JP2017224923A - Autofocus device and autofocus program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a proper autofocus at a small load irrespective of a number of a target of focus or a fluctuation of a position.SOLUTION: An autofocus device of an imaging system comprises an evaluation value calculation part and a focusing position determination part. The evaluation value calculation part calculates an evaluation value for determining a focusing position in each image data obtained by imaging a subject at a focal position. The focusing position determination part determines the focusing position on the basis of the evaluation value of a plurality of focal positions. The evaluation value calculation part generates evaluation data contributed by a positive weight to a first score component and contributed by a negative weight to a second score component by using the first score component calculated for a first region which has the size including a target and the second score component calculated to a second region which has the size including a peripheral part of the target in each local area of image data, and calculates the evaluation value on the basis of the evaluation data in the plurality of local regions.SELECTED DRAWING: Figure 3

Description

  The present invention relates to an autofocus device and an autofocus program for detecting a focus position by a contrast method.

  2. Description of the Related Art Conventionally, an autofocus device captures an object while changing the lens focal position of an optical system in stages, and acquires an evaluation value such as contrast information based on a photographed image of the object at each stage. Then, the lens focus position at the stage where an optimum evaluation value (for example, the maximum evaluation value) can be obtained is detected as the in-focus position. Various improvements have been made to the autofocus device in order to detect a more optimal focus position according to the environment and subject.

  For example, the autofocus camera of Patent Document 1 has an arithmetic control unit that controls the focus position of the photographing lens depending on distance information to the subject. Then, the calculation control unit determines the focus position based on the distance information and background information other than the distance.

  In addition, the autofocus device disclosed in Patent Document 2 includes an AF evaluation value generation unit that generates AF evaluation values for the central portion of the photographing field and the entire two AF areas. Further, when the focus lens in-focus positions calculated from the AF evaluation values of these AF areas are substantially equal, a CPU for determining the focus lens in-focus position from the difference between the AF evaluation values of the central AF area and the entire AF area is provided. Provide.

JP 2000-321483 A JP 2004-309722 A

  However, in a conventional autofocus device, when the focus target is a small area and the subject has a large surrounding contrast, the contrast information (evaluation value) when the surrounding is focused is maximized. There is a problem that the focus is easily drawn to the surroundings. In this case, the actual shooting is performed at the in-focus position where the surrounding is in focus, and an image in which the target is blurred is generated. For example, in a machine vision system that inspects an electronic component or the like, an electronic component having a protruding member such as an electrode may be photographed from the tip side of the protruding portion to determine the position of the protruding portion. At this time, the tip of the projection becomes the focus target, but it protrudes from the periphery, so that when viewed from the tip, the surrounding contrast is small and the surrounding contrast is large, making it difficult to focus.

  On the other hand, by setting a region of interest (ROI) only in the vicinity of the target and covering it so as not to refer to the surroundings, it is naturally possible to shoot at a focus position where the target is in focus. is there. However, in general, under shooting conditions where it is desired to automatically control the focus, the position of the target within the angle of view is often unknown accurately, and fluctuations may occur between the timings of imaging. In addition, the smaller the size of the target, the higher the probability that the target will deviate from the fixed region of interest due to the position variation.

  In addition, in order to avoid such a situation, a technique for adaptively identifying a target position by pattern matching or the like and tracking a dedicated field angle or region of interest to the position has been proposed. Such pattern matching and tracking processing complicates the configuration and processing, and increases the load. Furthermore, when there are a plurality of targets and their positional relationship is unknown, it becomes more difficult to detect contrast information in a region matched to the target.

  Furthermore, for example, a focus mechanism or the like used in a manufacturing apparatus may be required to have very high performance with respect to detection accuracy and speed instead of the physical conditions of a subject to be focused being limited to some extent. . However, in the conventional focusing apparatus as described above, when the focus target is limited to a small area, there is a possibility that the accuracy and speed performance of target detection may be reduced.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to realize appropriate autofocus with a small load regardless of the number and position of focus targets.

  In order to achieve the above object, an autofocus device according to the present invention is an autofocus device that determines a focus control state for the subject in accordance with a contrast of an image of the subject, wherein the subject is detected at each of a plurality of focus positions. For each of a plurality of images obtained by imaging, an evaluation value calculation unit that calculates an evaluation value used for determination of an in-focus position, and a plurality of the evaluation values calculated for each of the plurality of focus positions A focus position determination unit that determines a focus position, and the evaluation value calculation unit includes a focus target for each of the plurality of local regions included in each image. Including a first score component calculated for the first region of the target and at least a part of a peripheral portion of the target around the first region. Using a second score component calculated for a second region of size, so as to contribute to the first score component with a positive weight and to contribute to the second score component with a negative weight The evaluation data calculated in the above is generated, and the evaluation value is calculated based on the plurality of evaluation data generated for each of the plurality of local regions in each image.

  For example, the evaluation value calculation unit provides a positive coefficient corresponding to the positive weight to the first region and corresponds to the negative weight for the pixel data of each local region in each image. It is preferable to perform a convolution operation using a spatial weighting filter that gives a negative coefficient to the second region, and to generate the evaluation data based on a result of the convolution operation.

  Further, the evaluation value calculation unit performs a frequency filter process using a spatial frequency filter having a predetermined bandpass characteristic on the pixel data of each local region in each image, and after the frequency filter process A conversion process for converting the amplitude component into a value representing the magnitude of the amplitude component may be performed based on the data, and the first score component and the second score component may be calculated based on the result of the conversion process. Alternatively, the evaluation value calculation unit may perform the convolution operation using a combined spatial filter in which the weighting filter and the frequency filter are combined in advance.

  Further, for example, the evaluation value calculation unit uses, as pixel data in each image, information corresponding to a gradation having a linear or nonlinear relationship with the luminance of the subject imaged at each pixel position, An index corresponding to a contrast in which the gradation is distributed for each of the plurality of local regions in each image is calculated.

  The evaluation value calculation unit may set a maximum value of the plurality of evaluation data as the evaluation value for each image, and the focus position determination unit may determine the plurality of focus positions with respect to each of the plurality of focus positions. The focus position where the maximum value can be obtained from the evaluation values is preferably determined as the focus position.

  According to such a configuration, the autofocus device detects the in-focus position where the target is in focus even when the focus target and the peripheral portion of the target have different focus positions. Can do. For this reason, for example, when shooting a subject having a target with a smaller area than the peripheral part, such as an electronic component having a protruding member such as an electrode, from the tip side of the protrusion, the tip of the protrusion is focused. The in-focus position can be detected. Alternatively, even when an image of a component having a recess is taken from the front of the recess, it is possible to detect the in-focus position where the deep portion of the recess is in focus. In addition, the autofocus device can quickly and surely detect the in-focus position of the target without providing complicated processing and configuration even when the position of the target fluctuates according to the shooting timing. Therefore, when this autofocus device is employed in a machine vision system or the like, it is possible to improve the processing speed of photographing and inspection, reduce the processing load, and further reduce the equipment cost.

  Further, the evaluation value calculation unit performs a pixel size reduction process on the data after the frequency filter process, and calculates the first score component and the second score component based on the data after the reduction process Good.

  According to such a configuration, the autofocus device can improve the processing speed in the autofocus operation and reduce the processing load.

  Further, the evaluation value calculation unit may use the weighting filter having a third region as a dead zone having a weight of 0 between the first region and the second region.

  According to such a configuration, for example, when shooting a subject in which the tip of the target of the projection has a minute area when viewed from the shooting side and the area increases from the tip to the base end, Even when shooting a subject whose deep part of the concave target has a very small area and the area increases from the deep part to the edge of the concave part, focusing on other than the tip or deep part of the target is avoided to ensure that the target is The in-focus position can be detected.

  The autofocus device further includes an appropriate distribution acquisition unit that acquires information on an appropriate distribution range or polarity with respect to the focus position, and the focus position determination unit includes the plurality of evaluation values for the plurality of focus positions. One or more local peak positions may be calculated from the above, and the focal position at which the local peak position most suitable for the appropriate distribution is obtained may be determined as the in-focus position.

  According to such a configuration, even when there are two or more protrusions or recesses having different focal positions in the subject, it is possible to select an optimum target and focus.

  Furthermore, the autofocus device may include an area setting unit that can change at least one of the size of the first area and the size of the second area, and contributes to the first score component. A contribution rate setting unit that can change at least one of the positive weight and the negative weight that contributes to the second score component may be provided.

  According to such a configuration, the autofocus device can perform detailed setting for each subject (for example, an inspection product in a machine vision system or the like) for a target to be focused on the subject.

  Further, the autofocus program of the present invention is a program for causing a computer to function as any of the above autofocus devices, the evaluation value calculation unit and the in-focus position determination unit provided in the autofocus device, Each processing step has the same function as the appropriate distribution acquisition unit, region setting unit, and contribution rate setting unit.

  According to the present invention, it is possible to realize appropriate autofocus with a small load regardless of the number of focus targets and the variation in position.

1 is a block diagram illustrating an imaging system including an autofocus device according to an embodiment of the present invention. It is a schematic diagram showing an example of a photographed image of a subject photographed by an imaging system including an autofocus device according to an embodiment of the present invention. It is a flowchart which shows the autofocus operation | movement in the autofocus apparatus based on one Embodiment of this invention. It is a graph which shows the example of the weighting filter produced | generated by the weighting filter production | generation part of the autofocus apparatus which concerns on other embodiment of this invention. It is a schematic diagram which shows the example of the image data of the object image | photographed with the imaging system provided with the auto-focus apparatus which concerns on one Embodiment of this invention. It is a graph which shows the example of a part of image data of the subject image | photographed with the imaging system provided with the autofocus apparatus which concerns on one Embodiment of this invention. It is a graph which shows the example of a part of data after the spatial filtering process based on the image data of the object image | photographed with the imaging system provided with the autofocus apparatus which concerns on one Embodiment of this invention. It is a graph which shows the example of a part of data after the conversion process to the amplitude component based on the image data of the object image | photographed with the imaging system provided with the autofocus apparatus. It is a graph which shows the example of a part of data after the weighting process based on the image data of the object image | photographed with the imaging system provided with the auto-focus apparatus which concerns on one Embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. Although the autofocus device of the present invention may be an independent device, the autofocus device includes an imaging unit such as a camera, a smartphone, and a tablet terminal, and is provided as part of various devices capable of imaging control and image processing. Alternatively, it may be an autofocus device provided as a part of various devices such as a management device that is connected to an imaging unit and can perform imaging control and image processing in an imaging system such as a machine vision system. That is, the autofocus device of the present invention may be provided as a part of various devices or computers having an autofocus function based on a captured image. Similarly, the autofocus program of the present invention may be provided as a part of various devices or computers having an autofocus function based on a captured image. In the present embodiment, an imaging system 1 including an autofocus device according to the present invention will be described.

  As shown in FIG. 1, the imaging system 1 includes an optical system 10, an optical system driving unit 11, an imaging unit 12, an imaging driving unit 13, an A / D conversion unit 14, an image processing unit 15, and an area. A setting unit 16, a weighting filter generation unit 17, an evaluation value calculation unit 18, and an in-focus position determination unit 19 are provided. In addition, the imaging system 1 includes a control unit 30 including a CPU, a storage unit 31 including a ROM and a RAM, a recording medium 32 that records image data, an operation unit 33, and a display unit 34. In FIG. 1, subject images, image data, and the like exchanged between the units of the imaging system 1 are indicated by dotted lines, and control signals, drive signals, and the like are indicated by solid lines.

  The imaging system 1 includes, for example, a camera (imaging device) including an optical system 10, an optical system drive unit 11, an imaging unit 12, an imaging drive unit 13, and an A / D conversion unit 14, and a personal computer (management) including other units. Device) in a communicable manner. Alternatively, the imaging system 1 may be configured by including each unit in one camera. The autofocus device of the imaging system 1 is configured to include at least an area setting unit 16, a weighting filter generation unit 17, an evaluation value calculation unit 18, and a focus position determination unit 19, and the optical system 10 includes an optical system 10 based on input image data. The optical drive unit 12 is controlled.

  The imaging system 1 is configured such that the control unit 30 connected to each unit via the bus 30a operates by controlling each unit. The storage unit 31 stores, for example, programs and data necessary for shooting control, autofocus control, image processing control, and other programs and data for controlling other functions of the imaging system 1. The control unit 30 executes arithmetic processing according to each program stored in the storage unit 31 to control each unit connected to the control unit 30.

  For example, the image processing unit 15 may be configured by a program (that is, an image processing program) stored in the storage unit 31 and executed by the control unit 30. In addition, the region setting unit 16, the weighting filter generation unit 17, the evaluation value calculation unit 18, the in-focus position determination unit 19, a contribution rate setting unit 20 and an appropriate distribution acquisition unit 21 to be described later are stored and controlled in the storage unit 31. The program may be configured to be executed by the unit 30 (that is, an autofocus program applied to the imaging system 1 or the camera). In this case, the area setting unit 16, the weighting filter generation unit 17, the evaluation value calculation unit 18, the in-focus position determination unit 19, the contribution rate setting unit 20, and the appropriate distribution acquisition unit 21 described later are each processed by the autofocus program. Operates as a step. For example, the autofocus program is incorporated so as to operate when the shooting mode of the imaging system 1 or the camera is selected as a pin focus mode for focusing on the protruding tip of the subject or the deep part of the recess.

  The optical system 10 is a light incident mechanism that captures a subject image and enters the imaging unit 12. Although not shown, the optical system 10 includes a zoom lens, a focus lens, a shutter mechanism, and a diaphragm mechanism arranged on a predetermined optical axis. The optical system 10 is configured such that the focus lens is displaced in the optical axis direction in accordance with a drive signal from the optical system drive unit 11.

  The optical system drive unit 11 outputs a drive signal for changing the position of the focus lens of the optical system 10 to the optical system 10 based on a control signal from the control unit 30. For example, the optical system drive unit 11 sequentially outputs a drive signal that fluctuates the position of the focus lens (focal position) step by step during a contrast method (mountain climbing method) autofocus operation. Further, when the in-focus position (lens position to be in focus) is determined by an autofocus operation or the like, the optical system drive unit 11 outputs a drive signal for changing the focus lens to the determined lens position.

  Although not shown, the imaging unit 12 includes an imaging surface and a horizontal transfer path that form one screen of a captured image, and the imaging surface includes a plurality of photosensitive units and a plurality of vertical transfer paths corresponding to each pixel. Each photosensitive unit is an optical sensor that photoelectrically converts light into an electrical signal corresponding to the amount of received light when receiving incident light. For example, the imaging unit 12 includes an image sensor such as a charge coupled device (CCD), a metal oxide semiconductor (Metal Oxide Semiconductor: MOS), or the like. The imaging unit 12 photoelectrically converts a subject image formed on the imaging surface via the optical system 10 based on a drive signal from the imaging drive unit 13 and analogizes the signal charge of each pixel obtained thereby. An electrical signal is converted into a captured image and output to the A / D converter 14.

  The imaging drive unit 13 drives the imaging unit 12 based on a control signal from the control unit 30. For example, the imaging drive unit 13 outputs to the imaging unit 12 drive signals such as a charge readout pulse for reading out charges from each photosensitive unit and a transfer pulse for transferring charges in the horizontal transfer path and the vertical transfer path. To do.

  The A / D converter 14 converts the captured image of the analog electrical signal generated by the imaging unit 12 into a captured image of a digital image signal.

  The image processing unit 15 performs image processing on the captured image of the digital image signal obtained by the A / D conversion unit 14. For example, the image processing unit 15 generates image data in a format that can be recorded on the recording medium 32 or the like, and records the image data on the recording medium 32. Further, the image processing unit 15 generates image data in a format that can be displayed on the display unit 34 and the like, and causes the display unit 34 to display the image data. Further, the image processing unit 15 generates image data every time the focus lens of the optical system 10 is changed at each stage during the autofocus operation, that is, generates image data of each of a plurality of focal positions.

  The area setting unit 16 has a function of setting an area corresponding to the focus target 41 when photographing the subject 40 (see FIG. 2). As shown in FIG. 2, the region setting unit 16 includes a main target range 44 having a size including the target 41 in the entire image region 43 of the image data of the subject 40, and the target 41 around the main target range 44. A peripheral range 45 having a size including at least a part of the peripheral portion 42 is set. The area setting unit 16 determines the main target range 44 and the peripheral range based on reference data predetermined as the image data of the subject 40 or based on image data first taken when a plurality of subjects 40 are taken. 45 may be set. The area setting unit 16 may automatically set the main target range 44 and the peripheral range 45 based on the reference data, but may set the main target range 44 and the peripheral range 45 according to a user's manual operation via the operation unit 33. For example, the region setting unit 16 displays the entire image region 43 of the image data on the display unit 34 for manual operation by the user, and accepts and selects a range on the entire image region 43 via the operation unit 33. Are set as the main target range 44 and the peripheral range 45. As described above, the region setting unit 16 is configured to be able to change at least one of the sizes of the main target range 44 and the peripheral range 45, and thereby, the size of the first region 46 and the second region of the following local region At least one of the 47 sizes can be changed.

  The target 41 is a part of the entire image region 43, and has a smaller area than the peripheral portion 42, for example. The main target range 44 is an image size (main target size) including the target 41 on the assumption that the target 41 is known, and is set, for example, as a rectangle of 8 × 8 pixels or more. The peripheral range 45 is a size (peripheral size) of a region including the peripheral portion 42 (background) of the target 41 that is blurred when the target 41 is in focus, and is a rectangle that is symmetrically positioned around the target 41. Is set. The peripheral range 45 is set so as not to include other targets adjacent to the target 41 corresponding to the main target range 44 when there are a plurality of targets 41 in the entire image region 43 of the image data. Even when the subject includes a plurality of targets 41, the in-focus position of the target 41 can be reliably detected by setting the peripheral range 45 without including other adjacent targets.

  The weighting filter generation unit 17 generates a spatial weighting filter (kernel) corresponding to the main target range 44 and the peripheral range 45 set by the region setting unit 16. In the weighting filter, the weighting filter generation unit 17 sets a positive value (a positive coefficient corresponding to a positive weight) as a main target weighting factor in an area corresponding to the main target range 44, and corresponds to the peripheral range 45. A negative value (a negative coefficient corresponding to a negative weight) is set as a peripheral weighting coefficient in the region. The weighting factors set in the main target range 44 and the peripheral range 45 may be automatically set by the weighting filter generation unit 17, but may be adjustable by a user's manual operation via the operation unit 33. For example, the imaging system 1 includes a contribution rate setting unit 20 that can change at least one of the main target weighting factor set for the main target range 44 and the peripheral weighting factor set for the peripheral range 45. You may prepare. For example, the contribution rate setting unit 20 displays a contribution rate setting screen for setting the main target weighting factor and the peripheral weighting factor on the display unit 34, and contributes the setting input by the user of the main target weighting factor and the peripheral weighting factor. It may be accepted via the operation unit 33 on the rate setting screen.

  The main target weighting factor is a weighting factor for clarifying the contrast of the image focused on the target 41 when performing a relative comparison between images having different focal points during the autofocus operation, and the absolute value is not necessarily limited. However, here, the normalized value is set so that the sum of the sampling points in the main target range 44 is 1.0. The peripheral weighting factor is a weighting factor for suppressing the contrast of the image focused on the peripheral portion 42 of the target 41. That is, the peripheral range 45 penalizes the score (evaluation value) obtained from the contrast of the image data. It is an area to give. The peripheral weighting factor is not necessarily limited as long as it is a negative value, and may be selected in consideration of the penalty effect, but here it is set to a value that does not change the level after application depending on the size. Normalization is performed so that the sum of the sampling points in the range 45 is −2.0.

  The weighting filter is used for a convolution operation on each of a plurality of local area data (hereinafter simply referred to as local areas) having different pixel positions in the image data (all image areas 43) at each focal position. The local region has a size including the main target range 44 and the peripheral range 45, that is, includes a first region 46 corresponding to the main target range 44 and a second region 47 corresponding to the peripheral range 45. For example, a plurality of local regions can be obtained by scanning the entire image region 43 for each pixel position. Based on the result of this convolution operation, evaluation data for each local region is calculated. The evaluation data of each local area is calculated based on a score component (score component serving as an index of sharpness) having a high correlation with the sharpness, details of which will be described later. By performing a convolution operation using a weighting filter, a positive weight contributes to the first score component when the first score component of the first region 46 is calculated. When calculating the score component of 2, a negative weight contributes to the second score component.

  The evaluation value calculation unit 18 calculates an evaluation value for evaluating the optimum in-focus position based on the image data of each focal position imaged during the autofocus operation. That is, the evaluation value calculation unit 18 calculates an evaluation value at each focal position. Calculation of the evaluation value during the autofocus operation will be described with reference to the flowchart of FIG.

  Before starting the autofocus operation, a weighting filter is generated by the weighting filter generation unit 17 (step S1). The weighting filter generation unit 17 generates a weighting filter corresponding to the two-dimensional main target range 44 and the peripheral range 45, but in order to facilitate understanding, the one-dimensional (left-right direction) weighting filter is represented in the graph of FIG. Show. In FIG. 4, the vertical axis indicates the filter coefficient F that determines the level of the weighting coefficient, the horizontal axis indicates the pixel position coordinate i in the horizontal direction, and the origin of the horizontal axis indicates the center of the main target range 44. As shown in FIG. 4, the weighting filter is configured by setting a main target weighting factor in the main target range 44 and setting a peripheral weighting factor in the peripheral range 45.

  The evaluation value calculation unit 18 inputs, for example, two-dimensional image data constituting the entire image region 43 as shown in FIG. 5 (step S2), but in order to facilitate understanding, the evaluation value calculation unit 18 performs one-dimensional along the line B. The image data in the (left-right direction) is shown in the graph of FIG. In FIG. 5, the target 41 protrudes from the subject 40 to the photographing side (camera side). 6A shows data in a state in which the peripheral portion 42 of the target 41 is focused (peripheral focused state), and FIG. 6B shows data in a state in which the target 41 is focused (target focused state). Show. Hereinafter, in order to facilitate understanding, one-dimensional (left-right direction) data is also shown in FIGS.

  The evaluation value calculation unit 18 generates evaluation data for each of a plurality of local regions included in the image data in order to calculate the evaluation value of the image data at each focal position, and each evaluation data is calculated based on the plurality of evaluation data. An evaluation value of the focal position is calculated. Details of generation of evaluation data and calculation of evaluation values will be described later. The local region includes at least two regions, a first region 46 having the same size corresponding to the main target range 44 and a second region 47 having the same size corresponding to the peripheral range 45. The evaluation value calculation unit 18 uses, as the image data at each focal position, for example, information corresponding to a gradation having a linear or nonlinear relationship with the luminance of the subject 40 imaged at each pixel position. Calculates an index corresponding to the contrast distributed for each of the plurality of local areas in the entire image area 43, and uses the pixel data of the calculation result. The pixel data used here may be, for example, a monochrome signal (grayscale) that is linearly proportional to the luminance. Alternatively, only the G component data extracted from the image data composed of the three primary colors RGB, or linearly encoded data such as gamma characteristics may be used.

  In order to generate evaluation data, the evaluation value calculation unit 18 first performs frequency filter processing on the pixel data of each local region of each image data using a spatial frequency filter having a predetermined bandpass characteristic (step S1). S3) Only local relative light and dark changes centering on the edge portion are calculated as data. Then, based on the data after the frequency filter processing, a score component (score component serving as a sharpness index) having a high correlation with the sharpness is calculated for each of the plurality of local regions. For example, a conversion process is performed for converting the data of each local region into a numerical value representing the magnitude of contrast with an amplitude component. As a result, the first score component for the first region 46 and the second score component for the second region 47 of each local region are calculated.

  In such a spatial filtering process, for example, a low-cut filter removes contrast differences having low correlation with sharpness, and a high-cut filter suppresses the influence of low noise components related to focus determination. Alternatively, in this spatial filtering process, the data shown in FIG. 7 can be obtained by performing a calculation for extracting only the amplitude, excluding the average, as a minimum simple process for the data shown in FIG. Note that the frequency filter process is one of the techniques for detecting local contrast from image data, and any technique other than the frequency filter process may be applied as long as the local contrast can be detected.

Here, in general, in order to convert the amplitude component into a numerical value indicating the magnitude of the contrast, for example, it can be calculated by a mean square or a difference between the maximum value and the minimum value. The average processing using the absolute value data as in Figure 8, each value Y _{i (i} is an integer number of data n), when the average value of each value and Y _mean, the following equation ( 1).

  However, with the data simply averaged in this way, the peripheral in-focus state in FIG. 8A has a higher absolute value as a whole than the target in-focus state in FIG. In some cases, it is determined that the contrast is higher in the peripheral in-focus state in FIG.

Accordingly, the evaluation value calculation unit 18 additionally performs a weighting filter process before the conversion process into a numerical value representing contrast (step S5). Here, a convolution operation that applies the above weighting filter is performed on the amplitude component A _i after the frequency filter (for example, on the level | Y _i −Y _mean | absolute value as described above). , Can be expressed by the following formula (2).

  At this time, in the same local region, there is data (target data) corresponding to the target 41 in the first region 46 with respect to the main target range 44 of the weighting filter, and within the second region 47 with respect to the peripheral range 45. When there is data corresponding to the peripheral portion 42 (peripheral portion data), the level of the data is lowered as a whole by the convolution operation, but the peak of the target data is maintained (see FIG. 9B). On the other hand, in the same local region, when there is no data (target data) corresponding to the target 41 in the first region 46 with respect to the main target range 44 of the weighting filter, the overall data level is determined by the convolution operation. That is, the overall level of the image focused on the peripheral portion 42 of the target 41 is suppressed (see FIG. 9A).

  Therefore, when the above conversion processing is performed on the data after such a convolution operation (step S6), the first score component calculated for the first region 46 of each local region is weighted. For the second score component calculated for the second region 47 of each local region, the main target weighting factor set in the main target range 44 of the filter contributes, and the peripheral range 45 of the weighting filter. Evaluation data calculated (integrated) so as to contribute to the peripheral weighting coefficient set in (1) is obtained.

  Then, the evaluation value calculation unit 18 obtains one evaluation value that serves as an index for determining the in-focus position from a certain range of interest based on the evaluation data of a plurality of local regions generated using the convolution operation. Statistical processing for calculation is performed (step S7). For example, the evaluation value calculation unit 18 acquires the maximum value from a plurality of evaluation data over the entire image area of each image data and sets it as the evaluation value. At this time, since the evaluation data of the target in-focus state as shown in FIG. 9B has a peak at a position corresponding to the target 41, the evaluation data in the peripheral in-focus state as shown in FIG. In comparison, a high evaluation value can be obtained.

  If the evaluation value calculation unit 18 can maintain a resolution that does not impair the maximum value of the evaluation value when focused on the target 41 in the stage before statistical processing as shown in FIG. By performing image data reduction processing (pixel size reduction processing) after the processing (step S3), the subsequent calculation amount may be reduced (step S4).

  The focus position determination unit 19 analyzes the evaluation value calculated by the evaluation value calculation unit 18 at each of a plurality of different focus positions, and detects the focus position at which the maximum evaluation value is obtained as the focus position (step) S8). For example, among the evaluation data of each focal position, when the evaluation data in the target in-focus state in FIG. 9B has a peak at a position corresponding to the target 41 and the maximum evaluation value is obtained, this target The focus position (focus position focused on the target 41) when the image data in focus is obtained is detected as the focus position.

  In the present embodiment, the example in which the spatial filtering process using the frequency filter and the additional filtering process using the weighting filter are performed separately has been described, but the present invention is not limited to this example. For example, in another embodiment, the convolution operation may be performed together with the spatial filtering process using a combined spatial filter in which a frequency filter and a weighting filter are combined in advance.

  In the present embodiment, a positive main object weighting coefficient is contributed to the first score component in the first area 46 by convolution using a weighting filter, and the second area 47 second Although an example in which a negative peripheral weight coefficient is contributed to the score component has been described, the present invention is not limited to this example. For example, a filter that contributes a positive main object weighting factor to the first score component in the first region 46 and a negative peripheral weighting factor to the second score component in the second region 47. The evaluation data of each local region may be calculated by separately providing a filter to be contributed and adding each filter processing result. Alternatively, a positive main object weight coefficient is made to contribute to the first score component of the first area 46, and a negative peripheral weight coefficient is made to the second score component of the second area 47. As long as it can contribute, methods other than the convolution calculation using filters, such as a weighting filter, may be applied.

  As described above, according to the present embodiment, the autofocus device of the imaging system 1 is a device that determines the focus control state for the subject 40 according to the contrast of the image of the subject 40. The autofocus device includes an evaluation value calculation unit 18 and a focus position determination unit 19. The evaluation value calculation unit 18 calculates an evaluation value used for determining the in-focus position for each of a plurality of image data obtained by imaging the subject 40 at each of a plurality of focal positions. The focus position determination unit 19 determines the focus position based on a plurality of evaluation values calculated for each of the plurality of focus positions. Further, the evaluation value calculation unit 18 calculates the first score component calculated for the first region 46 having a size including the focus target 41 of the subject 40 for each of the plurality of local regions included in each image data. And the second score component calculated for the second region 47 having a size including at least a part of the peripheral portion 42 of the target 41 around the first region 46, Based on a plurality of evaluation data generated for each of a plurality of local regions in each image data, generating evaluation data that contributes with a positive weight and contributes to a second score component with a negative weight To calculate an evaluation value.

  For example, the evaluation value calculation unit 18 gives a positive coefficient corresponding to a positive weight to the first area 46 and a negative coefficient corresponding to a negative weight for the pixel data of each local area in each image data. It is good to perform the convolution operation using the spatial weighting filter which gives to the 2nd area | region 47, and to generate evaluation data based on the result of the convolution operation.

  Further, for example, the evaluation value calculation unit 18 performs a frequency filter process using a spatial frequency filter having a predetermined bandpass characteristic on the pixel data of each local region in each image data, and after the frequency filter process A conversion process for converting the amplitude component into a value representing the magnitude of the amplitude component may be performed based on the data, and the first score component and the second score component may be calculated based on the result of the conversion process. Alternatively, the evaluation value calculation unit 18 may perform a convolution operation using a combined spatial filter in which a weighting filter and a frequency filter are combined in advance.

  Further, for example, the evaluation value calculation unit 18 uses, as pixel data in each image data, information corresponding to a gradation having a linear or nonlinear relationship with the luminance of the subject 40 imaged at each pixel position, An index corresponding to contrast in which gradation is distributed for each of a plurality of local regions in each image data is calculated.

  For example, the evaluation value calculation unit 18 may set the maximum value among a plurality of evaluation data as the evaluation value for each image data, and the focus position determination unit 19 may perform a plurality of evaluations for each of the plurality of focus positions. The focal position where the maximum value is obtained from the values may be determined as the in-focus position.

  With such a configuration, the autofocus device detects the in-focus position where the target 41 is in focus even when the focus target 41 and the peripheral portion 42 of the target 41 are in different focus positions. can do. For this reason, for example, even when an object having a target 41 having a smaller area than the peripheral portion 42 is photographed from the distal end side of the projection portion, such as an electronic component having a projecting member such as an electrode, the tip of the projection portion is focused. The in-focus position can be detected. Alternatively, even when an image of a component having a recess is taken from the front of the recess, it is possible to detect the in-focus position where the deep portion of the recess is in focus. In addition, the autofocus device can detect the in-focus position of the target 41 quickly and reliably without providing complicated processing and configuration even when the position of the target 41 varies depending on the timing of shooting. Therefore, when this autofocus device is employed in a machine vision system or the like, it is possible to improve the processing speed of photographing and inspection, reduce the processing load, and further reduce the equipment cost.

  Further, according to the present embodiment, the evaluation value calculation unit 18 performs a pixel size reduction process on the data after the frequency filter process, and based on the data after the reduction process, the first score component and the second score Components may be calculated.

  Thus, the autofocus device can improve the processing speed in the autofocus operation and reduce the processing load.

  In addition, according to the present embodiment, the autofocus device of the imaging system 1 includes the area setting unit 16 that can change at least one of the size of the first area 46 and the size of the second area 47. It is good to change at least one of a positive main weighting factor (positive weight) contributing to the first score component and a negative peripheral weighting factor (negative weight) contributing to the second score component. You may provide the contribution rate setting part 20 which can do.

  Accordingly, the autofocus device can make detailed settings for each subject 40 (for example, an inspection product in a machine vision system or the like) for the target 41 to be focused on in the subject 40.

  In the present embodiment, the configuration in which the weighting filter generation unit 17 generates the weighting filter by uniformly setting the negative weighting coefficient for the peripheral range 45 around the main target range 44 has been described. It is not limited to. In another embodiment, the weighting filter generation unit 17 has a third region as a dead zone having a weight of 0 between the first region 46 and the second region 47 in each local region of each image data. There is a dead band range in which a zero weighting factor is set between a main target range 44 in which a positive main weighting factor is set and a peripheral range 45 in which a negative peripheral weighting factor is set. Thus, a weighting filter may be generated. According to such another embodiment, for example, when shooting the subject 40 in which the tip of the projection target 41 has a very small area when viewed from the shooting side and the area increases from the tip to the base end. Even when shooting the subject 40 in which the deep portion of the target 41 in the concave portion has a very small area when viewed from the photographing side and the area increases from the deep portion to the edge of the concave portion, focusing on the tip of the target 41 other than the deep portion is possible. By avoiding this, the in-focus position of the target 41 can be reliably detected. Alternatively, the weighting filter generation unit 17 divides the peripheral range 45 from the inner side (side closer to the target 41) to the outer side (side farther from the target 41) into multi-step regions, and reduces the negative range with respect to the outer step region. A weighting filter may be generated by setting a weighting factor.

  Further, in the present embodiment, an example has been described in which the focus position determination unit 19 determines, as the focus position, the focus position at which the maximum value is obtained from a plurality of evaluation values for each of the plurality of focus positions. The invention is not limited to this example. For example, in another embodiment, the in-focus position determination unit 19 calculates one or more local peak positions from among a plurality of evaluation values for a plurality of focal positions, and a local peak position that is most suitable for a predetermined appropriate distribution. The focal position at which is obtained may be determined as the in-focus position. In this case, the imaging system 1 may be configured to include an appropriate distribution acquisition unit 21 in order to acquire a predetermined appropriate distribution indicating information on an appropriate distribution range or polarity with respect to the in-focus position of the target 41. The information on the appropriate distribution range and polarity is information for evaluating an optimum one of one or more local peak positions seen over a plurality of focal positions. For example, the “distribution range” is a range that can be assumed in advance, such as a short distance side or a long distance side, among a plurality of focal positions. Thus, even when there are two or more protrusions or recesses having different focal positions in the subject 40, it is possible to select the optimum target 41 and achieve the focus.

  In the present embodiment, the region setting unit 16 has described the configuration in which the main target range 44 having a size including the target 41 and the peripheral range 45 around the main target range 44 are described. It is not limited to the configuration. For example, in another embodiment, the region setting unit 16 may set a plurality of regions including at least a part of the peripheral portion 42 of the target 41 in addition to the main target range 44 having a size including the target 41. . Further, the region setting unit 16 is not limited to a rectangle as the shape of the main target range 44 and the peripheral range 45, and can also set other shapes as long as it is a region corresponding to the target 41 and its peripheral portion 42. .

  In addition, the present invention can be appropriately changed without departing from the gist or concept of the invention that can be read from the claims and the entire specification, and an autofocus device and an autofocus program that involve such a change are also included. It is included in the technical idea of the present invention.

DESCRIPTION OF SYMBOLS 1 Imaging system 10 Optical system 11 Optical system drive part 12 Imaging part 13 Imaging drive part 14 A / D conversion part 15 Image processing part 16 Area | region setting part 17 Weight filter generation part 18 Evaluation value calculation part 19 Focus position determination part 20 Contribution Rate setting unit 21 Proper distribution acquisition unit 30 Control unit 31 Storage unit 32 Recording medium 33 Operation unit 34 Display unit 40 Subject 41 Target 42 Peripheral part 43 Total image area 44 Main target range 45 Peripheral range 46 First area 47 Second area

Claims (13)

An autofocus device that determines a focus control state for the subject according to a contrast of an image of the subject,
An evaluation value calculation unit that calculates an evaluation value used to determine the in-focus position for each of a plurality of images obtained by imaging the subject at each of a plurality of focal positions;
A focus position determination unit that determines a focus position based on the plurality of evaluation values calculated for each of the plurality of focus positions;
The evaluation value calculation unit includes, for each of a plurality of local regions included in each image, a first score component calculated for a first region having a size including a focus target of the subject, Using a second score component calculated for a second region having a size including at least a part of the periphery of the target around one region, and contributing to the first score component with a positive weight And generating evaluation data calculated to contribute to the second score component with a negative weight, and based on the plurality of evaluation data generated for each of the plurality of local regions in each image An autofocus device characterized by calculating an evaluation value.   The evaluation value calculation unit gives a positive coefficient corresponding to the positive weight to the first region and a negative coefficient corresponding to the negative weight for the pixel data of each local region in each image. The autofocus device according to claim 1, wherein a convolution operation using a spatial weighting filter that gives a coefficient to the second region is performed, and the evaluation data is generated based on a result of the convolution operation.   The evaluation value calculation unit performs frequency filter processing using a spatial frequency filter having a predetermined bandpass characteristic on the pixel data of each local region in each image, and the data after the frequency filter processing is performed. The conversion processing for converting the amplitude component into a value representing the magnitude of the amplitude component is performed, and the first score component and the second score component are calculated based on the result of the conversion processing. 2. The autofocus device according to 2.   4. The autofocus device according to claim 3, wherein the evaluation value calculation unit performs the convolution operation using a combined spatial filter in which the weighting filter and the frequency filter are combined in advance.   The evaluation value calculation unit uses, as pixel data in each image, information corresponding to a gradation having a linear or nonlinear relationship with luminance of the subject imaged at each pixel position, 5. The autofocus device according to claim 1, wherein an index corresponding to a contrast distributed for each of the plurality of local regions in each image is calculated.   The evaluation value calculation unit performs a pixel size reduction process on the data after the frequency filter process, and calculates the first score component and the second score component based on the data after the reduction process. The autofocus device according to claim 3.   The said evaluation value calculation part uses the said weighting filter which has a 3rd area | region as a dead zone whose weight is 0 between the said 1st area | region and the said 2nd area | region. Autofocus device.   The autofocus device according to any one of claims 1 to 7, wherein the evaluation value calculation unit sets a maximum value among the plurality of evaluation data as the evaluation value for each image.   The focus position determination unit determines a focus position at which a maximum value is obtained from the plurality of evaluation values for each of the plurality of focus positions as a focus position. The autofocus device according to claim 1. An appropriate distribution acquisition unit for acquiring information on an appropriate distribution range or polarity with respect to the focal position;
The focus position determination unit calculates one or more local peak positions from the plurality of evaluation values with respect to the plurality of focus positions, and the focus at which the local peak position most suitable for the appropriate distribution is obtained. The autofocus device according to claim 1, wherein the position is determined as a focus position.   The autofocus device according to any one of claims 1 to 10, further comprising an area setting unit capable of changing at least one of a size of the first area and a size of the second area. .   2. A contribution rate setting unit capable of changing at least one of the positive weight contributing to the first score component and the negative weight contributing to the second score component. The autofocus device according to any one of 11 to 11.   An autofocus program for causing a computer to function as the autofocus device according to any one of claims 1 to 12.

Images