position is well matched, the SIFT amount in this case is obtained as S. The above is the outline procedure of the process related to the correlation calculation.
[0013]
Further, when the output distribution of both sensor arrays is illustrated with the above S taken into account, as shown in FIG. 2 (b), it is the same as the R side sensors with corresponding subscripts from the sensors shifted by S on the L side. Is obtained.
[0014]
Next, the “interpolation calculation” process will be described in detail with reference to FIGS. 2B to 2D. The actual image shift amounts on the two sensor arrays are not exactly shifted by the sensor pitch. For accurate distance measurement, the amount of image shift must be detected with a precision finer than the pitch. Therefore, an interpolation operation is performed. R and L in FIGS. 2B and 2C respectively represent a part of sensor outputs constituting the sensor arrays 2a and 2b in FIG.
FIG. 2 (d) shows a graph that has been shifted to the above-described S for which the “correlation calculation” has already been completed, and has been made easier to compare. That is, L0 to L4 should be accurately described as Ls to Ls + 4, but this S is omitted so as not to become complicated in description.
Here, it is assumed that light shifted by x with respect to the R reference is incident on the L sensor after shifting by S. At this time, for example, the light incident on R0 and R1 is mixedly incident on the L1 sensor, and similarly, the light shifted by x on the R reference is sequentially incident on each L sensor. It can be seen that the outputs (L1 to L3) are expressed as shown in [Equation 1] in FIG.
[0015]
Fmin and FF values F-1 and F + 1 obtained by shifting the shift amount in the plus and minus directions from Fmin are expressed by using the outputs of the respective Rn and Ln. It is expressed as Further, when [Expression 2] is expanded using [Expression 1], values Fmin, F-1, and F + 1 are expressed as [Expression 3] in FIG.
In addition, if {| R0−R1 | + | R1−R2 | + | R2−R3 |} in [Equation 3] is expressed as (ΣΔR), the previous deviation amount x does not depend on (ΣΔR). Is obtained by the calculation shown in [Expression 4] in FIG. This is the “interpolation calculation”.
[0016]
These calculations are performed by the calculation means 5 in FIG. 1, but may be performed by a calculation control means (CPU) 10 such as a one-chip microcomputer according to a predetermined program.
[0017]
An outline of the above-described calculation process can be expressed as shown in the flowchart of FIG. That is, first, it is specified which region of the sensor array is to be subjected to the “correlation calculation” using light incident thereon (S101). Next, the “correlation calculation” described with reference to FIG. 2A by the calculation means 5 is performed to obtain the shift amount (S102).
Further, from the result, the “interpolation calculation” described in [Expression 4] is further performed to obtain the above x (S103).
[0018]
An autofocus (AF) camera can be provided if the CPU 10 calculates and controls the amount of extension of the focusing lens based on the values S and x thus obtained. Although this arithmetic means can be constituted by a logic circuit, as described above, these arithmetic operations may be performed according to a predetermined program built in the CPU 10 without using such hardware arithmetic means. .
Such a case is represented by a flowchart as shown in FIG. That is, the CPU 10 inputs the output of each sensor of the sensor array (S105), and the CPU 10 performs the correlation calculation (S106) and the interpolation calculation (S107) as described above according to a predetermined program, and zooms based on the result. Focusing control is performed by calculating the feeding amount of the lens or the like (S108).
[0019]
Note that the operation by the logic circuit (that is, the hard logic circuit) can reduce the scale of the circuit and increase the speed, but it is difficult to determine the data contents and switch the width of the distance measurement area. On the other hand, such control is possible with microcomputer control, but it has the advantages and disadvantages that speeding up is difficult.
[0020]
For example, in a scene of a person and a mountain as shown in FIG. 4, if the subject 6a (person) is to be correctly measured, the distance measurement area W should be narrower than the width K of the person's face, that is, W < K relation holds. The reason is that, as the distance measurement area W is wider, the optical information of the background mountain 6b is added to the calculation, so that highly accurate distance measurement becomes difficult (that is, because of the occurrence of mixed perspective errors). Accuracy decreases).
[0021]
The distance measurement area W has the following relationship depending on the pitch SP of the sensor array, the number E of sensors to be used, the distance L, and the distance f between the sensor and the lens.
W = L × SP × E / f (formula e)
As can be seen from this relationship, the face width K is constant with respect to the distance L, while the distance measuring area W is proportional to the distance L. That is, even if the distance L to the subject changes, SP and f are constant so that the distance measurement area W can be made equal to or less than the face width K, so that there is only a method of switching the number of sensors E. On the other hand, if there is no perspective, the larger the correlation area is, the higher the accuracy is because the sensor noise is offset. Therefore, a wider correlation area is advantageous for short-distance shooting. As described above, if the width of the distance measurement area is variable, distance measurement can be performed with high accuracy over a wide range from a short distance to a long distance.
[0022]
However, in many cases, the above-mentioned problem of mixing of perspectives occurs at the level of the interpolation calculation, and the influence on the correlation calculation is small.
Therefore, in the first embodiment of the present invention, even if the circuit is configured with a relatively simple logic circuit, or when there is a mixture of perspectives, a correlation operation that does not cause a large error is performed with a high-speed hard logic circuit, while strict accuracy is achieved. The interpolation calculation to be performed in (1) is configured to be performed precisely by calculation according to software by the CPU. As a result, it is possible to realize a distance measuring device that achieves both high speed and high accuracy.
[0023]
The flowchart of FIG. 3 (c) shows a procedure related to the arithmetic processing as the first embodiment of the present invention compared to the flowcharts of FIG. 3 (a) and FIG. 3 (b). That is, in order to achieve both high speed and high accuracy, the arithmetic processing is controlled as shown in FIG.
First, a correlation area is designated (S109), and a correlation calculation is performed by a hard logic calculation circuit (S110). Based on the correlation value S as a result, the CPU 10 reads the data of the sensor array (S111), and performs software-like interpolation calculation by the microcomputer (S112).
Based on the obtained result, the zoom lens feed amount is calculated, and appropriate focus control is performed on the subject (S113).
[0024]
(Modification 1)
For the purpose of the present invention, the correlation calculation is not necessarily performed by the hard logic circuit. For example, if the speed can be increased by reducing the number of data in the area to be correlated, as in step S106 in FIG. It may be performed by software.
Similarly, the interpolation calculation may be performed by hardware as in step S103 of FIG. Regardless of which method is used, in correlation calculation and interpolation calculation, data to be used, work area, or the like is separated to increase the processing speed and accuracy.
[0025]
Here, for example, if only a correlation between sensor pitch units of simple light and dark light patterns is obtained, this function can be satisfied even with a circuit configuration as shown in FIG.
Sensor arrays 20a and 21b are arranged behind the optical axis of the light receiving lens 1a, and arrays 20b, 21b and 22b are arranged behind the light receiving lens 1b. Consider the case where light is obtained.
The transistors 23 and 24 constitute a current mirror circuit, and the photocurrent of the sensor 20 a becomes the collector current of the transistor 24. Therefore, the difference photocurrent of the sensors 20a and 21a is input to the amplifier 28 following the latter stage. When the photocurrent of the sensor 20a is larger than the photocurrent of the sensor 21a, the difference current flows into the amplifier 28. When the magnitude relationship is reversed, a difference current flows out from the amplifier 28. In other words, the amplifier output becomes L (Low) or H (High) depending on the magnitude relationship.
[0026]
Similarly, the sensor on the lens 1b side is also connected to the current mirror circuit of the transistors 25, 26 and 27, and the difference photocurrent flows into and out of the amplifiers 29 and 30 with reference to the output of the sensor 21b. These amplifier outputs become H (High) or L (Low) in the same manner as the amplifier 28 depending on the brightness on the sensor array.
[0027]
The amplifier 29 becomes L when the photocurrent of the sensor 21b is larger than that of the sensor 20b, but is inverted by the action of the inverter 31 and input to the AND circuit 32 following the subsequent stage. Further, when the photocurrent of the sensor 21b is larger than that of the sensor 22b, the output of the amplifier 30 becomes L. When this is input to the AND circuit 33, the magnitude relationship of the light pattern on the lens 1a side is determined on the sensors 20b and 21b. When the AND circuit 32 becomes equal on the sensors 21b and 22b, the AND circuit 33 outputs an H signal.
With the above-described configuration, it is possible to determine where the change in the light / dark pattern occurs in the sensor array by using a logic circuit. A simple correlation operation can be performed by making such a circuit complicated as hardware.
[0028]
Here, FIG. 6 shows a configuration of a camera as a specific example of the embodiment of the present invention. However, since the ranging unit 100 including the lenses 1a and 1b and the sensor arrays 2a and 2b has already been described, detailed description thereof will be omitted. The correlation circuit 5 is composed of a hard logic circuit as described with reference to FIG.
The A / D conversion circuit 4 is a circuit that converts the analog output of the sensor array into digital data. Here, the A / D conversion circuit 4 also has a function of storing the converted data. The output circuit 7 outputs this stored data to the CPU 10 by serial communication. The CPU 10 includes a port means 19 including a serial communication port, a terminal for controlling each function of the camera, and the like, a register group 14 and 18 for controlling the function, a calculation unit 17, a RAM 15 for temporarily storing data, It is composed of a ROM 16 or the like containing a program of instructions for controlling these with a predetermined algorithm.
[0029]
Further, by the port control, the feeding control of the focusing lens 13 is performed via the feeding control means, and the shutter 8 and the like are driven and controlled. An electrically writable ROM (EEPROM) 9 is provided to store correction coefficients for correcting errors caused by variations in camera parts and variations in assembly accuracy. The CPU 10 refers to the correction data while referring to the correction data. Control the camera.
[0030]
Next, the operation of the camera distance measuring apparatus configured as described above will be described with reference to the flowchart of FIG.
First, a correlation area is designated (S1). Here, the A / D conversion means 4 performs A / D conversion on the data based on the light pattern incident on the sensor array by designating correlation areas as shown by R1 to R3 in FIG.
Correlation is calculated for each sensor pitch by the correlation means 5 in FIG. 6 (S2), and this correlation result S is obtained. Since this value S is a value that depends on the distance of the subject, a rough determination of the subject distance can be made by making a determination based on this value (S3). For example, if it is larger than the predetermined value S1, the subject is considered to be in "short distance", branching to step S16, performing interpolation calculation in the "wide" area (S16), and proceeding to step S11.
[0031]
On the other hand, if the value is equal to or smaller than the predetermined value S1 in step S3, the process proceeds to step S4 to perform interpolation calculation in the “narrow” correlation area. Specifically, first, data (R0 to R4) for measuring a distance in a predetermined narrow area is read from the R side sensor (S4).
Next, reading, which is one of the features of the present invention, is performed. In other words, the data on the L side is not read indiscriminately, but only the data (L1, L2, L3) used in each calculation formula shown in [Formula 3] is input (S5). In this method, since the data is a value that takes into account the correlated shift amount S, the RAM of the microcomputer (CPU) is not wasted more than necessary, and the time required for reading data by serial communication is shortened. it can.
From the data thus read, a calculation based on each equation shown in [Equation 2] is performed to obtain the minimum Fmin and F-1 and F + 1 on both sides (S6 to S8).
[0032]
Next, it is confirmed whether or not there is a difference in the correlation value S by switching the area range between the correlation calculation and the interpolation calculation (S9). In this determination, it is determined whether or not Fmin has a minimum value among Fmin, F-1, and F + 1 obtained by the correlation calculation in steps S6 to S8.
If Fmin is not minimum, the process further branches to step S13 to determine the magnitude relationship between F-1 and F + 1. If F-1 is minimum, S is decreased by one. If it is minimum, the correlation value S is increased by one and S is corrected (S14, S15), and then the process returns to step S3.
[0033]
Subsequently, steps S10 to S12 are steps for performing an interpolation operation (S10) based on the equation shown in the figure, calculating a distance L (S11), and focusing on a subject at the distance L (S12). Note that the above-described calculation of x in step S16 is basically performed in the same manner as in steps S6 to S10 only by obtaining the sensor area over a wide range. However, the processing step S9 for confirming the magnitude of Fmin again is unnecessary.
[0034]
Usually, a one-chip microcomputer uses a calculation register 14 to calculate data and puts the data in and out according to a predetermined order. Therefore, when performing a complicated calculation using a lot of data, It takes a lot of time and instructions to calculate data by putting it in and out one by one. Therefore, the “FF operation” as shown in FIG. 2C, in which the difference between the R and L data is taken while shifting the data little by little and added over a predetermined area, is a very inefficient process. Become. In addition, since the subsequent interpolation calculation uses only Fmin and F-1 and F + 1 on both sides as shown in [Equation 4], even if many FFs are calculated, high-precision measurement is possible. I can't get away. Although the load is heavy for the capacity of the ROM 16 provided as shown in FIG. 6, it is a big problem from the viewpoint of the RAM capacity for temporarily storing data.
[0035]
Therefore, in contrast to the RAM usage pattern of the conventional distance measuring device shown in FIG. 8A, FIG. 8B illustrates the RAM usage pattern of the present embodiment. As shown in the figure, R0 to R3, L1 + S, L2 + S, and L3 in the present invention, compared with the prior art where much of R0 to R4, L0 to L4 + S,. It can be understood that only the calculation of + S is sufficient and the memory space can be saved, so that the use space of the RAM and the like is greatly reduced.
[0036]
(Operation effect 1)
As described above, according to the present embodiment, the distance measuring device has high speed and high accuracy, that is, the influence of mixing of perspectives is small, and also has the effect of reducing the program in the CPU and the memory space used. Can be provided.
[0037]
(Second Embodiment)
Next, a second embodiment according to the present invention will be described. The distance measuring device of the present embodiment is also implemented in the device configured as shown in FIG. The feature of the present embodiment is one embodiment for dealing with an offset error between the left and right R and L sensor arrays as shown in FIG. 8, and the flowchart of FIG. 9 shows the feature of the present embodiment. The details of typical arithmetic processing are exemplified as described later.
[0038]
FIG. 11 is a graph showing the output of each sensor corresponding to FIG. As illustrated, assuming that L is a reference and R is assumed to have errors of C11, C22, and C33 on each sensor, it is preferable to perform correlation calculation and interpolation calculation after correcting these. However, when this error is small, the correlation calculation with the sensor pitch as “resolution” can be performed accurately. Therefore, in the second embodiment of the present invention, this correction is performed prior to the interpolation calculation, and the interpolation is accurate and the number of correction coefficients is minimized, so that the capacity of the EEPROM for storing the correction can be reduced. . Further, since such an error tends to affect when the distance measurement area is small, the correction coefficient is further reduced by performing the above correction only when the interpolation calculation is performed in a narrow area.
[0039]
Further, as shown in the graph showing the probability of the distance of the person of the main subject shown in FIG. 10A, when a photograph taken in general is analyzed, a person is less likely to be a subject at a distance L1 or more. On the other hand, as shown in FIG. 4, it is not particularly necessary to narrow the area beyond a certain distance L2. Therefore, it is possible to design that the sensor error should be corrected only when the subject located at a distance of L1 to L2 is measured. This rough distance determination can be predicted to some extent by the correlation calculation result S. Therefore, the flowchart shown in FIG. 9 is a specific example designed on the assumption that the correlation value S of the subject at a distance of L1 to L2 takes a value between predetermined S2 and S1.
[0040]
Steps S1 to S5 in FIG. 9 are the same as the flowchart of FIG. 7 of the first embodiment already described. However, for the reason described above, in step S3, control is performed so as to perform distance measurement in a narrow area only when the correlation value S is between S1 and S2 (that is, S2 <S <S1).
Following the reading (S5) of the L side sensor data (LS + 1 to LS + 3) following the reading of the R side sensor data (R0 to R4) (S4), in step S5b, L1 and L2 in FIG. , L3 reference R1, R2, R3 errors C11, C22, C33 and other correction coefficients are read out from the EEPROM.
In steps S6 to S8, errors between Rn, Rn-1, Rn + 1 and Ln (errors which do not result in the same output when the same luminance is given) Cnn, Cn (n-1), In this step, Fmin, F-1, and F + 1 are calculated while correcting Cn (n + 1).
From Fmin, F-1, and F + 1 obtained in this way, x is obtained by interpolation as in step S10 of FIG. 7, and focusing is performed (S12).
[0041]
(Operation effect 2)
With the configuration as described above, the memory capacity of the correction coefficient conventionally required for each of the R and L sensors as shown in FIG. 10B is the same as that shown in FIG. As shown, the memory capacity can be reduced. Therefore, according to this embodiment, the subject distance can be determined using the correlation result S obtained in advance at high speed, and error correction can be performed with emphasis on a person at a predetermined long distance. This makes it possible to reduce the correction program and memory.
[0042]
(Third embodiment)
Next, a third embodiment according to the present invention will be described. The distance measuring device of this embodiment is also implemented in the device configured as shown in FIG. As shown in FIGS. 12 (a) and 12 (b), the feature of this embodiment is that a state in which a part of the distance measurement area deviates from the person 6 is detected. In this case, the distance measurement is performed except for the deviated sensor. FIG. 5 is an embodiment for dealing with the mixed perspective error described with reference to FIG. 4 by performing calculations. Compared to the first and second embodiments described above, the distance measurement area is not narrowed unnecessarily, so it is more accurate for subjects that are larger than human faces even at long distances. This is an embodiment in which the distance measurement can be performed.
In the shooting scene as shown in FIG. 12B, a sensor that is out of the subject has a poor correlation. Therefore, the result of the difference | Rn−Ln + s | corresponds to this subject as shown in the graph of FIG. The value is different compared to the sensor. Utilizing such a principle, in the example of FIG. 12A, the sensor output exceeding the difference Fc between the central sensors R2 and Ls + 2 is removed and interpolation calculation is performed. In order to have such a function, a program having the algorithm of the flowchart shown in FIG. 13 is built in the CPU.
[0043]
Steps S1 to S5 are the same processing steps as those already described. Subsequently, Fmin, F-1, F + 1, etc. for interpolation calculation are initialized to 0 (S20), 2 is substituted for i to select the center sensor (S21), and Ri or R2 Data is read (S22).
The difference Fi is taken (S23), and the obtained difference is set as a difference reference value Fc (S24). Since the sum of the differences for each sensor is Fmin, this is added to Fi (S25).
In subsequent calculation step groups, Ri + 1 and Ri-1 are read in order to obtain F + 1 and F-1, the difference from Li is obtained, and F + 1 and F-1 are calculated (S26-). S29).
[0044]
Next, in steps S30 to S34, first, i is subtracted by 1 (S30), then Ri, Ri + 1, Ri-1 are read for the younger number on the left side of the center (S31), and the difference from Li is calculated. Then, F + 1 and F-1 are calculated respectively (S32). Here, the magnitude relationship with the reference value Fc is determined so that the difference Fi between Ri and Li greatly deviates from the reference value Fc is removed from the target of the calculation process (S34), and the difference is not taken into account. (S33).
[0045]
Similarly, in steps S35 to S40, Ri, Ri + 1, Ri-1 are read (S37) for the sensor i which is incremented by one from the center on the right (S35, S36), and the difference from Li is obtained. , Fmin, F + 1, and F-1 are calculated (S38). Similarly, if the difference Fi between Ri and Li deviates significantly from the reference value Fc, it is removed from the calculation process (S40). In this manner, the magnitude relationship with the reference value Fc is determined (S39).
Interpolation is performed by using Fmin, F-1, and F + 1 for interpolation calculation obtained in this way to obtain x (S10), and then the distance L is obtained and focused (S12). A series of processing ends.
[0046]
(Operation effect 3)
As described above, in the third embodiment, measures are taken for mixed perspective errors, and in the case of a subject other than a person, the distance measurement area can be widened, and the optimum distance measurement area for the subject can be selected for higher accuracy. It is possible to focus on.
[0047]
(Other embodiments)
In the first to third embodiments exemplified above, the case where a hard logic circuit is used as the means for obtaining the correlation amount S in units of sensor pitch has been described as an example, but this means is calculated by a microcomputer using a predetermined program. Even software means can be applied. Therefore, a specific example in this case is shown in a flowchart in FIG. This processing procedure is a process for obtaining the correlation amount S. The minimum value of the FF described in FIG. 2 is represented by “FF1”, and the shift amount on the horizontal axis of the graph is represented by “SFT”. N is an integer representing the sensor number.
[0048]
First, FF1 is initialized to 100 (S51), and SFT, n, and FF are initialized to 0 (S52).
The subsequent steps S53 to S56 are a difference calculation processing loop. That is, the difference FF1 is taken (S53), and the obtained difference is added to the FF (S54). After incrementing the sensor number n by 1 (S55), the repetition of this processing loop is determined by comparison with a predetermined limit number nc (that is, the number of target areas) (S56).
[0049]
When the sum of the differences for each sensor is obtained as FF by the above processing loop, next, FF and FF1 are compared in size (S57). If FF1 is larger, FF is substituted into FF1 (S58). , SFT is substituted into S (S59).
After incrementing SFT by one (S60), the process returns to step S53 and repeats the above steps until a predetermined limit value Smax is reached. When Smax is reached, the process proceeds to step S3 in FIG.
[0050]
(Operation effect 4)
As described above, the means for obtaining the correlation amount can also be realized by software means by the program composed of steps S51 to S61. As is apparent from this flowchart, if the target area number nc for adding the correlation is small, the number of times the process loop is repeated is reduced, and the speed can be increased. Further, since the correct S value can be obtained even if the correlation calculation in pitch units is somewhat rough, higher-speed processing is possible than calculation processing without area switching as shown in FIG.
[0051]
(Other variations)
In addition, the present invention can be variously modified without departing from the above gist as an implementation method.
[0052]
As mentioned above, although demonstrated based on embodiment, the following invention is contained in this specification.
[1] Light receiving means for receiving a subject image by a pair of light receiving element rows having parallax and outputting a pair of subject image signal rows;
Correlation means for receiving the output of the light receiving means and outputting a correlation value of a first portion of the pair of subject image signal sequences;
Only the partial data of the second portion narrower than the first portion of all the subject image signal rows of one of the pair of light receiving element rows and the subject image signal row of the other light receiving element row. A data input means for selectively inputting
Interpolation means for performing a predetermined interpolation calculation based on the input data;
A distance measuring device comprising:
[2] Light receiving means for receiving a subject image by a pair of light receiving element rows having parallax and outputting a pair of subject image signal rows;
Means for dividing the pair of subject image signal sequences into a plurality of blocks having a first number of pixels;
Correlation means for receiving an output of the light receiving means and outputting a correlation value of the pair of subject image signal sequences;
Means for resetting the block to a second number of pixels greater than the first number of pixels if the correlation value satisfies a predetermined condition;
Only the data of the second number of pixels is selectively input from all the subject image signal rows of one of the pair of light receiving element rows and the subject image signal row of the other light receiving element row. Data input means,
Interpolation means for performing a predetermined interpolation calculation based on the input data;
A distance measuring device comprising:
[3] Two line sensors that output two light pattern signals according to the luminance distribution of the subject image observed from different fields of view having parallax;
A pitch difference calculating means for comparing the two light pattern signals and calculating a relative position difference based on a parallax of the subject image in units of an arrangement pitch of the line sensors;
An arithmetic control unit that selects an output signal of the line sensor based on an output result of the pitch difference calculating unit, and calculates a relative position difference between the two light patterns based on the result with a finer accuracy than the sensor pitch. When,
Comprising
The distance measuring device, wherein the pitch difference calculating means is constituted by a hard logic circuit, and the arithmetic control means is constituted by a microcomputer and software.
The following invention is also included.
(1) two line sensors that output two light pattern signals according to the luminance distribution of the subject image observed from different fields of view having parallax;
A pitch difference calculating means for comparing the two light pattern signals and calculating a relative position difference based on a parallax of the subject image in units of an arrangement pitch of the line sensors;
Calculation control means for selecting the output signal of the line sensor based on the output result of the pitch difference calculation means and calculating the relative position difference between the two light patterns based on the result with a finer accuracy than the sensor pitch. When,
A distance measuring device comprising:
(2) The pitch difference calculating means includes a hard logic circuit,
The distance measuring apparatus according to (1), wherein the arithmetic control unit is configured by a microcomputer and software.
(3) The arithmetic control means includes a storage means for storing the line sensor output, and selects one output of the two line sensors based on the output result of the pitch difference calculation means. The distance measuring apparatus according to (1), wherein a control is performed so that a difference from one output is calculated and the result is stored in the storage means.
(4) In the process of calculating the relative position difference between the two light patterns with an accuracy finer than the sensor pitch, the arithmetic control unit compares the difference between the two light patterns, It has a function to judge the output result,
The distance measuring apparatus according to (1), wherein when the output result of the pitch difference calculating means is determined to be incorrect, the output result is changed.
[0053]
(5) Two line sensors that output two light pattern signals according to the accuracy distribution of the subject image observed from different fields of view having parallax, and a relative position based on the parallax of the subject image by comparing the two light pattern signals A pitch difference calculation unit that calculates the difference in units of the arrangement pitch of the line sensors, and a relative position difference between the two light patterns is calculated with a finer accuracy than the pitch based on an output result of the pitch difference calculation unit. A distance measuring device comprising arithmetic control means,
The arithmetic control means selects and outputs the output signals of the two line sensors, and determines a sensor portion for calculating the relative position difference based on the subtraction result. .
(6) comprising means for storing output unbalance correction information of the two line sensors;
The distance measuring apparatus according to (5), further comprising a correcting unit that selects the output imbalance correction information based on the calculated pitch difference and corrects the subtraction result.
(7) two line sensors that output two light pattern signals based on the luminance distribution of the subject image observed from different fields of view having parallax;
A pitch difference calculating means for comparing the first portions of the two light pattern signals and calculating a relative position difference based on the parallax of the subject image in units of a line sensor;
An arithmetic control unit that calculates a relative position difference between the two light patterns according to an output result of the pitch difference calculating unit with an accuracy finer than a pitch of the line sensor;
When calculating the relative position difference with a finer accuracy than the pitch, the arithmetic control unit determines a second portion different from the first portion of the two line sensors based on the pitch difference calculation result. A distance measuring device that selects and compares the output of this portion to calculate the relative position difference.
(8) means for storing output imbalance correction information of the two line sensors;
Correction means for determining whether or not to correct the subtraction result based on the output of the selected second portion for position difference calculation and the output imbalance correction information;
The distance measuring apparatus according to (7), further comprising:
[0054]
In addition, the following inventions are also included.
(1 ′) two line sensors that output two light pattern signals according to the luminance distribution of the subject image observed from different fields of view having parallax;
A pitch difference calculating means for comparing the two light pattern signals and calculating a relative position difference based on the parallax of the subject image in units of the pitch of the line sensor;
According to the output result of the pitch difference calculating means, the output signal of the line sensor is selected, and the arithmetic control means for calculating the relative position difference between the two light patterns with a finer precision than the pitch of the line sensor according to the result;
A distance measuring device comprising:
(2 ′) The distance measuring device according to (1 ′), wherein the pitch difference calculating means is constituted by a hard logic electronic circuit, and the arithmetic control means is constituted by a microcomputer and software. .
(3 ′) The arithmetic control means includes storage means for storing the output of the line sensor, selects one output of the two line sensors based on the output result of the pitch difference calculation means, and The distance measuring device according to (1 ′), wherein a control is performed so as to calculate a difference from the output of the first output and store the result in the storage means.
(4 ′) In the process of calculating the relative position difference between the two light patterns with an accuracy finer than the pitch of the line sensor,
A function of comparing the difference between the two light patterns to determine an output result of the pitch difference calculating means; and if the output result of the pitch difference calculating means is determined to be incorrect, the pitch difference calculating result The distance measuring device according to (1 ′), wherein the distance measuring device is changed.
[0055]
(5 ′) two line sensors that output two light pattern signals according to the accuracy distribution of the subject image observed from different fields of view with parallax;
A pitch difference calculating means for comparing the two light pattern signals and calculating a relative position difference based on the parallax of the subject image in units of the arrangement pitch of the sensors;
In a distance measuring apparatus comprising: an arithmetic control unit that calculates a relative position difference between the two light patterns based on an output result of the pitch difference calculation unit with an accuracy finer than the pitch.
The arithmetic control means selects the output signals of the two line sensors, performs a predetermined subtraction, and determines a sensor portion for calculating the relative position based on the subtraction result. apparatus.
(6 ′) A means for storing output unbalance correction information of the two line sensors is provided, and the output unbalance correction information is selected based on the calculated pitch difference to correct the subtraction result. The distance measuring device according to (5 ′), further comprising a correcting means.
[0056]
(7 ′) two line sensors that output two light pattern signals according to the luminance distribution of the subject image observed from different fields of view having parallax;
A pitch difference calculating means for comparing the first portions of the two light pattern signals and calculating a relative position difference based on the parallax of the subject image in units of the pitch of the line sensor;
In the distance measuring apparatus configured with calculation control means for calculating the relative position difference between the two light patterns with an accuracy finer than the sensor pitch based on the output result of the pitch difference calculation means,
When the relative position difference is calculated with an accuracy finer than the pitch of the line sensor, the arithmetic control unit is different from the first part of the two line sensors based on the pitch difference calculation result. A distance measuring device, wherein a second part is selected, and the position difference is calculated by comparing outputs of the two parts.
(8 ′) Further comprising means for storing output unbalance correction information of the two line sensors, and using the output unbalance correction information based on the selected second portion for calculating the position difference, The distance measuring apparatus according to (7 ′), further comprising a correcting unit that determines whether or not to correct the subtraction result.
[0057]
【The invention's effect】
As described above, according to the present invention, it is possible to appropriately switch the distance measurement range of the passive type distance measuring device and perform high-speed calculation processing without measuring an unnecessary area of the subject. A distance measuring device can be provided, and by this application, it is possible to provide a photographing device such as a camera or a video that can perform accurate focusing at high speed.
In addition, as an electronic system capable of high-speed and high-precision distance measurement, it is possible to provide a distance measurement device for a camera or the like that can be configured even with a small memory capacity and a small storage capacity of a program.
[Brief description of the drawings]
FIG. 1 is a configuration diagram showing a basic structure of a distance measuring apparatus according to an embodiment of the present invention.
2 (a) to 2 (d) show the calculation principle of distance measurement according to the present invention;
(A) is a graph showing the relationship between the focus and the shift amount;
(B) is a graph showing the signal output for each sensor portion of the right sensor R;
(C) is a graph showing the signal output for each sensor portion of the left sensor L,
(D) is a graph showing the relationship between the left and right sensor outputs and the outputs of adjacent sensor portions.
FIGS. 3 (a) to 3 (c) show a procedure consisting of correlation calculation and interpolation calculation;
(A) is a flowchart showing a processing procedure by hardware;
(B) is a flowchart showing a processing procedure by software;
(C) is a flowchart showing a processing procedure by a combination of hardware and software.
FIG. 4 is a conceptual diagram showing a positional relationship between a line sensor and a subject.
FIG. 5 is a circuit diagram showing a hard logic circuit as correlation means of the distance measuring device.
FIG. 6 is a block diagram showing in detail the structure of the distance measuring device of the present invention.
FIG. 7 is a flowchart showing a correlation calculation processing procedure according to the first embodiment of the distance measuring apparatus;
8 (a) and 8 (b) show a comparison of calculation areas used in the prior art and the present invention;
(A) is a conceptual diagram showing an area where the conventional method is used;
(B) is a conceptual diagram which shows the area used with this distance measuring device.
FIG. 9 is a flowchart showing a correlation calculation processing procedure according to the second embodiment of the distance measuring apparatus;
10 (a) to 10 (c) show a graph relating to a subject and an area for storing errors (correction values) of the conventional and the left and right sensors of the present invention in comparison with each other;
(A) is a graph showing the probability when the subject is a person,
(B) is a conceptual diagram showing an area of correction coefficients used in the conventional method;
(C) is a conceptual diagram which shows the area used with this distance measuring device.
FIG. 11 is a graph showing the output of each sensor corresponding to FIG.
FIGS. 12A and 12B show the outputs of the reference right sensor and the left and right sensors,
(A) is a graph showing the output and difference of the left and right sensors,
(B) is explanatory drawing which shows the positional relationship between the sensor unit of the right sensor used as a reference, and the subject.
FIG. 13 is a flowchart showing a processing procedure of interpolation calculation as a third embodiment of the distance measuring apparatus;
FIG. 14 is a flowchart illustrating a correlation calculation processing procedure according to another embodiment of the distance measuring apparatus;
FIG. 15 is a diagram illustrating [Equation 1] representing an output of each left sensor portion when a deviation amount of light is incident on the left sensor with the right sensor as a reference;
FIG. 16 is a diagram illustrating [Expression 2] based on a difference between right and left sensors;
FIG. 17 is a diagram illustrating [Expression 3] in which [Expression 2] is further developed.
FIG. 18 is a diagram showing [Expression 4] representing the relationship between the minimum value Fmin obtained by interpolation calculation and the SIFT amount;
[Explanation of symbols]
1a, 1b ... lenses (for sensors),
2a, 2b ... line sensors,
3a, 3b ... signal output waveform,
4 A / D converter,
5 ... operation part (correlation means),
6 ... Subject,
7 ... Output section,
8 ... Shutter
9… EEPROM,
10: Control means (CPU),
12 ... Feeding means,
13 ... Zoom lens,
14: Register for calculation,
15 ... RAM,
16 ... ROM,
17 ... calculation means,
18: Register (general purpose),
19 ... Port.
S1 to S16 ... processing steps of the first and second embodiments of the present invention,
S1 to S40: processing steps of the third embodiment of the present invention,
S51 to S61 ... processing steps of other embodiments of the present invention,
S101 to S104 ... hardware processing steps,
S105 to S108: Calculation processing steps by software,
S109 to S113: Arithmetic processing steps by hardware and software.

Claims (3)

Light receiving means for receiving a subject image by a pair of light receiving element rows having parallax and outputting a pair of subject image signal rows;
Correlation means for receiving the output of the light receiving means and outputting a correlation value of a first portion of the pair of subject image signal sequences;
Partial data of a second portion narrower than the first portion of all the subject image signal rows of one of the pair of light receiving element rows and the subject image signal row of the other light receiving element row Data input means for selectively inputting only,
Interpolation means for performing a predetermined interpolation calculation based on the input data;
A distance measuring device comprising: Light receiving means for receiving a subject image by a pair of light receiving element rows having parallax and outputting a pair of subject image signal rows;
Means for dividing the pair of subject image signal sequences into a plurality of blocks having a first number of pixels;
Correlation means for receiving an output of the light receiving means and outputting a correlation value of the pair of subject image signal sequences;
Means for resetting the block to a second number of pixels smaller than the first number of pixels if the correlation value satisfies a predetermined condition;
Only the data of the second number of pixels is selectively input from all the subject image signal rows of one of the pair of light receiving element rows and the subject image signal row of the other light receiving element row. Data input means,
Interpolation means for performing a predetermined interpolation calculation based on the input data;
A distance measuring device comprising: Two line sensors that output two light pattern signals according to the luminance distribution of the subject image observed from different fields of view with parallax;
A pitch difference calculating means for comparing the two light pattern signals and calculating a relative position difference based on a parallax of the subject image in units of an arrangement pitch of the line sensors;
Based on the output result of the pitch difference calculating means, a part of the light pattern signal output from the line sensor used for comparison in the pitch difference calculating means is selected, and the subject is selected based on the selected light pattern signal. Calculation control means for calculating a relative position difference based on parallax of the image with an accuracy finer than an arrangement pitch of the line sensors;
Comprising
The distance measuring device, wherein the pitch difference calculating means is constituted by a hard logic circuit, and the arithmetic control means is constituted by a microcomputer and software.

Images