JP6087754B2 - Lens tilt detector - Google Patents

Description

  The present invention relates to a lens tilt detection device for detecting a tilt direction and a tilt amount of a lens unit of a camera module for a mobile phone, for example.

  In camera modules for mobile phones, strict positioning adjustment accuracy is required when the lens group is incorporated into the optical unit due to the downsizing of the image sensor and the diversification of aspherical lenses. After the lens group is incorporated in the lens unit, it becomes an optical unit on which an actuator for driving the lens is mounted. At this time, the lens unit is often tilted and adjusted with respect to the reference plane of the optical unit, and the lens group is tilted with respect to the optical axis of the optical unit, and is mounted without adjusting the image sensor. Even in such a case, a camera module having desired performance cannot be obtained. Therefore, it is necessary to measure the tilt of the lens group in the state of the optical unit and adjust the generated tilt.

  As a method of measuring the lens tilt amount, by irradiating the lens unit of the optical unit with ring illumination from the ring light source, the light reflected on the upper surface side of the first lens of the lens unit forms a virtual image in the back of the lens unit. The light reflected from the lower surface side of the first lens forms an image in front of the lens unit. For this reason, when the lens unit is observed with a camera arranged on the optical axis connecting the center of the ring illumination and the center of the lens unit, two different sizes of ring illumination images can be confirmed on concentric circles. At this time, when the first lens is tilted, the two ring illumination images are shifted in the lens tilt direction. A lens unit, ring illumination, and an imaging camera are arranged on a central axis that is perpendicular to the imaging sensor surface and passes through the center thereof, and an upper ring image and a lower ring image are detected by imaging the upper surface of the lens unit with this imaging camera. When the lens is tilted in the X-axis direction, the ring image is shifted in the vertical direction on the paper surface, and when the lens is tilted in the Y-axis direction, the ring image is shifted in the horizontal direction on the paper surface. There is a method in which the lens tilt is detected from the direction and amount in which the center of gravity of each of the two ring images is shifted from the central axis, and is adjusted using this tilt amount.

  At this time, in the image processing technique for detecting the two ring images and estimating the position of the center of gravity, a method of detecting an edge from the density information of the ring and detecting a circle, a desired reference pattern image prepared in advance A method of detecting a circle by pattern matching using, etc. can be considered.

  In Patent Document 1, a desired shape is detected with high accuracy by calculating a center position in an image and detecting edge information of the outer shape.

JP 2011-1003003 A

  When detecting the edge of a circle, noise such as dust adhering to the lens or illumination and the sharpness of the edge have a large effect on the detection accuracy of the circle, and if the edge sharpness is insufficient, When it is difficult to obtain sufficient detection accuracy, and there are two ring-shaped imaged images as inspection objects as in this case, and two ring-shaped imaged images overlap, a circle with high accuracy is obtained. Is difficult to detect.

  Even in the method using pattern matching, the state of the two ring-shaped formed images is indefinite, and it is difficult to detect the position with high accuracy when the ring-shaped formed images overlap.

  Accordingly, an object of the present invention is to provide a lens tilt detection device that can accurately detect the tilt direction and the tilt amount of a lens with respect to an optical unit even when two ring-shaped formed images overlap.

In order to solve the above problems, a lens tilt detection device of the present invention is
Illumination light from a ring light source whose light emitting surface is substantially ring-shaped is applied to a plurality of lens surfaces of the lens unit of the optical unit, and at least two or more of the ring light sources generated by reflected light from the plurality of lens surfaces A lens tilt detection device that detects a tilt amount and a tilt direction of the lens unit based on a relative shift amount of each center position of a ring-shaped formed image,
An imaging means for capturing the ring-shaped image;
An image processing device for processing an image signal output from the imaging means,
The image processing apparatus includes:
A contour extraction unit for extracting a contour line of the ring-shaped image;
An area calculation unit for calculating the area of the figure surrounded by the contour line;
A ring detection unit that detects the center of a ring-shaped imaging image surrounded by the contour line based on the contour line and the area;
The center of the large ring is detected based on the outer contour line of the large ring in the imaged image and the area of the figure surrounded by the outer contour line. A ring detection correction unit that detects the center of the small ring based on the inner contour line of the small ring having a smaller diameter than the ring and the area of the figure surrounded by the inner contour line is provided.

  According to the present invention, since the center of the outer contour line of the large ring and the center of the inner contour line of the small ring having a smaller diameter than the large ring are detected, one of the contour lines of the two ring-shaped imaging images is detected. Even when the portions overlap, it is possible to accurately detect the center of two ring-shaped contour lines (the center of a circle or a rectangle) and accurately detect the tilt direction and the tilt amount.

It is a schematic block diagram of the lens tilt detection apparatus of Embodiment 1 of this invention. 2A and 2B are schematic views of a ring-shaped light source and a lens structure in the first embodiment. FIGS. 3A and 3B are a conceptual diagram and an observation image for explaining the principle of the present invention. (B), (C), (D), and (E) in FIG. 3-2 are conceptual diagrams for explaining the principle of the present invention. 4A, 4 </ b> B, and 4 </ b> C are conceptual diagrams illustrating lens tilt inclination and ring image shift in the first embodiment. 5A, 5 </ b> B, and 5 </ b> C are conceptual diagrams illustrating the lens tilt inclination and the shift of the ring image when the holder is displaced in the first embodiment. 6A, 6 </ b> B, and 6 </ b> C are conceptual diagrams of camera module assembly using the lens tilt adjustment method according to the first embodiment. FIG. 3 is an image processing flowchart in the first embodiment. FIGS. 8A and 8B are processed image diagrams according to the first embodiment of the present invention. (C), (D), (E), and (F) in FIG. 8-2 are processed image diagrams according to the first embodiment of the present invention. It is a figure which shows the overlapping state of the ring-shaped image formation image in Embodiment 1 of this invention. FIG. 10 is an image processing flowchart according to the second embodiment. It is a figure explaining operation | movement of the circle | round | yen detection part using a template. FIGS. 12A and 12B are diagrams illustrating a schematic configuration of the lens tilt detection apparatus according to the third embodiment and a ring-shaped light source. FIG. 10 is a schematic diagram of a band-shaped image formed in the third embodiment.

  Hereinafter, the present invention will be described in detail with reference to the illustrated embodiments.

(Embodiment 1)
FIG. 1 shows a schematic configuration of a lens tilt detection apparatus 1 according to Embodiment 1 of the present invention.

  The lens tilt detection device 1 includes a ring light source 2, an imaging camera 3 as an example of an imaging unit, a signal processing unit 8, an image processing unit 5, an arithmetic processing unit 6, a display unit 7, and the like.

  The ring light source 2 is a ring-shaped light source having a substantially circular light emitting surface. FIG. 2A is a schematic view seen from the light irradiation surface 21 side of the ring light source 2. The irradiation surface 21 is a circular ring-shaped fluorescent tube. In addition to this, as a ring light source, LEDs (light emitting diodes) that are point light sources are provided on the circumference at the same interval, or a diffuser plate that is a circular ring-shaped surface light source is attached thereto, or A light bulb may be used. Further, a type in which EL (Electro Luminescence) light emitters are continuously formed in a circumferential shape may be used.

  The light from the ring light source 2 irradiates the optical unit 4 as shown in FIG. As shown in FIG. 2B, the optical unit 4 includes a lens group 41 (consisting of three lenses L1, L2, and L3 in FIG. 2B) and a lens to which the lens group 41 is attached. It comprises a holder 42 and an actuator 43 for driving the lens holder 42 up and down. The lens group 41 and the lens holder 42 constitute a lens unit 45.

  The light from the ring light source 2 is reflected by the upper surface S1, which is the lens surface of the first lens L1 of the lens group 41, as shown in FIGS. 1, 2B, and 3A. An image and an image reflected by the lower surface S2, which is a lens surface, are formed. The imaging camera 3 causes the lens unit 31 to make this image incident on the light receiving element 32. The light receiving element 32 is a light receiving element generally used in an imaging camera, such as a CCD (Charge Coupled Devise) or a CMOS (Complementary Metal Oxide Semiconductor).

  On the imaging surface of the light receiving element 32, an image of the ring light source 2 reflected by the upper surface S1 and the lower surface S2 of the first lens L1 is formed. The shape of the image of the ring light source 2 is determined based on the positional relationship between the light receiving element 32, the lens unit 31, the ring light source 2, the first lens L 1, and the specification settings of the imaging camera 3.

  The output signal from the light receiving element 32 is sent to the signal processing unit 8, which generates a video signal corresponding to the light intensity of the received image and transmits it to the image processing unit 5. The image processing unit 5 converts the received video signal into an image signal. The signal processing unit 8 may be provided in the light receiving element 32 or in the arithmetic processing unit 6. The image signal is transmitted to the arithmetic processing unit 6 composed of a microcomputer, and the center of gravity of the light intensity of the ring image is obtained by calculation.

  The arithmetic processing unit 6 calculates the tilt angle of the first lens L1 by performing arithmetic processing on the shift amount of the center of the formed image based on the obtained output signal of the light receiving element 32.

  The image processing unit 5 and the arithmetic processing unit 6 constitute an image processing apparatus. This image processing apparatus may include the signal processing unit 8.

  The image and the calculation result obtained by the image processing apparatus are displayed on the display unit 7. The display unit 7 employs a liquid crystal display or a CRT display.

  In general, the lens group 41 including the plurality of first to third lenses L1, L2, and L3 is assembled with tilt adjustment performed with high accuracy. Among these, in particular, the first lens L1 is often a lens that greatly affects the optical performance. Therefore, the tilt of the first lens L1 greatly affects the optical performance of the lens group 41.

  Therefore, a detection method for detecting the tilt angle of the first lens L1 used as an index for adjusting the tilt of the lens group 41 will be described in detail below. This detection method obtains the state of eccentricity that occurs when the first lens L1 is tilted, and detects the tilt angle of the first lens L1.

  The relationship between the eccentricity and the image will be described with reference to FIGS.

  First, when the first lens L1 is a biconvex lens, the light emitted from the ring light source 2 is divergently reflected by the upper surface S1 of the first lens L1, and a virtual image is formed at the point P1. Next, the light transmitted through the upper surface of the first lens L1 and converged and reflected by the lower surface forms a real image at the point P2. When the image formed at these points P1 and P2 is observed by the imaging camera 3, the virtual image at the point P1 is at the back and the real image at the point P2 is at the front with respect to the upper surface S1 of the first lens L1.

  At this time, it is desirable that the depth of field of the imaging camera 3 is such that only points P1 to P2 are observed simultaneously. However, even if the depth of field is not sufficient from the points P1 to P2, there is no problem if an image capable of measuring the center of the ring image is obtained.

  When the first lens L1 is a convex meniscus lens as viewed from the illumination side as shown in FIG. 3B, the light emitted from the ring light source 2 first diverges on the upper surface S1 of the first lens L1. Reflected to form a virtual image at the point P1. Next, the light transmitted through the upper surface of the first lens L1 and divergently reflected from the lower surface forms a virtual image at the point P2. When the images formed at these points P1 and P2 are observed by the imaging camera 3, the virtual images at the points P1 and P2 can be made deeper with respect to the upper surface S1 of the first lens L1.

  The positional relationship between the points P1 and P1 in the direction of the optical axis m is determined by the curvature of the lens surface, but is not important in the present invention as described later.

  When the first lens L1 is a concave meniscus lens as seen from the illumination side as shown in FIG. 3C, the light emitted from the ring light source 2 first converges on the upper surface S1 of the first lens L1. Reflected to form a real image at the point P1. Next, the light transmitted through the upper surface of the first lens L1 and converged and reflected by the lower surface forms a real image at the point P2. When the images formed at these points P1 and P2 are observed by the imaging camera 3, the images at the points P1 and P2 can be positioned above the upper surface S1 of the first lens L1.

  As shown in FIG. 3D, when the first lens L1 is a biconcave lens, the light emitted from the ring light source 2 is first reflected and reflected by the upper surface S1 of the first lens L1. A real image is formed on P1. Next, the light transmitted through the upper surface S1 of the first lens L1 and divergently reflected at the lower surface S2 forms a virtual image at the point P2. When an image formed at these points P1 and P2 is observed by the imaging camera 3, the image at the point P1 can be on the upper side and the image at the point P2 can be located behind the upper surface S1 of the first lens L1. In this way, a reflection elephant from the front and back surfaces of one lens is formed on the optical axis m regardless of the shape of the lens.

  In addition, even when there are a plurality of lenses other than the first lens L1, an image by reflected light from each surface is generated, and an image within the depth of field of the imaging camera 3 can be observed.

  For example, as shown in FIG. 3E, when the first lens L1 is a biconvex lens and the second lens L2 is a concave meniscus lens, an image of the ring light source 2 is generated at points P1 to P4. At this time, the depth of field W of the imaging camera 3 may be designed so that all images of the points P1 to P4 can be observed at a time, but a design capable of observing at least two images is desirable. (In (E) of FIG. 3-2, the depth of field W is a design example in which two points P1 and P2 can be observed at one time.)

  Next, FIGS. 4A, 4 </ b> B, and 4 </ b> C show how the ring illumination images are tilted. In each of FIGS. 4A, 4B, and 4C, an image obtained by the light receiving element 32 is shown in the upper part of the drawing, and the inclination state of the lens holder 42 and the first lens L1 is shown in the lower part of the drawing.

  FIG. 4A shows a case where the first lens L1 is not tilted. At this time, ring images R1 and R2 (centers P1 and P2) of the ring-shaped light source 2 with respect to the lens holder 42 appear in the image obtained by the light receiving element 32. When the lens holder 42 is rotated in the direction of the arrow W1 from this state, as shown in FIG. 4B, the center positions P1 and P2 of the image of the ring light source 2 are observed shifted to the left and right in the figure. Next, as shown in FIG. 4C, when the lens holder 42 is rotated in the direction of the arrow W2, the center positions P1 and P2 of the image of the ring light source 2 are observed shifted in the opposite directions in the figure.

  As described above, when the first lens L1 is tilted, the lens surfaces S1 and S2 are decentered, and the decentered state can be detected from the relative positional relationship of the obtained image. Thus, the tilt angle of the first lens L1 can be obtained by reverse calculation.

  That is, by detecting the shift direction of the center positions P1 and P2 of the images R1 and R2 of the ring light source 2 from the image of the light receiving element 32, the tilt direction of the first lens L1 can be detected, and the image of the ring light source 2 can be detected. By detecting the distance between the center positions P1 and P2, the tilt angle of the first lens L1 can be detected. That is, the tilt of the first lens L1 can be corrected by adjusting the images R1 and R2 of the ring light source 2 to be concentric circles and overlapping the center positions P1 and P2. This is effective for a lens mounted on the lens unit 45 where the lens outer shape cannot be seen because the tilt amount is detected by detecting the relative positional relationship between two or more images of the lens.

  Further, as shown in FIG. 5A, the ring images R1 and R2 are concentric without being affected even when the aperture AP of the lens holder is formed to be shifted from the lens center. Thus, when the centers P1 and P2 overlap, it can be determined that there is no tilt.

  Next, as shown in FIG. 5B, when the lens holder 42 is rotated in the direction of the arrow W1, the center positions P1 and P2 of the image of the ring light source 2 are left and right in the figure regardless of the position of the aperture AP. Observed by

  Next, as shown in FIG. 5C, when the lens holder 42 is rotated in the direction of the arrow W2, the center positions P1 and P2 of the image of the ring light source 2 are observed shifted in the opposite directions in the figure. The As described above, when the first lens L1 is tilted, the lens surfaces S1 and S2 are decentered, and the decentered state can be detected from the relative positional relationship of the obtained image. Thus, the tilt angle of the first lens L1 can be obtained by reverse calculation.

  In other words, the tilt direction of the first lens L1 can be detected by detecting the shift direction of the center positions P1 and P2 of the images R1 and R2 of the ring light source 2 from the image formed by the light receiving element 32, and the image of the ring light source 2 can be detected. By detecting the distance between the center positions P1 and P2, the tilt angle of the first lens L1 can be detected.

  In the above description, the ring-shaped image formed only by the first lens L1 has been described. However, in the case of a lens unit including two or more lenses, two or more ring-shaped image formed are formed. Even when the image is obtained, the state where the ring-shaped image is concentrically is a state where there is no lens tilt, and the tilt angle is detected by the sum of the differences between the center positions of the respective ring-shaped image images. The tilt direction can be recognized by detecting the direction in which each ring-shaped formed image is decentered due to the occurrence.

  Next, a description will be given of a flow in which the lens tilt adjustment method according to the first embodiment is implemented in the imaging camera module assembling apparatus using the lens tilt detection apparatus 1. This flow is executed by an instruction from the program in the arithmetic processing unit 6. This program is stored in a ROM or RAM (not shown) in the arithmetic processing unit 6.

  As shown in FIG. 6A, first, the image sensor 9 for the camera module is set in an assembly device (not shown), and the image sensor with respect to the reference axis m is tilted by the tilt sensor 10 disposed on the assembly reference axis m of the assembly device. The tilt amount α1 of the light receiving surface 9 is measured. k is a normal line of the light receiving surface.

  Next, as shown in FIG. 6B, the optical unit 4 for the camera module is set in the assembling apparatus, and the lens tilt detection apparatus 1 disposed on the assembling reference axis m of the assembling apparatus is used with respect to the reference axis m. The tilt amount α2 of the lens group 41 is measured.

According to the above measurement results,
(I) As the camera module, the image sensor 9 is corrected from the tilt α1 to the lens tilt α2 and the relative position is adjusted with respect to the optical unit 4 and then assembled to assemble the camera module, or (ii) the optical unit 4 is assembled. After correcting the tilt α2 to the tilt α1 of the image sensor 9 and adjusting the relative position to the image sensor 9, the camera module is assembled by assembling, or (iii) the tilt amount α1 of the image sensor 9 with respect to the reference axis m Is corrected so that the tilt amount α1 of the lens unit 41 becomes 0, the posture of the optical unit 4 is corrected, and the relative position adjustment between the corrected image sensor 9 and the optical unit 4 is performed. After that, the camera module is assembled by assembling.

  With respect to the camera module assembled by any of the above methods, the camera module is arranged perpendicular to the reference axis m as shown in FIG. 6C (when the figure is assembled by the method of (i) above). The image pickup chart C is picked up by the image pickup sensor 9, the image pickup characteristic is confirmed, and the product inspection is performed.

  A flow of basic processing in the image processing device (5, 6) of the lens tilt detection device 1 will be described with reference to FIG.

  First, two ring-shaped formed images as shown in FIG. 8-1 (A) are captured from the imaging camera 3 as the imaging means (step # 1 in FIG. 7). First, unnecessary background information is cut and removed by the mask processing unit (step # 2).

  Next, noise information of the image is cut by the noise cut processing unit (step # 3). In this noise cut processing, general noise cut processing such as filtering processing such as Gaussian, blur, and median, opening, and closing processing is performed to generate a desired image suitable for circle detection.

  Next, as shown in (B) of FIG. 8A, after the binarization processing unit performs binarization processing in which the detection candidate is a white image with a predetermined luminance value as a threshold value ( In step # 4), mask processing is performed by the mask processing unit (step # 5), and the contour extraction unit that extracts the contour of the binarized image extracts the contour in the binarized image (step # 6).

  Next, it is determined whether or not the inspection of all the contour images has been completed (step # 7). If not, the process proceeds to step # 8. If YES, the process proceeds to step # 9.

  In step # 8, for the contour for which the inspection of the contour image has not been completed, the contour extracted by the contour extraction unit is filled so as to be a white image, the area is calculated, and the process returns to step # 7.

  In step # 9, as shown in (C), (D), (E), and (F) of FIG. 8-2, circle detection as an example of ring detection is performed on all individually extracted contour images. Do. Here, only the contour within the desired area range is extracted, and the center of gravity coordinates are calculated from the filled contour image in the circle detection unit as an example of the ring detection unit (step # 9). The center of gravity (x, y) of the contour image is calculated by dividing the first moment in a general image in image processing by the calculated contour area.

  Next, in the circle detection correction unit as an example of the ring detection correction unit, each circle is estimated from the outer contour line of the large ring and the inner contour line of the small ring having a smaller diameter than the large ring, respectively. The center coordinates are estimated, and each circle detection is corrected (step # 10). More specifically, in the binarized image in the binarization processing unit, the center of gravity position information of the outer contour of the large ring and the inner contour of the small ring having a smaller diameter than the large ring is used as an example of the ring estimation unit. The circle estimation unit estimates a circle by shape detection using, for example, Hough transform, estimates center coordinates, and corrects circle detection. In addition to the method described above, the circle may be estimated by creating a template image and calculating the similarity.

  Using these methods, correction of circle detection is performed to extract the inner and outer contours of the large circle and the small circle, and the center or centroid (x1, y1) of the large ring and the center or centroid of the small ring ( x0, y0) is calculated (step # 11). Further, by obtaining the average value of the center of gravity positions of the large ring and the small ring, it becomes possible to calculate the center of gravity of the large and small rings with high accuracy.

  As described above, as a reason for using the circle detection correction unit as an example of the ring detection correction unit, as the lens tilt amount increases, as shown in FIG. 9, the inner contour of the great circle and the outer contour of the small circle overlap. It may be mentioned that there are cases. In this case, in the case of only circle detection by contour extraction and area calculation, the contours inside the large circle and the contour outside the small circle are integrated, making it impossible to detect the circle and calculate the center of gravity. Further, an error becomes very large only by circle estimation by Hough transform or pattern matching using only the diameter of the circle to be detected as a parameter. Therefore, if circle estimation is performed by the circle detection correction unit using the center of gravity information of the outer contour of the large ring and the inner contour of the small ring, the center of gravity position of the outer contour of the large ring and the inner contour of the small ring is increased. It becomes possible to calculate with accuracy.

  As described above, the image processing device (5, 6) calculates the two ring-shaped formed images in the image, that is, the position of the center of gravity of the ring image with high accuracy. It is possible to calculate the amount of deviation of the ring with high accuracy, and to adjust the lens tilt with higher accuracy.

(Embodiment 2)
FIG. 10 is a flowchart of the image processing unit of the lens tilt detection apparatus according to the second embodiment.

  The lens tilt detection apparatus according to the second embodiment is different from the first embodiment only in the configuration of the image processing apparatus illustrated in FIG. 10. The other elements are described in detail with reference to the drawings and the description of the first embodiment. Description is omitted.

  Further, steps # 1 to # 8 in the flowchart of FIG. 10 showing the processing of the image processing apparatus of the second embodiment are steps # in the flowchart of FIG. 7 showing the processing of the image processing apparatus (5, 6) of the first embodiment. 1 to # 8, which are the same as those in FIG. 1, the same reference numerals are assigned and description thereof is omitted, and only different steps # 19 to # 22 will be described below.

In step # 19 of FIG. 10, a circle is detected from the extracted circle candidates of the contour image. Specifically, as shown in FIG. 12, in a circle detection unit as an example of a ring detection unit,
(I) calculating the center of gravity position g (x, y) of the circle candidate 71;
(Ii) Calculate the area A of the circle candidate 71,
(Iii) Assuming that the circle candidate 71 is a perfect circle, the radius r = sqrt (A / PI) is calculated from the area A;
(Iv) Create a template circle image 72 with the center of gravity g (x, y) as the center of gravity,
(V) Template matching between the created template circle image 72 and the extracted circle candidate 71 is performed to calculate the similarity of the images. Only when it is equal to or greater than the threshold value, it is adopted as the detection circle contour. In template matching, a feature amount consisting of moment information in an image is calculated to obtain a degree of coincidence.

  In step # 20, it is determined whether the inner contour line of the large ring and the outer contour line of the small ring have not been detected when the circle is detected. If not, the process proceeds to step # 21, where the center of gravity ( x1, y1) and the center of gravity (x0, y0) of the small ring are calculated. If YES, the process proceeds to step # 22.

  In step # 22, the circle detection correction unit as an example of the ring detection correction unit performs circle estimation of each contour and corrects the circle detection in exactly the same manner as in step # 10 of FIG. 7 of the first embodiment. This circle detection correction is performed only when the inner contour of the large ring and the outer contour of the small ring are leaked from the above detection. By doing so, it is possible to improve the efficiency of image processing, and it is possible to detect the direction and amount of lens tilt with high accuracy and high speed.

(Embodiment 3)
12A and 12B are diagrams showing a schematic configuration of the lens tilt detection apparatus of the third embodiment and a ring light source 22. FIG.

  The lens tilt detection device 11 of the third embodiment is different from the configuration of the first embodiment shown in FIG. 1 only in the configuration of the ring light source 22 shown in FIG. 12, and the other components are the same as those of the first embodiment. The same reference numerals as those of the elements are attached, and detailed description is omitted.

  As shown in FIG. 12B, the ring light source 22 is not a circular ring shape but a rectangular ring shape having a space at the center.

  By irradiating the optical unit 4 with the irradiation light from the rectangular ring light source 22, as shown in FIG. 13, substantially rectangular strip-shaped imaging images 61 and 62 having different sizes are generated. The lens tilt amount can be calculated by calculating the shift between the center positions 63 and 64 of the two strip-shaped formed images in the same image processing flow as in the first and second embodiments.

  Thus, the ring-shaped light source does not necessarily have a circular shape, and the lens tilt amount can be detected by any light source as long as the formed image is a band-shaped image.

(Embodiment 4)
The lens tilt detection apparatus of the fourth embodiment differs from the configuration of the first embodiment only in the configuration of a circle detection correction unit (# 10 in FIG. 7 in the first embodiment) as an example of a ring detection correction unit. Since the other configuration is the same as the configuration of the first embodiment, FIG. 7 and the description of the first embodiment are used.

  The circle detection correction unit # 10 includes roundness calculation means for calculating the roundness of the contour line extracted by the contour extraction unit # 6 (see FIG. 7), and the roundness is determined by a predetermined threshold. The correction operation is performed only in the following cases.

  According to the fourth embodiment, the correction operation is performed only when the roundness of the contour line is equal to or less than a predetermined threshold value. Therefore, there is no unnecessary correction, and the tilt direction and the tilt amount are not performed. Detection speed can be improved.

  The ring light sources 2 and 22 of the first to fourth embodiments are substantially circular rings having continuous light emitting surfaces or continuous rectangular rings having substantially continuous light emitting surfaces. It may be circular or substantially rectangular with a portion cut.

  The present invention and the embodiment are summarized as follows.

The tilt detection device of the present invention is
Illumination light from the ring light sources 2 and 22 having a substantially light-emitting surface is irradiated onto the plurality of lens surfaces S1 and S2 of the lens unit 45 of the optical unit 4, and generated by reflected light from the plurality of lens surfaces S1 and S2. The lens tilt for detecting the tilt amount and the tilt direction of the lens unit 45 based on the relative shift amount of the center position of each of the ring-shaped formed images of the ring light sources 2 and 22 A detection device,
Imaging means 3 for imaging the ring-shaped image,
An image processing device (5, 6) for processing an image signal output from the imaging means,
The image processing device (5, 6)
An outline extraction unit (# 6) for extracting an outline of the ring-shaped image;
An area calculation unit (# 8) for calculating the area of the figure surrounded by the outline,
A ring detector (# 9, # 19) for detecting the center of the ring-shaped image surrounded by the contour line based on the contour line and the area;
The center of the large ring is detected based on the outer contour line of the large ring in the imaged image and the area of the figure surrounded by the outer contour line. Ring detection correction units (# 11, # 22) for detecting the center of the small ring based on the inner contour line of the small ring having a smaller diameter than the ring and the area of the figure surrounded by the inner contour line It is characterized by providing.

  According to the tilt detection apparatus having the above-described configuration, the heights of two ring-shaped formed images having different sizes are different, the depth of field of the imaging unit 3 is not sufficient, and the contour line of the formed image is Even if the image is not clear, the contour extraction unit (# 6), area calculation unit (# 8), and ring detection unit (# 9) can be used to easily change the shape of a ring such as a circle or a rectangle without being affected by edge information. The center can be detected.

  In the case of the large ring, the ring detection correction unit (# 11, # 22) detects the center of the outer contour line of the large ring, while in the case of the small ring, the inner contour line of the small ring. Since the center of the two ring-shaped contour lines overlaps, the center of the two ring-shaped contour lines (the center of a circle or a rectangle) is accurately obtained, and the tilt direction and The amount of tilt can be accurately detected.

  In addition, according to the present invention, if the two ring-shaped formed images having different sizes do not overlap, the contour extracting unit (# 6) and The ring detection unit (# 9, # 19) can easily detect the center of the ring-shaped formed image without being affected by the edge information, thereby improving the tilt detection speed. it can.

In one embodiment,
The ring detection unit (# 19) performs template matching between a template ring image having an area equivalent to the area calculated by the area calculation unit (# 8) and a figure surrounded by the contour line, thereby obtaining a threshold value. When having the above similarity, the contour line is adopted as the contour of the ring.

  According to the above embodiment, template matching is performed between the template ring image having an area equivalent to the area calculated by the area calculation unit (# 8) and the figure surrounded by the outline, so only the area information is included. Even in the case of ring detection, the detection accuracy of the tilt direction and the tilt amount can be improved.

In one embodiment,
The ring detection correction unit (# 11, # 22) includes a ring estimation unit that estimates the center of the ring-shaped formed image by the Hough transform of the contour line.

  According to the embodiment, since the center of the ring-shaped image is estimated by the Hough transform of the contour line, the tilt direction and the tilt amount can be detected with higher accuracy.

In one embodiment,
The ring detection correction unit includes roundness calculation means for calculating the roundness of the formed image, and performs an operation for correction only when the roundness is equal to or less than a predetermined threshold.

  According to the above embodiment, the ring detection correction unit performs the correction operation only when the roundness of the formed image is equal to or less than a predetermined threshold value. The tilt detection speed can be improved.

In one embodiment,
The ring light sources 2 and 22 have a ring shape whose light emitting surface is substantially circular or substantially rectangular.

  According to the embodiment described above, the ring light sources 2 and 22 can be adapted to various structures because the light emitting surface has a substantially circular or substantially rectangular ring shape.

  Of course, the constituent elements described in the first to fourth embodiments and the modification examples may be combined as appropriate, and may be appropriately selected, replaced, or deleted.

DESCRIPTION OF SYMBOLS 1 Lens tilt detection apparatus 2,22 Ring light source 3 Imaging camera 4 Optical unit 5 Rectangular illumination 10 Tilt sensor 21 Irradiation surface 31, 45 Lens unit 32 Light receiving element 41 Lens group L1, L2, L3 Lens 42 Lens holder 43 Actuator 71 Circle candidate 72 template circle image

Claims (5)

Illumination light from a ring light source whose light emitting surface is substantially ring-shaped is applied to a plurality of lens surfaces of the lens unit of the optical unit, and at least two or more of the ring light sources generated by reflected light from the plurality of lens surfaces A lens tilt detection device that detects a tilt amount and a tilt direction of the lens unit based on a relative shift amount of each center position of a ring-shaped formed image,
An imaging means for capturing the ring-shaped image;
An image processing device for processing an image signal output from the imaging means,
The image processing apparatus includes:
A contour extraction unit for extracting a contour line of the ring-shaped image;
An area calculation unit for calculating the area of the figure surrounded by the contour line;
A ring detection unit that detects the center of a ring-shaped imaging image surrounded by the contour line based on the contour line and the area;
The center of the large ring is detected based on the outer contour line of the large ring in the imaged image and the area of the figure surrounded by the outer contour line. A lens comprising: an inner contour line of a small ring having a smaller diameter than the ring, and a ring detection correction unit that detects a center of the small ring based on an area of a figure surrounded by the inner contour line Tilt detection device. The lens tilt detection apparatus according to claim 1,
The ring detection unit performs template matching between a template ring image having an area equivalent to the area calculated by the area calculation unit and a figure surrounded by the outline, and has a similarity equal to or greater than a threshold value. A lens tilt detection device, wherein the contour line is adopted as the contour of a ring. The lens tilt detection device according to claim 1 or 2,
2. The lens tilt detection device according to claim 1, wherein the ring detection correction unit includes a ring estimation unit that estimates a center of the ring-shaped formed image by a Hough transform of the contour line. In the lens tilt detection device according to any one of claims 1 to 3,
The ring detection correction unit includes a roundness calculation unit that calculates the roundness of the imaged image, and the lens tilt that operates for correction only when the roundness is equal to or less than a predetermined threshold value. Detection device. In the lens tilt detection device according to any one of claims 1 to 4,
The ring light source has a light emitting surface in a ring shape having a substantially circular or rectangular shape.