JP5351095B2 - Focusing device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a focusing device which has further improved sureness of determination of a main object. <P>SOLUTION: The focusing device is caused to perform the following processing, that is, the focusing device detects a subject image signal, determines the head part of a person in the subject image signal, to decide the head part of the person as a focusing point candidate, and detects a person's face direction, in the decided focusing point candidate. Then, the focusing device decides a focusing point where focusing processing is performed based on the detection result and performs focusing processing based on the decided focusing point. At this time, in the detection of the face direction, it is determined that the face of the person is turned to the front or sideways with respect to the focusing device and the person of the face determined as the face turned to the front with respect to the focusing device is decided as the focusing point more preferentially than the person of the face determined as the face turned to the sideways with respect to the focusing device. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a camera autofocus (AF) technique, and relates to a focusing device having a function of detecting a main subject from a subject pattern.

  Conventionally, there has been known a proposal for a camera that can be focused regardless of where the subject is in the screen, and the present applicant has also attempted to improve the performance of this type of technology by using cited reference 1 and the like. Further, a technique related to detection of a main subject in a screen is known from, for example, cited document 2 and the like. Although this is not a technique related to autofocus, it uses a color information or the like to determine and detect a main subject. In such a conventional technique, a circular object in the screen is regarded as a face, and such a shape is recognized as a main subject such as a person.

Japanese Patent No. 2871734 JP 05-041830 A

  The subject detection function described above often functions correctly when used in limited scenes such as portrait photography or snap photography mainly of people. However, for example, when there are many people at a wedding or the like, the main subject cannot be specified, and there are many scenes where it is difficult to know where to focus (focus).

  The technique taught by the above cited reference 2 uses color information in the screen, and thus requires a complicated configuration for color information input / processing and the like. Further, in the color determination, if there are many same-colored objects on the screen, erroneous distance measurement may occur. Conventionally, the main subject detection technology has been known to prioritize people or prioritize close subjects, but there was no idea of giving the person who is correctly facing the camera the highest priority. .

In view of the above disadvantages of the prior art, the distance measuring device of the present invention can determine a portion in the shooting screen where the main subject is present and take a beautiful photograph by correctly focusing on that portion. The aim is to realize a more advanced AF camera. An object of the present invention is to provide a focusing device with further improved accuracy of main subject determination.

Autofocus that allows you to take a photo that is correctly focused on the subject In many cases, the main subject is a person, so the person's head is identified from the image pattern (image shape) on the screen and focused (
Focusing technology prevents focusing on the background and miscellaneous subjects and speeds up by eliminating the extra distance measurement processing step. A focusing device is proposed that has a function of appropriately determining which person (head) of them can be focused even if the heads of the heads are scattered.

  In the present invention, the priority order of focusing on the person in the screen is determined by the determination technique having such characteristics. If the priority order is determined, the speed can be increased as described above if distance measurement is not performed for the lower priority. In addition, the object with higher priority is first measured, and further subject selection is performed in consideration of the distance data according to the obtained distance result, so that it is possible to select the main subject with higher accuracy and focus on the point. Is.

In order to achieve the above object, a focusing apparatus according to a first aspect of the present invention includes a detection unit that detects a subject image signal, and a head that determines and detects a human head among the subject image signals. Determination means; candidate determination means for determining the head of the person as a focus point candidate based on the determination result of the head determination means; and contrast detection for detecting the contrast of the image signal in the center of the face area of the person And the focus point candidate determined by the candidate determining means, it is determined from the contrast of the image signal in the center of the face area of the person whether or not the face of the person is facing the front. and determining the face direction detecting means whether sideways or not the symmetry of the image of the region, based on the detection result by the face direction detecting means, for determining an in-focus point to focus process if Asepo A cement determining means, anda focusing means for performing focusing processing based on the focus point determined by the focal point determining means, upper Kigoase point determination means, by said face direction detecting means The face person who is determined to be facing the front with respect to the focusing device is preferentially determined as a focusing point.

  According to the present invention, it is possible to provide a focusing device that further improves the accuracy of main subject determination.

1 (a) to 1 (c) show an outline of the distance measuring device of the present invention, FIG. 1 (a) is a principle diagram for explaining the distance measuring principle of the distance measuring device, and FIG. FIG. 1C is a schematic configuration diagram of a distance measuring device, and FIG. 1C is a perspective view of a camera incorporating the distance measuring device. FIG. 2 is a flowchart showing an imaging sequence related to the distance measuring apparatus according to the first embodiment of the present invention. FIGS. 3A to 3C show head detection in a shooting scene, FIG. 3A is an explanatory diagram showing a screen with a plurality of subjects, and FIG. 3B shows an image of head detection. Explanatory drawing and FIG.3 (c) are explanatory drawings which show the image of the determined candidate point. 4A and 4B show the head detection method, FIG. 4A is an explanatory diagram showing image data detection points in the horizontal direction, and FIG. 4B is a change at each detection point. A graph showing a pattern. 5 (a) and 5 (b) show the head detection method, FIG. 5 (a) is an explanatory diagram showing a contour pattern obtained by processing the change of the image signal, and FIG. 5 (b) is obtained. Explanatory drawing which shows the head area | region selected from the selected outline pattern. FIG. 6 is a flowchart showing a procedure of main object detection and determination performed by the distance measuring apparatus according to the second embodiment of the present invention. 7A to 7D show the orientation of the face in the face area, FIG. 7A shows an example of a subject screen facing backward, FIG. 7B shows an example of a subject screen facing forward, and FIG. ) Is an example of a front-facing subject screen with glasses, and FIG. 7D is an example of a side-view subject screen. 8A to 8C show a method of discriminating backward and forward, FIG. 8A is a flowchart for determining backward, FIG. 8B is a graph of sensor data without contrast, and FIG. c) is a graph of sensor data with contrast. FIG. 9 is a flowchart showing a procedure of low contrast determination. 10A to 10E show a face area vertical line determination method, FIG. 10A is a flowchart showing a face area vertical line determination procedure, and FIGS. FIG. 10E is a graph showing the sensor data of y2, and FIG. 10E is an explanatory diagram showing the position of the face area. FIG. 11 is a flowchart showing a procedure for determining a horizontal line in the center of the face area. FIG. 12 is a flowchart of asymmetric determination for determining the face orientation. FIG. 13 is a scene showing an example of selecting a nearby subject as a main subject.

The gist of the present invention will be described in detail below with reference to a plurality of embodiments.
(First embodiment)
1A to 1C, a configuration of a distance measuring apparatus according to an embodiment and a principle of distance measuring will be described. When shooting with the camera facing the subject, in the passive distance measurement shown in FIG. 1A, the image of the subject 9 has a pair of light-receiving lenses 3a and 3b arranged with a base line length B apart. Then, the light enters the sensor arrays 4a and 4b that are arranged in parallel and perpendicularly to the optical axes of the respective lenses. Since each sensor constituting the sensor arrays 4a and 4b outputs an electrical signal corresponding to the brightness of the image by photoelectric conversion, a pair of image signals is obtained from each of the sensor arrays 4a and 4b. However, due to the effect of the base line length B, the relative image position having these lens optical axes as the origin has a difference X. This difference X changes according to the relationship expressed by the following equation according to the subject distance L and the lens focal length f.
x = (B · f) / L (Expression 1).

  As shown in FIG. 1B, after the outputs of the sensor arrays 4a and 4b are converted into digital values by the A / D converter 4, the CPU 1 as an arithmetic control means (control unit) composed of a one-chip microcomputer or the like Based on Equation 1, a relative position difference X is calculated. Here, since (B · f) is a known value by the camera distance measuring device 2, the subject distance L is obtained.

  The CPU 1 is responsible for the shooting sequence of the entire camera, detects manual operation of the release switch 7 and appropriately controls the distance measuring device 2, and moves the shooting lens 6 via the focusing mechanism 5 based on the result, to the subject 9. Focusing and shooting control are performed. In addition, the light emission control of the flash light emitting unit 8 and the like are appropriately performed as necessary for exposure.

The sensor arrays 4a and 4b are two-dimensionally arranged as shown in FIG. 1B and function as area sensors, so that not only a one-dimensional image signal but also a two-dimensional image signal of the subject 9 is obtained. It is configured as follows.
An external view of a camera body incorporating the distance measuring device 2 of the present embodiment is shown. As shown in FIG. 1C, a photographing lens 6, AF light receiving lenses 3a and 3b, and a flash are provided on the front surface of the camera body 10. A light emitting unit 8 and the like are arranged. When the release switch 7 on the upper surface of the camera body 10 is pressed, the shooting sequence is started. This photographing sequence is appropriately controlled by a command from the CPU 1 in accordance with a predetermined program (referred to as a “camera sequence program”) written in a built-in ROM. The program includes a head determination unit that determines the head of a person from the detected image signals, and a determination unit that determines a focus point based on the determination result. A software direction determination unit for determining the direction is also provided.

  FIG. 2 is a flowchart showing an example of this photographing sequence (subroutine). When the release switch is operated, first, in step S1, the image of the subject in the shooting screen is detected by the area sensor 4a or 4b (S1). The head of the person is detected from the result. Although there are several methods for detecting the head, detection is performed in the order of the images of the photographic scene shown in FIGS. 3A to 3C (details will be described later).

  In step S2, the head of the person is detected and candidate points for distance measurement are determined (S2), then the back of the head is detected in step S3, and those corresponding to this are deleted from the candidate point group. (S3). In step S4, the distance is measured only for the determined candidate point (S4), and the main subject of the candidate point is focused in step S5 (S5). In step S6, a predetermined photographing operation is performed (S6), and the process returns to the main routine of the camera sequence program.

  Here, the above-described procedure will be described in more detail with reference to the explanatory diagrams of FIGS. 3A to 3C. On the shooting screen 11 in FIG. 3A, two persons 9a and 9b are subjects. It has become. In this case, as shown in FIG. 3B, the portions 9a1 and 9b1 of the subject are detected as the heads of the respective persons, and are set as candidates for the main subject.

However, if the portion 9a1 is deleted from the distance measurement candidate group by the back head detection in step S3 in FIG. 2, only the subject portion 9b1 is selected as shown in FIG. Narrow down from multiple to one.
If only this point is measured in step S4 and the photographing optical system is driven based on the obtained distance value to focus on the main subject, it is possible to take a photograph that is correctly focused at high speed. . As described above, the premise here is that, when there are a plurality of persons on the screen, the person who faces forward is given priority over the person facing backward.

  The head detection method which is the premise of the present embodiment will be described in detail with reference to FIGS. 4 (a) and 4 (b) and FIGS. 5 (a) and 5 (b). Referring to FIGS. 4A and 4B, the method will be described. The procedure of head detection and determination is as follows. First, the image data change is performed at a plurality of points (circle number 1, circle number 2, circle number 3). The head change (circle number 2 in FIG. 4 (b)) from the top, the overhead pattern (circle number 1 in FIG. 4 (b)) to the shoulder width pattern (circle number 3 in FIG. 4 (b)) ) And the difference in the shape of the image signal. By this method, it is possible to determine that the range in which a pattern such as the circled number 2 in FIG. 4B (the waveform starts to protrude upward) is obtained is the region where the head of the subject exists.

  Further, as illustrated in FIG. 5A, after detecting the change of the image signal in the screen by using the differential signal, if the result is binarized, the contour of the person image can be obtained. Judge the item closest to. As a result, as shown in FIG. 5B, the head region 9b2 of the person can be recognized and the head can be specified.

  (Effect 1) As described above, according to the first embodiment, the head of a person in the screen is detected and identified, and the main subject is determined by the orientation of the head (forward / backward). In order to determine the order that will be, the priority order of focusing on the person in the screen is determined. If the priority to be measured is determined, those with a low degree are excluded from the distance measurement target so as not to measure the distance. In particular, a backward-facing person usually cannot be a main subject and can be excluded from focus candidates, and accordingly, the processing time required for distance measurement on the main subject can be reduced accordingly. Note that the detection of the back of the head may be determined based on the contrast of the hair (black, brown, etc.). If such a distance measuring function is used for an AF camera, it is possible to speed up AF shooting including a distance measuring operation as a whole and to achieve accurate focusing.

  (Second Embodiment) A distance measuring apparatus as a second embodiment of the present invention can be mechanically implemented by an apparatus equivalent to that of the first embodiment described above, but is characterized by a method for determining a main subject. There is. The flowchart in FIG. 6 shows an example of main subject detection and determination performed by the distance measuring apparatus of the second embodiment (procedure for discarding candidate points). In this example, unlike the first embodiment, not only the backward-facing head but also the side face is excluded from the focusing candidates, and the forward-facing head is preferentially incorporated into the focusing candidates. The accuracy is further improved.

Step S11 is performed when heads are detected by the method described in FIGS. 4A, 4B, 5A, 5B, etc., and then a plurality of n heads are detected. In order to determine the orientation of the face with respect to the head (point), this variable n is first initialized (S11).
In step S12, the image signal of the selected n-th point is determined (S12). In step S13, if this signal is low contrast (hair: black, brown, etc.) and it is determined that it is facing backward, the process proceeds to step S23. Branch and set the coefficient P (n) indicating the priority to a low value (ie P (n) = 1). Whether or not it faces backward is determined by whether the contrast of the face area surrounded by a circle is low (that is, low contrast) or high as shown in FIG.

  If the face is facing the front, it is determined in steps S14 and S15. That is, whether or not a nose line (i.e., a vertical line) is recognized at the center of the face area as shown in FIG. 7B (S14), or the hair, glasses, or eyebrows line as shown in FIG. Whether or not (that is, a horizontal line) is recognized (S15). When this determination result is YES, it is regarded as a face facing front, and the process branches to step S20, and the priority is set to a high value of P (n) = 10.

  On the other hand, in the case of the subject head as shown in FIG. 7D, if it is determined in step S16 that the image of the face area is asymmetrical, the face is determined to be sideways (S16). In this case, in step S22, the priority of distance measurement is set as a priority coefficient P (n) = 5 between the front and the rear (S22). Otherwise, the process branches to NO in step S16, and in step S21, it is determined that the face orientation is not well understood, and P (n) = 2 is set (S21). In this way, all the points (1 to n) are processed while incrementing the variable n one by one in step S26.

  When the orientation of the face portion of each point is determined, three points with large priorities P (n) are selected in the following step S27, and distance measurement is performed (S27). As a result, in step S28, the closest distance measurement results are selected, and candidate points are narrowed down to one, that is, selected (S28). If such a process of the flowchart is performed in step S3 in FIG. 2 in the shooting sequence, the distance measuring point can be accurately determined, and the correct focus can be achieved no matter where the main subject is on the screen.

  The low contrast determination is performed, for example, according to the procedure shown in FIG. The graph patterns shown in FIGS. 8B and 8C show sensor data corresponding to the image pattern of the face area as shown in FIGS. 7A and 7B. In other words, the sensor data does not have contrast in the back subject as shown in FIG. 8B, and conversely in the forward subject, contrast occurs due to the nose line or the like as shown in FIG. 8C. According to this principle, the backward determination is made along the flowchart of FIG.

  In this example, an average value (SAV) of image signals in the face area is calculated (S30), a difference from this SAV is obtained from each data, and a deviation σ is calculated (S31). The deviation σ is compared with a predetermined value σ0 (S32), and if the deviation σ is larger than the predetermined value σ0, it is determined that “contrast is present”. (S33), the CPU 1 recognizes that this low contrast region is the back of the head.

  Further, as in the low contrast determination according to another flowchart shown in FIG. 9, the MAX value and the MIN value of the sensor data of the face area are respectively obtained (S35, S36), and the difference (MAX value−MIN value) is obtained. It may be compared with a predetermined value ΔD (S37) to determine whether the contrast is low (S38).

  These predetermined values, σ0, ΔD are not fixed values, and if they can be switched by the sensor data average value (SAV) of the face area, they are less affected by the environment such as backlight and direct light, and are more effective. It is. These flowcharts correspond to step S13 in FIG.

  Next, a face area vertical line determination method will be described with reference to FIGS. This method is executed according to the procedure shown in FIG. As shown in FIG. 10 (e), it is assumed that the face area is detected as a substantially circular shape, the start coordinates in the X and Y directions are xS, yS, and the end coordinates of the face area are xE, yE (S40, S41). ). From this, the center coordinates x2, y2 of the face are obtained (S42).

  In order to detect the nose line, reference lines y1 and y3 are provided above and below the diameter of the circle from the center (S43, S44). Further, when the coordinates corresponding to the lines on both sides of the nose line x2 as a reference for the data change determination of the sensor detecting the nose line are x1 and x3 (S45, S46, S47), in the subsequent steps S48, S49 This is calculated. More specifically, x1 and x3 are places left and right from the center line by a position that is 1/4 of the lateral width of the face (S48, S49).

  The determination step group after step S50 uses the above y1, y2, y3, x1, x2, and x3, and uses each sensor data S (X, Y) at the coordinates, and FIGS. 10 (b), (c), As shown in (d), it is determined whether or not the data change along the line Y = y1, the line Y = y2, and the line Y = y3 is “dented” in the central portion (x2). For this purpose, S (x1, y) >> S (x2, y) when comparing data of x1 and x2 of the same Y coordinate, and S (x2, y) << S (x3, y) when comparing data of x2 and x3. By using this, it is determined that the line passes through black at the branches in steps S50 and S51, steps S52 and S53, and YES in steps S54 and S55 (S56).

  Note that the flowchart shown in FIG. 10 of the second embodiment corresponds to step S14 (face area center vertical line determination) in FIG. 6 described above. Moreover, step S15 (face area center part horizontal line determination) in FIG. 6 is realizable with a flowchart like FIG. Next, a horizontal line is determined using x1, x2, x3, y1, y2, y3 obtained in steps S60 and S61 in FIG. 10 and the sensor data average value (SAV) of the face area. That is, Steps S71, S72, S73, Steps S74, S75, S76 and Steps S77, S78, S79 are each a set, and the Y coordinate is changed, and a line (line ) Is a step for determining whether or not. This determination is made based on whether or not the data on the line (Y = y1, y2, y3) is at a lower level than the average value SAV of all the sensor data of the face.

  When branching to YES, YES, and YES in steps S71, S72, and S73, respectively, when crossing in the horizontal direction at the y1 coordinate, it indicates that shadows are formed at all points x1, x2, and x3, and this shadow is connected. Judging from the fact that it becomes a horizontal line. The line crossing the face portion may be formed not only at the sensor y1, y2 as shown in FIG. 11B but also at any one of the sensors y1, y2, y3. Steps S71 to S73, S74 Branches from S76 and S77 to S79, and proceeds to the horizontal line determination step of Step S80. Thereby, shadows of eyebrows and glasses can be determined. Such a state is a situation that cannot occur when the face is facing backwards or sideways.

  FIG. 12 shows a flowchart of asymmetric determination for determining that the face is sideways as shown in FIG. This corresponds to step S16 in FIG. When the face is sideways, the sensor data is not symmetrical along the X direction as shown in FIG. This determination is made based on the y2 coordinate at the center in the vertical direction of the face.

  First, in steps S90 and S91, the data at x1 and x2 are compared, and if either is greater, the direction of change is determined at steps S92 and S93. Assuming that it cannot be symmetric, asymmetric determination is performed in subsequent step S94. Here, if the pattern of the face area is asymmetric, it is determined that the pattern is not at least backward, and the priority coefficient is set to P (n) = 5 in step S22 in FIG.

  (Effect 2) As described above, according to the second embodiment, in addition to the method of “person priority” in the main subject detection technique, based on the idea that the person who is correctly facing the camera is given the highest priority, The person who faces forward is given the highest priority, the person who faces horizontally is prioritized next, and the person who faces backward is excluded from the distance measurement as much as possible, so the main subject can be determined at high speed without the user being aware of the focus. In addition, it is possible to focus on a person with priority given to a person who is positive and accurate.

  In addition, in consideration of the concept of selecting “close” in step S28 in FIG. 6, even in a scene where a plurality of persons (9a2, 9a3) facing the front as illustrated in FIG. In consideration of the above, it becomes possible to focus correctly with “close subject priority” to the main subject. In particular, this function can be said to be a very effective technique because it can focus correctly even when shooting with a self-timer or shooting with a remote control without looking through the viewfinder. In other words, a subject having a high priority (for example, the persons 9a2 and 9a3) is first measured, and further subject selection is performed in consideration of the distance data according to the obtained distance result, and the main subject having higher accuracy (for example, After selecting the person 9a2), the correct focusing (focusing operation) for that point becomes possible.

  Therefore, this distance measuring device can determine a main subject with a simple configuration suitable for an AF camera, increase the priority at the time of distance measurement of the subject, determine the distance of each subject, and instantaneously perform correct focus control. Since this is a device, it is possible to provide a cheaper and higher performance AF camera by incorporating it.

  (Modification) The plurality of embodiments described above may be modified as follows. For example, the present invention is not limited to the distance measurement method, and may be an active method in addition to the passive method. The present invention can be similarly applied as long as an image image can be captured and analyzed by a sensor array. . Therefore, the present invention can be easily applied to a camera that already has an electronic imaging system such as an area sensor. Thereby, the effect equivalent to or more than the above can be expected. In addition, various modifications can be made without departing from the scope of the present invention.

As mentioned above, although demonstrated based on multiple examples, the following invention is included in this specification.
(1) Provided with a built-in distance measuring device, and a camera for detecting and focusing a person's head, comprising a discrimination means for discriminating whether the person's head is facing forward or backward it can.

(2) The camera according to (1), wherein the determination unit excludes the back head of the person facing backward (low contrast head) from the distance measurement object even if the head of the person is detected.
(3) The camera according to (1), wherein the discrimination unit preferentially discriminates a glasses-like pattern, a nose-like pattern, and a mayu-hair-like pattern after detecting a human head. .

  (4) A sensor array that detects a subject image signal, a control unit that determines and recognizes a main subject based on the subject image signal, and performs distance measurement on the main subject, and according to the control of the control unit In a camera including a photographing optical system that performs a focusing operation on the main subject, the control unit determines a person's head from the detected subject images and determines a candidate for distance measurement. A step of selecting a head facing backward out of the heads and deleting the head from the candidates; a step of instructing a subject as a remaining candidate to perform a ranging operation and a focusing operation; The camera characterized by having can be provided.

  As described above, according to the present invention, it is possible to provide a focusing device in which the accuracy of main subject determination is further improved. Therefore, since the position of the main subject is correctly determined and focused there, it is possible to realize a camera with a simple configuration that allows anyone to enjoy taking a picture without worrying about the composition.

    DESCRIPTION OF SYMBOLS 1 ... CPU (control part: Calculation control means), 2 ... Distance measuring device (ranging optical system), 3a, 3b ... Light receiving lens, 4 ... A / D conversion part, 4a, 4b ... Sensor array (detection means), 5 ... Focusing mechanism (focusing part), 6 ... Shooting lens, 7 ... Release switch, 8 ... Flash light emitting part, 9 ... Subject (person), 10 ... Camera body, 11 ... Shooting screen (viewer field of view), S1- S2 ... Shooting sequence, S11-S28 ... Candidate point selection sequence (head determination means), S30-S38 ... Low contrast determination sequence (direction determination unit), S40-S56 ... Face area vertical line determination sequence (direction determination unit), S70 to S80: Face area horizontal line determination sequence (orientation determination unit).

Claims (2)

Detecting means for detecting a subject image signal;
A head determination means for determining and detecting a head of a person from the subject image signal;
Candidate determination means for determining the head of the person as a focus point candidate based on the determination result of the head determination means;
Contrast detection means for detecting the contrast of the image signal at the center of the human face area;
For the in-focus point candidate determined by the candidate determining means, it is determined whether or not the person's face is facing the front from the contrast of the image signal at the center of the person's face area, and the image of the person's face area Face direction detection means for determining whether or not it is horizontal from the symmetry of
Based on the detection result by the face direction detection means, a focus point determination means for determining a focus point for performing the focusing process;
Focusing means for performing focusing processing based on the focusing point determined by the focusing point determination means;
Comprising
Upper Kigoase point determination means, by the face direction detecting means, the determined face of a person as a front direction to the focusing device preferentially, and determines as an in-focus point Focusing device. The focusing apparatus according to claim 1, wherein the focusing apparatus is a focusing apparatus mounted on a camera.

Images