JP3938989B2 - Ranging device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a distance measuring device.
[0002]
[Prior art]
There is a so-called passive distance measuring device that obtains two image signals from the luminance distribution of an object and obtains the distance to the object according to the difference in position based on the parallax. The two image signals are based on light beams that have passed through different optical paths, and the image position changes according to the distance to the object by the principle of triangulation.
[0003]
In such a system, the image signal is likely to change depending on the state of the background, and it is necessary to measure the distance of the object image separately from the background in order to increase the accuracy of distance measurement. Also, in a backlight scene, the image becomes unbalanced and causes accuracy degradation.
[0004]
Therefore, when the above-described method is applied to an apparatus that has to perform stable distance measurement on subjects in various scenes, such as an autofocus camera, various devices are required. For example, Japanese Patent Application Laid-Open No. 62-133410 discloses switching an image signal to be used according to a screen.
[0005]
Further, the image signal is obtained by integrating the output current of the sensor array. However, in a scene where the luminance difference is large as in the case of backlighting, a technique for keeping the integration level within a predetermined dynamic range is also important. For example, Japanese Patent Laid-Open No. 5-264887 discloses increasing the integration time during backlighting. Japanese Patent Laid-Open No. 5-264892 discloses a method of overcoming accuracy degradation due to non-uniformity of light rays incident on two sensor arrays due to flare.
[0006]
[Problems to be solved by the invention]
Japanese Patent Application Laid-Open No. 62-133410 and Japanese Patent Application Laid-Open No. 5-264892 do not disclose any countermeasures against the problem of backlighting. Japanese Laid-Open Patent Publication No. 5-264887 discloses measures against backlighting, but there is a problem that the integration time becomes long and a time lag occurs.
[0007]
The present invention has been made paying attention to such a problem, and the object of the present invention is to perform accurate distance detection without a time lag even in backlit conditions, and in addition, overcome the problem of flare. Accordingly, it is an object of the present invention to provide a distance measuring device with further improved distance measuring accuracy.
[0008]
Another object of the present invention is to provide a distance measuring device that can improve a focusing rate even in a scene where focusing cannot be performed when applied to an AF camera.
[0009]
[Means for Solving the Problems]
In order to achieve the above object, the first invention provides a first sensor array that outputs a luminance distribution of a subject as a first image signal, and a second sensor that outputs the luminance distribution of the subject as a second image signal. A sensor array, a determination means for determining whether or not the backlight state is based on the first and second image signals, and integration regions of the first and second sensor arrays when the backlight state is determined. An area switching means for switching so as to be larger than that in the non-backlight state, and an area where the difference between the main subject brightness and the background brightness is large from the first and second image signals output from the switched area. The distance of the subject is calculated based on the edge signal output means for extracting and outputting the luminance difference of this area as an edge signal corresponding to at least one sensor array and the image signal from the area specified by the edge signal. Operator Comprising the door.
[0010]
In a second aspect based on the first aspect, the arithmetic means removes a flare signal from one of the two image signals to create a correction signal. The subject distance is calculated based on the other of the two image signals.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. First, the first embodiment will be described. First, an external light passive distance measuring device will be described with reference to FIGS. 1 (a), (b), and (c). Reference numerals 1a and 1b denote light receiving lenses that guide the luminance distribution of the object 3a to the sensor arrays 2a and 2b. These sensor arrays 2a and 2b convert the change in brightness of the object 3a into a current signal. The A / D conversion means 6 receives each output signal from the sensor arrays 2a and 2b and converts it into a digital value, thereby inputting an image signal to a digital arithmetic control circuit (hereinafter referred to as CPU) 10. Reference numeral 7 denotes a focusing means. Reference numeral 5 denotes a monitor / integration control circuit described later.
[0013]
Here, assuming that the distance between the light receiving lenses is the base line length B and the focal length is f, the image of the object 3a is placed on the two sensor arrays 2a and 2b as indicated by 14a and 14b in FIG. , X with a relative positional difference. Here, when the distance to the object 3a is L, these relations are x = (B · f) / L (1).
It becomes. That is, the larger the distance L, the smaller the relative position difference x, and the smaller the distance L, the larger the relative position difference x. When applied to a camera or the like, a one-chip microcomputer or the like may be used as the CPU 10, which obtains the relative position difference x between the images 14a and 14b, obtains the distance L from equation (1), and adjusts the focusing means. If the photographing lens is controlled in step 7, an AF camera can be designed.
[0014]
In addition to the above functions, the CPU 10 includes a backlight determination unit 8 that determines whether or not the object is in a backlight condition, an edge detection unit 9 that detects an edge portion of the obtained image signal, and the like. Yes.
[0015]
Under the condition of backlight, strong noise light is mixed into the sensor arrays 2a and 2b from the background of the object 3a, so that the image of the object 3a does not have a correct shape and the distance measurement tends to be inaccurate. . Therefore, in the present embodiment, it is determined whether or not the backlight is in a backlight condition, and when the backlight is in the backlight, the edge portion is extracted from the obtained image signal and is based on the image signal corresponding to the extracted region. Accurate distance measurement is performed by performing distance measurement. In addition, the signal component based on the mixed noise light is removed to further improve the ranging accuracy.
[0016]
Further, as shown in FIG. 1C, a plurality of points 15 in the camera screen 13 are measured, and the person 12 is correctly focused even if the person 12 as a distance measuring object does not exist in the center of the screen. There is a technique called multi-AF that is devised to fit. For example, when measuring the object 3a in FIG. 1A using this technique, the sensor on the optical axis of the sensor array 2a is used as a reference, and the sensor slightly shifted to the left side from this sensor position is used as a reference. By measuring the distance, the distance to the object 3a can be obtained. Ranging can be basically performed in the same way as the one-point AF except that the reference sensor position is changed. The sensor switching means 4 switches the reference sensor position. The same principle can be used when the edge portion is extracted from the image signal and distance measurement is performed.
[0017]
FIG. 2 shows the relationship between the person 12, the light receiving lens 1a, and the sensor 2a under backlight conditions.
Ideally, it is assumed that only light rays based on the shadow of the face of the person 12 enter the sensor array 2a. However, depending on the position of the sun, strong light 16 reflected from a part of the person reaches the sensor array 2a, and strong light is incident only on one of the two sensor arrays 2a and 2b. The image signal becomes unbalanced, or the sun light 17 is directly incident on the wall surface of the holding unit 11 holding the light receiving lens 1a and the sensor array 2a, and these noise lights are mixed on the original image signal. As a result, the contrast is lowered, and an image signal different from the expected one may be obtained.
[0018]
In a scene as shown in FIG. 2, the difference between the image signals obtained at the time of direct light and back light is shown in FIGS. 3 (a) and 3 (b). Assuming that the incident light is stronger, the output of the sensor array 2a is smaller, and as the incident light is weaker, the output of the sensor array 2a is larger. As shown in FIG. The contrast obtained with the width of ΔI1 on the face of the person 12 is also reduced to the width of ΔI2, as shown in FIG. Changes in the background due to forward light and backlight appear as changes at both ends of the image signal (changes in area). The change in contrast is caused by the fact that noise light overlaps the original image signal as described above, or that the luminance difference from the background becomes too large.
[0019]
FIG. 4 is a diagram showing a configuration of a processing circuit of the sensor array for obtaining the above image signal. The operation of this processing circuit will be described below.
20a and 20b are obtained by taking out two of the sensors constituting the sensor array, each of which is composed of a photodiode, and the outputs thereof are connected to integrating amplifiers 21a and 21b. Switches 22a, 22b, 24a, and 24b are connected to the integrating capacitors 23a and 23b. Among these, the switches 22a and 22b are controlled by the integration control circuit 28, and the integration start and end are determined by this control. The switches 24a and 24b are controlled by the integration reset circuit 25, and a reset operation is performed prior to starting integration again. The CPU 10 causes the above control to be performed sequentially.
[0020]
The integral outputs VINTa and VINTb are determined from the capacitances of the integrating capacitors 23a and 23b, the intensity of the incident light, and the integration time. If the integration control circuit 28 does not set an appropriate integration time at this time, VINTa, b cannot be stored in the dynamic range. Therefore, the monitor / integration control circuit 5 that monitors the integral outputs VINTa and VINTb via the output circuit 26 and performs integral control is required. Here, it is possible to select which sensor output is to be monitored from among the sensors constituting the sensor array. The sensor switching means 4 performs this selection. Therefore, if the control is performed by the CPU 10 according to the situation, it is possible to switch which sensor in the sensor array is focused on the integral control.
[0021]
For example, the image signal shown in FIG. 3 is obtained by switching the sensor for integration monitoring in the forward light state and the backlight state. Since FIG. 3 (a) shows a time of direct light, a contrast is obtained with the shadow of the face of the person 12 and normal distance measurement is possible. Therefore, integration control is performed in a narrow area so as not to be affected by the background as much as possible. However, since the contrast of the face is low in the backlight scene, the contrast is obtained by enhancing the luminance difference from the background as shown in FIG. In other words, in this case, the integration control area is made wider than the width of the face, and the integration control is performed in consideration of the background.
[0022]
FIG. 5A shows an example of an image signal when the integration control area (monitor range) is kept narrow in a backlight scene. The horizontal axis represents the order of the sensors in the sensor array (sensor No.), and the vertical axis represents the result of converting the sensor data into digital values by the A / D conversion means 6. In this example, since the integration is stopped using the integration result of the brightest part in the monitor range, the bright background exceeds the dynamic range and no data is output. Here, when the integration control area (monitor range) is widened, integration control is performed in consideration of the output of sensor data that receives light from the background as shown in FIG. An image signal is formed. Therefore, although the dynamic range of the data at the center of the face is deteriorated, an image signal having a high contrast can be obtained due to the luminance difference between the background and the person.
[0023]
As described above, in this embodiment, the image signal of only the face portion is less affected by the background, but in order to eliminate the disadvantage that the contrast is too low to obtain only a low reliability result, the integration control area Is used to obtain an image signal having a sufficiently large contrast, and to correctly measure a subject in a backlight condition from the image signal.
[0024]
FIG. 6 is a flowchart showing the operation when the method of the present embodiment is applied to a camera.
Step S1 is a step of performing integration control with the central portion in the screen as the integration control area. Next, the sensor data of this integration control area is read (step S2), and the peripheral part and the center part of the integration control area are compared to determine whether or not the backlight is in the backlit state (step S3). When it is not backlit, the process branches to step S11, a known multi-AF operation is performed, and sensor switching is repeated to measure a plurality of points on the screen, and a main subject distance Lp is determined from them, and this focus is determined in step S10. Align.
[0025]
On the other hand, if it is determined in step S3 that the backlight is in the backlight state, the process proceeds to step S4, and the integration region is switched to be larger than that in the non-backlight state. That is, the determination area is expanded by increasing the number of integral determination sensors. Next, in step S5, the integration control is performed again, and the obtained image data is read out and input to the CPU 10 after A / D conversion (step S6). As described above, the main subject luminance and its background in step S7. A portion having a high contrast with the luminance of the portion is detected (see FIG. 5B). This detection is referred to as edge detection, and a portion with a large change in image data is obtained by the sensor arrays 2a and 2b in FIG. 1A based on the flow shown in FIG. 7, and is output as an edge signal. The distance L is obtained from the relative position difference x represented by
[0026]
In FIG. 7, in steps S40 to S43, a number n indicating the order of the sensors constituting the sensor array 2a, a number n1 indicating the order of the sensors located at the edge portion, and a variable ΔSS indicating a difference between adjacent sensor data A variable ΔS indicating the maximum value is initialized. Next, in step S44, the difference ΔSS between the data of the nth sensor and the data of the (n + 1) th sensor is taken, and then in step S45, it is compared with the maximum value ΔS. If the determination result in step S44 is yes, the process branches to step S46, stores the sensor number n1 at that time, and stores ΔSS as the maximum value in step S47.
[0027]
Next, the sensor number to be compared is incremented in step S48, and the determination in step S49 is yes, that is, the step in step S44 is repeated until n reaches a predetermined number no. A position n1 of the sensor showing a large change is obtained. By such a flow, the position of the edge portion of the image signal is obtained as a number indicating the order of arrangement of the sensors.
[0028]
By the way, the output of the two sensor arrays tends to be unbalanced in the image data of the edge part due to the influence of background noise light. Therefore, in the flow of FIG. 6, after the edge detection (step S7), the flare correction step S8 is added to remove the influence of the noise light. As shown in FIG. 8, flare correction is a method of obtaining the magnitude of noise light when noise light is superimposed on one of a pair of sensor arrays 2a and 2b for AF, for example, the sensor array 2b. This is a technique for obtaining correct image data by removing by calculation.
[0029]
That is, as shown in FIG. 8, when noise light is evenly distributed, the same number (n2) of sensors is selected from each sensor array, and the noise light is obtained from the difference of the sum of these sensor data. it can. If the sum (Δh) obtained by dividing the sum difference by n 2 is added to each sensor data of the sensor array 2a, the shapes of the images incident on the two sensor arrays 2a and 2b can be made uniform. Δh may be subtracted from the sensor data of the sensor array 2b. In this case, however, the data becomes negative and may cause inconvenience in calculation. The image shape is aligned with the addition of equivalents. If there is no inconvenience in calculation, subtraction may be performed from the side with noise.
[0030]
If the relative position difference of the image shift amount is determined by ignoring the noise light with the image shapes being not uniform, a detection error at the position Δx in FIG. Therefore, in the present embodiment, the CPU 10 performs the flow calculation as shown in FIG. 9 to calculate the distance after aligning the shapes of the two image signals.
[0031]
First, steps S60 and S61 are steps for obtaining the sum of n2 sensor data of portions where the same image is obtained in each sensor array as Σa and Σb. In this case, the sensor data of the portion indicated as n2 in FIG. 8 may be obtained. That is, sensor data of n2 / 2 before and after the edge may be used. Further, the sum may be taken over the entire sensor area. Step S62 is a step of determining which sum is larger. Based on this result, a difference ΔΣ is obtained in steps S63 and S67, and then an average value Δh is obtained to obtain a correction amount Δh (step S64, S68). In steps S65 and S69, this is added to each sensor data of the sensor array with no flare.
[0032]
As described above, when the sum is taken over the entire sensor area, Δh is obtained by dividing this by the total number of sensors. Then, similarly to S65 and S69, this Δh is added to the sensor data on which no flare is placed.
[0033]
By such a process, corrected sensor data is formed in the memory in the CPU 10, and the CPU 10 obtains the image position difference x in FIG. 1 using this correction data, not the raw data of the sensor array output, The main subject distance Lp is obtained from the equation (1) and focused (steps S9 and S10 in FIG. 6). As described above, in the present embodiment, accurate focusing can be performed even when image signals incident on the two sensor arrays become low contrast or unbalanced due to noise light. Therefore, a beautiful photograph can be taken even in a backlight scene.
[0034]
The second embodiment of the present invention will be described below. The sensor array for distance measurement is an element in which several tens to hundreds of light receiving elements (sensors) are juxtaposed. When the above edge detection and flare correction are performed on the output data of these many sensors, the calculation is performed. Takes time, and the release time lag becomes longer. The time for performing the reintegration in step S5 in FIG. 6 also becomes a time lag. Therefore, as shown in the flow of FIG. 10, distance measurement is performed by the first integration, and determination based on distance is also adopted, so that edge detection and flare correction are not performed when the first distance measurement is at low luminance or long distance. In many scenes, the release time lag can be prevented from becoming long.
[0035]
For example, even in a backlight scene as shown in FIG. 11A, when the subject distance is a long distance, the subject image position difference on the sensor array due to the parallax of the two light receiving lenses is also small. The unbalance when the strong light 16 enters the two sensors via the light receiving lens 1a is also reduced, and the subject image is also reduced and enters the sensor array 2a. As shown in FIG. 11B, an image signal with high contrast can be obtained in consideration of the background.
[0036]
For these reasons, in the flow of FIG. 10, distance measurement is performed by the first integration (step S20), and if it is determined that the backlight is backlit in the next step S21, then the first distance measurement result is the short distance. Whether or not (step S22), only at short distance, reintegration (S23, S24), edge detection (step S25), and flare correction (S26) are performed by switching the integral determination range. Less.
[0037]
That is, in this embodiment, when the backlight is close and the subject distance is short as in the scene of FIG. 2, the two sensors that occur because the incident form of the noise light 16 differs due to the parallax of the two sensor arrays 2a and 2b. It overcomes the imbalance between the sensor data in the arrays 2a, 2b, or the low contrast problem caused by noise light overlapping the signal.
[0038]
In other words, when it is determined that the subject is backlit and the subject distance is close, the flow from step S23 is executed to reliably measure the subject distance. In other cases, the reference area is switched in step S29. However, multi-AF is performed. Accordingly, as shown in FIG. 1C, correct focusing can be performed at high speed even in a scene where the subject 12 does not exist in the center of the screen.
[0039]
Here, as an example of the method for determining the backlight, an example in which the integration result of the AF sensor array is monitored is described here. However, as shown in FIGS. It goes without saying that the backlight can be determined by using a photometric result (SPOT) and an average photometric result (AVE) of the entire screen using a known method at the time of exposure control instead of AF.
[0040]
The specific embodiments described above include inventions having the following configurations (1) to (5).
(1)
In a distance measuring device having a plurality of luminance distribution measurement points in an imaging region and outputting one or a plurality of luminance distributions of the luminance distribution measurement points as two image signals by two sensor arrays,
Determining means for determining whether or not the backlight state is based on an image signal from a predetermined luminance distribution measurement point;
When it is determined that the backlight is in the backlight state, an area where the difference between the main subject brightness and the background brightness is large is extracted from the two image signals of the predetermined brightness distribution measurement point, and the brightness difference of this area is determined by at least one sensor. Edge signal output means for outputting as an edge signal corresponding to the array;
When the non-backlight state is determined, the subject distance is calculated based on the outputs from the plurality of luminance distribution measurement points. When the backlight state is determined, the image from the region specified by the edge signal is calculated. Computing means for computing the distance of the subject based on the signal;
A distance measuring device comprising:
(2)
In a distance measuring device having a plurality of luminance distribution measurement points in an imaging region and outputting one or a plurality of luminance distributions of the luminance distribution measurement points as two image signals by two sensor arrays,
Determining means for determining whether or not the backlight state is based on an image signal from a predetermined luminance distribution measurement point;
An area switching means for switching the integration area of the two sensor arrays at the predetermined luminance distribution measurement point so as to be larger than that in the non-backlight state when it is determined that the backlight is in the backlight state;
A region where the difference between the main subject luminance and the background luminance is large is extracted from the two image signals in the integration region of the switched sensor array, and the luminance difference in this region is first and second corresponding to each sensor array. Edge signal output hand throw to be output as the edge signal of 2,
When the non-backlight state is determined, the distance of the subject is calculated based on the outputs from the plurality of luminance distribution measurement points, and when the backlight state is determined, the first and second edge signals are used. Computing means for computing the distance of the subject based on the image signal from the designated area;
A distance measuring device comprising:
(3)
In a distance measuring device that outputs luminance distribution of a subject as two image signals by two sensor arrays,
Determination means for determining whether or not the backlight state is based on the two image signals;
A computing means for computing the distance of the subject;
Determination means for determining whether or not the first subject distance calculated based on the two image signals in a backlight state is closer than a predetermined value;
When it is determined that the subject distance is close to a predetermined value in the backlight state, an area where the difference between the main subject brightness and the background brightness is large is extracted from the two image signals, and the brightness difference between these areas is calculated. Edge signal output means for outputting first and second edge signals corresponding to each sensor array;
Calculation control means for causing the calculation means to calculate a second subject distance based on two image signals from the two areas designated by the first and second edge signals;
A distance measuring device comprising:
(4)
The arithmetic means removes a flare signal from one of the two image signals to create a correction signal, and based on the correction signal and the other image signal of the two image signals. The distance measuring device according to any one of (1) to (3), wherein the distance of the subject is calculated.
(5)
The correction signal is generated by superimposing edge signals of specific portions of the two image signals and then removing a flare-related noise signal from one edge signal. Distance device.
[0041]
【The invention's effect】
According to the present invention, accurate distance detection can be performed even during backlighting, and in addition to this, the flare problem can be overcome and distance measurement accuracy can be further improved.
In addition, when applied to an AF camera, the focusing rate can be improved even in a scene where focusing cannot be performed.
[Brief description of the drawings]
FIG. 1 is a diagram for explaining an external light passive type distance measuring device;
FIG. 2 is a diagram illustrating a relationship between a person, a light receiving lens, and a sensor under backlight conditions.
FIG. 3 is a diagram showing a difference between image signals obtained in a normal light and a backlight in a scene as shown in FIG.
FIG. 4 is a diagram showing a configuration of a processing circuit of a sensor array for obtaining an image signal.
FIG. 5A shows an example of an image signal when the integration control area is kept narrow in a backlight scene, and FIG. 5B is a diagram showing an example of an image signal when the integration control area is enlarged. .
FIG. 6 is a flowchart showing an operation when the method of the first embodiment of the present invention is applied to a camera.
FIG. 7 is a diagram showing a flow of edge detection.
FIG. 8 is a diagram for explaining flare correction.
FIG. 9 is a diagram showing a flow of flare correction.
FIG. 10 is a flowchart showing a method according to a second embodiment of the present invention.
FIG. 11 is a diagram illustrating an example of a backlight scene.
FIG. 12 is a diagram for explaining another method for determining backlight.
[Explanation of symbols]
1a, 1b ... light receiving lens,
2a, 2b ... sensor array,
3a, 3b ... object,
4 ... Sensor switching means,
5 ... Monitor / integration control circuit,
6 ... A / D conversion means,
7: Focusing means,
8 ... Backlight determination means,
9: Edge detection means,
10 ... CPU,
11 ... holding part,
12 ... person,
13 ... screen.

Claims (2)

A first sensor array that outputs a luminance distribution of a subject as a first image signal;
A second sensor array that outputs the luminance distribution of the subject as a second image signal;
Determination means for determining whether or not the backlight is in a backlight state based on the first and second image signals;
A region switching means for switching the integration region of the first and second sensor arrays to be larger than that in the non-backlight state when it is determined to be in the backlight state;
A region having a large difference between the main subject luminance and the background luminance thereof is extracted from the first and second image signals output from the switched region, and the luminance difference in this region corresponds to at least one sensor array. Edge signal output means for outputting as an edge signal;
A computing means for computing the distance of the subject based on the image signal from the region specified by the edge signal;
A distance measuring device comprising:   The arithmetic means removes a flare signal from one of the two image signals to create a correction signal, and based on the correction signal and the other image signal of the two image signals. The distance measuring apparatus according to claim 1, wherein the distance of the subject is calculated.

Images