JP2014145626A - Image processing device, method and program for measuring shape of spiral spring - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique enabling endpoints of a spiral spring to be accurately identified using a photographed image of the spiral spring.SOLUTION: An image processing device includes: image input means for inputting a photographic image obtained by photographing a spiral spring; outline extraction means for extracting an outline of the spiral spring from the input photographic image; endpoint candidate extraction means for extracting a plurality of endpoint candidates on at least one of the end faces of the spiral spring based on the extracted outline; endpoint pair candidate creation means for creating a plurality of endpoint pair candidates constituted by two endpoint candidates selected from the plurality of the extracted endpoint candidates; and endpoint pair selection means for selecting a single pair of endpoint candidates from among the plurality of created endpoint pair candidates as two endpoints on the at least one of end faces of the spiral spring.

Description

  This specification discloses the technique which measures the shape of a spiral spring. The spiral spring here means a spring formed into a spiral shape when seen in a plan view.

  In the quality inspection process of industrial products such as spiral springs, the difference between the shape of the product and the shape defined in the design drawing is examined, and it is inspected whether or not the product is molded according to the design drawing. A product whose deviation from the shape specified in the design drawing is an allowable error is determined as a non-defective product, and a product whose deviation from the shape specified in the design drawing exceeds the allowable error is discarded as a defective product. In the spiral spring quality inspection process, the spiral spring is photographed, the shape of the spiral spring is measured from an image photographed using a computer, and the difference between the measured spiral spring shape and the shape specified in the design drawing is determined. inspect. In order to perform quality inspection appropriately, it is necessary to accurately measure the shape of the spiral spring from the captured image. For this reason, a technique for measuring the shape of an object using image processing has been developed (for example, Patent Document 1).

JP 2009-257950 A

  In order to accurately measure the shape of the spiral spring, it is necessary to accurately specify the end face of the spiral spring. That is, the spiral spring is manufactured by processing a spring steel material into a spiral shape and cutting the processed spring steel material. For this reason, the end surface (cut surface) by cut | disconnecting a spring steel material is formed in the edge part of a spiral spring. If the end face of the spiral spring cannot be specified accurately, the shape of the spiral spring cannot be specified accurately, so it cannot be confirmed whether or not the spiral spring can be manufactured according to the design drawing.

  Here, the end face of the spiral spring is defined by specifying a position where the end face is connected to the other surface of the spiral spring. For example, an end face, an outer peripheral surface, and an inner peripheral surface of the spiral spring appear as contour lines in a photographed image obtained by viewing the spiral spring in a plan view. The end face of the spiral spring is defined by a point (end point) where the end face and the outer peripheral face are connected and a point (end point) where the end face and the inner peripheral face are connected. Therefore, in order to specify the end face of the spiral spring, it is necessary to specify the points (two end points) at both ends of the end face of the spiral spring.

  In the conventional technique, the outline of the spiral spring is extracted from the captured image, and the end point of the end face of the spiral spring is extracted from the outline. Specifically, a point where the perpendicularity of the point on the contour line is maximized is extracted, and an end point is selected from the extracted plurality of end point candidates in consideration of the distance from the center of the spiral spring. Here, the perpendicularity of a point on the contour line means that the angle of the point on the contour line is a right angle, and the maximum squareness means that the point is closest to the surrounding point. doing. The angle of the point on the contour line is (1) the point, a straight line connecting the point on the contour line adjacent to the point by a minute distance, and (2) the point, An angle formed by a straight line connecting points on the contour line that are adjacent to the point on the other side by a minute distance.

  However, the end surface (cut surface) of the spiral spring is not flat due to burrs or the like generated during cutting, and dust or the like may adhere to the surface during manufacturing. For this reason, points other than the end points may be extracted as end point candidates. In the conventional technique, since only the distance from the center of the spiral spring is considered and the end point is specified from among the end point candidates, a point other than the end point may be erroneously recognized as the end point. This specification discloses the technique which can pinpoint the end point of the end surface of a spiral spring correctly from the picked-up image of a spiral spring.

  The present specification discloses an image processing apparatus for measuring the shape of a spiral spring formed in a spiral shape. The image processing apparatus includes an image input unit that inputs a captured image obtained by capturing a spiral spring, a contour line extracting unit that extracts a contour line of the spiral spring from the input captured image, and a spiral spring based on the extracted contour line. End point candidate extracting means for extracting a plurality of end point candidates for at least one end face; and end point pair candidate creating means for creating a plurality of end point pair candidates constituted by two end point candidates selected from the plurality of extracted end point candidates; End point pair selecting means for selecting one end point pair candidate as two end points of at least one end face of the spiral spring from among the plurality of end point pair candidates created.

  In this image processing apparatus, a contour line is extracted from the captured image of the spiral spring, and a plurality of end point candidates are extracted from the extracted contour line for at least one end face of the spiral spring. Then, two end point candidates are selected from the extracted plurality of end point candidates to create a plurality of end point pair candidates, and one end point pair candidate is selected from the created end point pair candidates. Since it is possible to comprehensively evaluate the possibility that the combination of the two end point candidates constituting the end point pair candidate is both ends of the end face, it is possible to accurately specify the end point of the end face as compared with the case where the end points are evaluated one by one. it can.

  The present specification also discloses an image processing method for measuring the shape of the spiral spring. The image processing method includes: an image input process for inputting a captured image obtained by photographing a spiral spring to a computer; a contour line extraction process for extracting a contour line of the spiral spring from the input captured image; and an extracted contour line. End point candidate extraction processing for extracting a plurality of end point candidates for at least one end face of the spiral spring, and end point candidate generation for generating a plurality of end point pair candidates constituted by two end point candidates selected from the plurality of extracted end point candidates Processing and end point pair selection processing for selecting one end point pair candidate as two end points of at least one end face of the spiral spring from among the plurality of end point pair candidates created are executed. By this method, the end point of the end face of the spiral spring can be specified accurately.

  The present specification further discloses a program for measuring the shape of a spiral spring. The program includes an image input process for inputting a photographed image of a spiral spring to a computer, a contour extraction process for extracting a contour line of the spiral spring from the input photographed image, and a spiral spring from the extracted contour line. End point candidate extraction processing for extracting a plurality of end point candidates on at least one of the end faces, and end point pair candidate creation processing for generating a plurality of end point pair candidates constituted by two end point candidates selected from the plurality of extracted end point candidates; Then, an end point pair selection process is performed in which one end point pair candidate is selected as two end points of at least one end face of the spiral spring from among the created end point pair candidates. With this program, the end point of the end surface of the spiral spring can be accurately specified from the captured image of the spiral spring using a computer.

The figure which shows the structure of an image processing apparatus. 6 is a flowchart showing a flow of image processing by the image processing apparatus. The flowchart which shows the flow of the process which pinpoints a hook area | region. The flowchart which shows the flow of the process which pinpoints an endpoint. The figure which shows the picked-up image of a spiral spring. The figure which shows the outline extracted from the picked-up image of a spiral spring. The graph of the angular coordinate value sequence obtained by carrying out polar coordinate conversion of the outline. The graph of the angular coordinate value sequence obtained by connecting the discontinuous point of the graph of FIG. 7a. The figure which expands and shows the specified inner hook area | region. The figure which expands and shows the specified outer hook area | region. The figure for demonstrating the process which trims the specified outer hook area | region. The figure which expands and shows the trimmed outer hook area | region. The figure for demonstrating the process which specifies an outer hook end point. The figure which shows an example of a squareness function. The figure which expands and shows the squareness function in the end point vicinity. The figure for demonstrating the method of approximating a squareness function by two Gaussian functions, and evaluating. The figure which shows an example which approximated the squareness function by two Gaussian functions.

  The main features of the embodiments described below are listed. The technical elements described below are independent technical elements and exhibit technical usefulness alone or in various combinations, and are not limited to the combinations described in the claims at the time of filing. Absent.

(Characteristic 1) In the image processing apparatus disclosed in the present specification, the end point pair selection unit includes, for each of the created end point pair candidates, the perpendicularity of the two end point candidates constituting the end point pair candidate and the end point candidates. One end point pair candidate may be selected by evaluating the end point pair candidate using the squareness of at least one point on the contour line connecting the two. At the end face of the spiral spring, the squareness is large at the end points (that is, the contour angle is close to a right angle), and the squareness is small at the points between the end points (that is, the contour angle is 180 °). Close to the value). Therefore, it is possible to comprehensively determine whether or not the end point of the spiral spring is by considering the perpendicularity of the two points of the end point pair candidates and the perpendicularity of the points on the contour line connecting these end point candidates.

(Characteristic 2) In the image processing apparatus disclosed in the present specification, the endpoint pair selection unit uses the perpendicularity of two candidate endpoints and the perpendicularity of the point having the lowest perpendicularity on the contour line connecting these endpoint candidates. The end point pair candidate may be evaluated. According to such a configuration, by using the perpendicularity of the point with the lowest perpendicularity between the two end point candidates, the flatness between the two end point candidates can be evaluated, and it is preferable whether or not the end point of the spiral spring. Can be determined.

(Characteristic 3) In the image processing apparatus disclosed in the present specification, the end point pair selecting unit further selects one of the end point pair candidates in which the perpendicularity of each point on the contour line connecting the two end point candidates is equal to or less than the set value. One end point pair candidate may be selected. According to such a configuration, an end point candidate pair having a high perpendicularity point between two end point candidates is excluded, and an incorrect end point pair candidate can be prevented from being selected.

(Characteristic 4) In the image processing apparatus disclosed in the present specification, the end point pair selecting unit further selects one end point pair candidate from end point pair candidates whose distance between the two end point candidates is within the setting range. May be. According to such a configuration, it is possible to prevent selection of an end point pair candidate in which the distance between the two end point candidates is far from the plate thickness of the spring steel material.

  As shown in FIG. 1, the image processing apparatus 10 according to the present exemplary embodiment captures the spiral spring 30, extracts a contour line of the spiral spring 30 from the captured image, and extracts end points (in this embodiment, It is a device that specifies a hook end point. The image processing apparatus 10 includes a stage 12, an illuminator 14 disposed on the stage 12, a CCD camera 16 fixed to the stage 12, and a computer 22 connected to the CCD camera 16 via a communication line 18. And a display 20 connected to the computer 22.

  The illuminator 14 is a surface light source, and a spiral spring 30 is placed on the light emitting surface thereof. The CCD camera 16 is disposed on the illuminator 14 and photographs the spiral spring 30 placed on the illuminator 14. That is, the spiral spring 30 is illuminated from below by the illuminator 14, and the transmitted light (shadow of the spiral spring 30) of the illuminator 14 is photographed by the CCD camera 16. The illuminator 14 is not limited to transmitted illumination as in the present embodiment, and may illuminate the spiral spring 30 from above. In this case, it is preferable to use a plurality of illuminators or ring-type illuminators so that the spiral spring 30 is illuminated uniformly from the entire circumferential direction.

  Image data captured by the CCD camera 16 is input to the computer 22 via the communication line 18. The computer 22 stores a program for executing various image processing described later. The computer 22 processes the image data of the photographed image taken by the CCD camera 16, creates a contour image of the spiral spring 30, and specifies the hook end point from the contour image. The computer 22 displays the captured image of the spiral spring 30, the analysis result, etc. on the display 20.

  FIG. 2 is a flowchart showing a flow of image processing of the spiral spring 30 by the image processing apparatus 10. FIG. 3 shows the hook area specifying process in step S18 shown in the flowchart of FIG. FIG. 4 is a flowchart showing a flow of hook end point calculation processing in step S20. The image processing apparatus 10 specifies the hook end point of the spiral spring 30 through the processes and processes shown in FIGS. 2, 3, and 4. Hereinafter, the flow of processing by the image processing apparatus 10 will be described with reference to the flowcharts shown in FIGS.

  First, in step S10 of FIG. 2, the spiral spring 30 is photographed by the CCD camera 16. The captured image captured by the CCD camera 16 is input to the computer 22. The input captured image can be, for example, a 256-gradation image in which the density value of the blackest pixel is “255” and the density value of the whitest pixel is “0”. FIG. 5 illustrates a captured image of the spiral spring 30. As shown in FIG. 5, the spiral spring 30 extends from the inner hook 32 to the outer hook 34 in a spiral shape. The photographed image is a density value image having a density value for each pixel (x, y).

  Next, in step S12 of FIG. 2, the captured image captured in step S10 is binarized. That is, for each pixel (x, y) of the input captured image, when the density value is equal to or greater than a preset set value, “1” is set, and the density value is less than the preset set value. In this case, “0” is set. As a result, the pixel group corresponding to the spiral spring 30 has the density value “1”, and the other pixel groups have the density value “0”. In addition, since the image image | photographed with the CCD camera 16 contains noise, the binarized image also contains noise. For this reason, some pixels that do not correspond to the spiral spring 30 have a density value “1”. The processing from step S12 to S20 described later is executed by the computer 22.

Next, in step S14 of FIG. 2, the center of gravity (c _x , c _y ) of the spiral spring 30 is calculated using the binarized image obtained in step S12. Specifically, the center of gravity (c _x , c _y ) is calculated from the coordinate values (x, y) of the pixel group having the density value “1” (that is, the pixel group corresponding to the spiral spring 30). To do. In the example shown in FIG. 5, the center of gravity (c _x , c _y ) is located approximately at the center of the spiral spring 30.

  Next, in step S16 in FIG. 2, a contour line is extracted from the binarized image obtained in step S12. The process of extracting the contour line from the binarized image is a known method (for example, Junichi Hasegawa, Yamato Usui, Akira Nakayama, Shige Yokoi, “Basic Techniques of Image Processing <Technical Introduction>”, Technical Review, p. .70-73) can be used. FIG. 6 shows an example of the contour line extracted from the binarized image of the spiral spring 30.

In this embodiment, the contour line extracted in step S16 is the contour line element number i (i = 0 to n) and the coordinate value (x _i , y _{i) of} the pixel corresponding to the contour line element number _i. ). The contour line element number i is a number given to the pixels on the contour line, and is given according to the following rules. That is, with one pixel set on the contour as the origin, the contour element number i = 0 is assigned to that pixel. The origin can be set at any point on the contour line. Next, contour element number 1 is assigned to the pixel adjacent to the origin. Here, there are two types of pixels serving as the origin: a pixel adjacent in the clockwise direction on the contour line and a pixel adjacent in the counterclockwise direction. In this embodiment, among the pixels adjacent to the origin, the contour element number 1 is given to the pixel adjacent in the direction of moving counterclockwise. Next, the contour element number 2 is assigned to the pixel adjacent to the pixel of the contour element number 1, and the contour element number is assigned in the same procedure. As a result, a contour element number is assigned to all the pixels constituting the contour. As is apparent from the above description, when the pixel is sequentially moved from the pixel with the contour element number i to the pixel with the contour element number n, the contour line is circulated in one direction. In this specification, the coordinate value sequences (x ₀ , y ₀ ), (x ₂ , y ₂ ),... (X _n , y _n ) from the contour element numbers 0 to n are represented by the contour line series (x _i , Y _i ) (i = 0 to n).

Next, in step S17 of FIG. 2, the contour line series (x _i , y _i ) (i = 0 to n) obtained in step S16 is converted into a contour line series (r _i , θ _i ) (i = 0). To n). That is, as shown in FIG. 6, the contour series (x _i , y _i ) in the XY coordinate system is set with the center of gravity (c _x , c _y ) calculated in step S14 as the origin and the radius r _i and the declination angle. It represents by (theta) _i . As shown in FIG. 6, the moving radius r _i is the distance between the origin (c _x , c _y ) and the contour line series (x _i , y _i ). The deflection angle θ _i is an angle formed by a straight line connecting the origin (c _x , c _y ) and the contour line series (x _i , y _i ) and the x axis. Therefore, in step S17, the contour line series (c _x , c _y ) (i = 0 to n) in the xy coordinate system is changed to the contour line series (r _i , θ _i ) (i = 0 to n) in the polar coordinate system. Converted. Here, when the angular coordinate value sequence θ _i (i = 0 to n) obtained from the contour line series (r _i , θ _i ) of the polar coordinate system is graphed, the graph shown in FIG. 7a is obtained. In the graph shown in FIG. 7A, the vertical axis is the deflection angle θ _i , and the horizontal axis is the contour element number i. As is apparent from FIG. 7a, the angular coordinate value sequence θ _i has a numerical value in the range of −180 ° to + 180 °, and is discontinuous at −180 ° and + 180 °.

Next, in step S18 of FIG. 2, a hook region is specified from the contour line series (r _i , θ _i ) (i = 0 to n) subjected to polar coordinate conversion in step S17. This process will be described along the flowchart shown in FIG.

First, in step S34, S36 in FIG. 3, the contour line of the polar coordinate system created in step S17 in FIG. 2 _{(r i,} θ _i) from (i = 0 to n), extracting the inner hook area and an outer hook region To do. In this embodiment, the contour line _{(r i,} θ _i) of the (i = 0 to n), by utilizing the characteristics of the angular coordinate value sequence θ _{i (i} = _0~n), an inner hook area A The outer hook area B is specified.

In the graph shown in FIG. 7A, the angular coordinate value sequence θ _i (i = 0 to n) is in a range of −180 ° to + 180 °, and the angular coordinate value sequence θ _i has discontinuous points. The processing after step S34 may be performed using an angular coordinate value sequence (FIG. 7b) obtained by connecting the discontinuous points of the graph shown in FIG. 7a, or an angular coordinate value sequence (FIG. 7a) in which the discontinuous points are left. ) May be used.

As is clear from the above, the contour line series (r _i , θ _i ) (i = 0 to n) starts to move counterclockwise from the origin on the contour line, and then completes the contour line by making a round of the contour line. Line element numbers are assigned. For this reason, when the spiral spring is set clockwise when viewed from the inside, the contour element number is given while moving counterclockwise from the origin to the inner hook end point (or outer hook end point), and the inner hook end point ( Or, the contour element number is given while moving clockwise from the outer hook end point to the outer hook end point (or inner hook end point), and it moves counterclockwise from the outer hook end point (or inner hook end point) to the origin. The contour element number is assigned. Therefore, in the graph (FIG. 7b) in which discontinuous points are connected, the angular coordinate value sequence θ _i (i = 0 to n) monotonously decreases (or monotonically increases) from the origin to the inner hook end point (or outer hook end point). From the inner hook end point (or outer hook end point) to the outer hook end point (or inner hook end point), the angular coordinate value sequence θ _i (i = 0 to n) monotonously increases (or monotonously decreases), and the outer hook end point (or The angle coordinate value sequence θ _i (i = 0 to n) monotonously decreases (or monotonously increases) from the inner hook end point) to the origin. Therefore, as shown in FIG. 7b, the function of the angular coordinate value sequence θ _i (i = 0 to n) takes a minimum value near the inner hook end point and takes a maximum value near the outer hook end point.

Therefore, in step S34, the area A including the portion where the angular coordinate value sequence θ _i (i = 0 to n) is the minimum value is specified as the inner hook area. For example, as shown in FIG. 7a, as the inner hook region A, an outline element adjacent to the minimum value and having an angular coordinate value _θa2 of 180 ° from an outline element number a1 having an angular coordinate value _θa1 of 180 ° The region up to the number a2 is specified, and the minimum value is included in this region (a1 to a2). FIG. 8 shows a contour line series of the identified inner hook area A. As is clear from FIG. 8, the inner hook end point is included in the identified inner hook region A.

In step S36, the region B including the portion where the angular coordinate value sequence θ _i (i = 0 to n) is the maximum value is specified as the outer hook region. For example, as shown in FIG. 7a, as the outer hook region B, a contour where the angular coordinate value _θb2 is −180 ° from the contour line element number b1 where the angular coordinate value _θb1 is −180 ° is adjacent to the maximum value. The region up to the line element number b2 is specified, and the maximum value is included in this region (b1 to b2). FIG. 9 shows a contour line series of the identified outer hook region B. As apparent from FIG. 9, the outer hook end point is included in the identified outer hook region B.

As is clear from the comparison between FIG. 8 and FIG. 9, the outer hook region B is wider than the inner hook region A. That is, when the specified outer hook region B (r _k , θ _k ) (k = b1 to b2) is wide, it is preferable to trim the outer hook region B as shown in FIG. In the example shown in FIG. 10, the angle range from the maximum value i _max to the angle α (for example, 40 °) is newly cut out as a new outer hook region B ′ (r _k , θ _k ) (k = b1 ′ to b2 ′). ing. FIG. 11 shows the outer hook region B ′ after trimming. As is clear from FIG. 11, the outer hook area B ′ after trimming is corrected to an appropriate size as compared with the outer hook area B before trimming. Thereby, the calculation amount of the hook end point calculation process (the process of step S20 in FIG. 2) to be performed next can be reduced.

  Further, in the graph shown in FIG. 7b, the local hook takes the local minimum value in the inner hook region A and the local maximum in the outer hook region B. And take a local minimum in the outer hook area. It should be noted that other known techniques may be used for the hook region specification processing (S16 to S18) from extraction of the outline of the spiral spring. For example, using the technique disclosed in Japanese Patent Application Laid-Open No. 2009-257950, the inner hook image and the outer hook image may be cut out from the spiral spring image, and the contour line may be extracted from the cut hook image.

Next, when returning to step S20 in FIG. 2, the inner hook end point (x _ac , y _ac ) is identified from the inner hook area A (x _k , y _k ) (k = a1 to a2) identified in step S18, Further, the outer hook end point (x _bc , y _bc ) is identified from the outer hook region B (x _k , y _k ) (k = b1 ′ to b2 ′) identified in step S18. Here, the process of specifying the inner hook end point from the inner hook area A and the process of specifying the outer hook end point from the outer hook area B are the same process except for the target area. For this reason, in the following description, the inner hook area A and the outer hook area B will be described as hook areas without being distinguished, and the inner hook end points and outer hook end points will be described as hook ends without being distinguished.

FIG. 4 is a flowchart showing a flow of processing for identifying a hook end point from a hook area. First, in step S42 in FIG. 4, end point candidates are extracted from the contour line represented by the identified hook region (x _k , y _k ). That is, since the spiral spring 30 is manufactured by cutting a spring steel material processed into a spiral shape, the contour line is bent (orthogonal) at approximately 90 ° at the end face. However, the end surface may not be flat due to burrs or the like generated when the spring steel material is cut, and dust or the like may adhere to the surface of the spring steel material during manufacturing. For this reason, there is a point (a point that looks like an end point) that is bent other than the end point in the outline of the hook region. Therefore, in step S42, a hook end point candidate (a point that looks like an end point) is extracted from the contour line of the identified hook region (x _k , y _k ).

A specific procedure will be described by taking as an example a case where endpoint candidates are extracted from the outer hook region B ′ (x _k , y _k ) (k = b1 ′ to b2 ′). As shown in FIG. 12, there are two outer hook end points in the outer hook region B ′ (x _k , y _k ) (k = b1 ′ to b2 ′), and a contour line is formed at these two outer hook end points. It is almost orthogonal. Therefore, first, the points where the contour lines are substantially orthogonal are extracted from the identified contour line series B ′ (x _k , y _k ) (k = b1 ′ to b2 ′). This processing is performed, for example, with respect to each of the contour line series B ′ (x _k , y _k ) (k = b1 ′ to b2 ′) constituting the outer hook region B ′ as shown in FIG. (K) and m (k) are set, and extraction can be performed based on whether these vectors p (k) and m (k) are orthogonal. p (k) and m (k) can be expressed by the following formula (1), for example.

As is apparent from the above equation (1), m (k) is the contour line sequence (x _k−N ) from the point (x _k , y _k ) on the contour line to each point on the contour line N pixels before. , Y _k−N ), (x _{k− (N−1)} , y _{k− (N−1)} ),..., (X _k , y _k ) are averaged. Further, p (k) is a contour line sequence (x _k , y _k ), (x _{k + 1} , y _{k + 1} ) from the point (x _k , y _k ) on the contour line to each point on the contour line N pixels behind. ,..., (X _{k + N} , y _{k + N} ) are averaged. In this embodiment, a range of 0 to N pixels is used to calculate the vectors m (k) and p (k). Here, the set value N is preferably set according to the thickness of the strand of the spiral spring 30 in the image taken in step S10. For example, when the thickness of the wire of the spiral spring 30 is about Z pixels in the captured image, the set value N can be set to Z / 2. Thus, it is preferable to set the number of pixels for calculating m (k) and p (k) depending on how many pixels the plate thickness of the spiral spring 30 corresponds to in the captured image. .

As described above, when the vector m (k) and the vector p (k) are calculated, it is next determined whether or not the vector p (k) and the vector m (k) are orthogonal. In order to determine whether the vector p (k) and the vector m (k) are orthogonal to each other, for example, cos θ _k calculated from the vector p (k) and the vector m (k) can be used (described below) (Refer Formula (2)). Here, θ _k is an angle formed by the vector p (k) and the vector m (k).

When the vector m (k) and the vector p (k) are substantially orthogonal, θ _k ≈90 ° and cos θ _k ≈0. Therefore, by searching for a point _where cos θ _k ≈0, a hook end point candidate can be extracted from the contour line series. Here, cos [theta] _k takes a value in the range of -1 to 1, theta _k is not cos [theta] _k is maximized or minimized when the 90 °. For this reason, in this embodiment, a function that takes the maximum value or the minimum value when θ _k is 90 ° is introduced to facilitate the extraction of hook end point candidates. For example, as a function that takes a maximum value when θ _k is 90 °, an evaluation function V (k) of the following equation (3) can be introduced, and a minimum value when θ _k is 90 °: An evaluation function V (k) of the following equation (4) can be introduced as a function that takes FIG. 13 shows the evaluation function V (k) of Expression (3). As is clear from FIG. 13, the evaluation function V (k) has a maximum value at the hook end point.

  By using the above formula (3) or formula (4), it is possible to evaluate whether the point on the contour line is an end point candidate (that is, whether the perpendicularity is high). Therefore, the value calculated by the above formula (3) or formula (4) corresponds to an example of “squareness”.

In addition, when using said Formula (3), the point which satisfies the following conditions (a) and (b) can be extracted as an end point candidate, for example. That is, (a) the evaluation function V (k) takes a maximum value (V (k−1) <V (k)> V (k + 1)), and (b) the evaluation function V (k) is equal to or greater than the threshold value p (p Is set as appropriate between 0 and 1). As a result, the points having a high squareness among the points constituting the contour line are selected as the end point candidates (x _p1 , y _p0 ),... (X _pm , y _pm ).

On the other hand, when the above equation (4) is used, for example, a point that satisfies the following conditions (c) and (d) can be extracted as an end point candidate. That is, (c) the evaluation function V (k) takes a minimum value (V (k−1)> V (k) <V (k + 1)), and (b) the evaluation function V (k) is equal to or less than the threshold value p (p Is set as appropriate between 0 and 1). Also by this, a point having a high _squareness among the points on the contour line is selected as the end point candidate (x _p0 , y _p0 ),... (X _pm , y _pm ).

When the end point candidates (x _p1 , y _p0 ),... (X _pm , y _pm ) are extracted in step S 42 in FIG. 4, two end points are selected from these end point candidates to create end point pair candidates ( Step S44 in FIG. That is, all possible combinations from all the end point candidates (x _p1 , y _p0 ),... (X _pm , y _pm ) are created as end point pair candidates. For example, when three points (x _p0 , y _p0 ), (x _p1 , y _p1 ), (x _p2 , y _p2 ) are extracted as end point candidates, {(x _p0 , y _p0 ), (x _p1 , y _p1 )} end point pair candidates, {(x _p0 , y _p0 ), (x _p2 , y _p2 )} end point pair candidates, and {(x _p1 , y _p1 ), (x _p2 , y _p2 )} End point pair candidates are created.

  In step S46, one end point pair candidate is selected from the plurality of end point pair candidates created in step S44. Then, in step S48, it is determined whether or not the distance between the two selected endpoint pair candidates is less than the set value. That is, when the distance between the two end point pair candidates is too large compared to the plate thickness of the spiral spring 30, the possibility that the end point pair candidate is a hook end point is low. For this reason, a pair of correct combinations is narrowed down from a plurality of end point pair candidates using the distance between the two end point pair candidates. It should be noted that the narrowing down using the distance between the two endpoint pair candidates may be determined based on whether or not the distance between the two endpoint pair candidates is within the set numerical range (that is, the upper limit value and the lower limit value). May be set).

When the distance between the two selected end point pair candidates exceeds the set value (NO in step S48), step S50 is skipped and the process proceeds to step S52. Therefore, the process of step S50 is not performed for the endpoint pair candidates that are determined not to be the correct endpoint pair. On the other hand, if the distance between the two selected end point pair candidates is less than the set value (YES in step S48), an end point pair candidate evaluation value is calculated (step S50). That is, an evaluation value for evaluating whether or not the end point pair candidate selected in step S46 is a correct end point combination is calculated. This evaluation value is calculated using, for example, the end point pair evaluation function D (a _j , b _j ) (a _j and b _j are end point candidates) using the above-described evaluation function V (k) representing the perpendicularity. Can do. When the evaluation function V (k) shown in the above equation (3) is used, for example, an evaluation function D (a _j , b _j ) shown in the following equation (5) can be used.

Here, V (a _j ) is an evaluation value (hereinafter, simply referred to as a _{perpendicularity} evaluation value) representing the _{perpendicularity} of the end point candidate a _j (x _aj , y _aj ), and V (b _j ) is the end point candidate. b _j (x _bj , y _bj ) is a squareness evaluation value, and V (k _min ) is the minimum value of the squareness evaluation values of the points located between the end point candidate a _j and the end point candidate b _j. Is the value. Therefore, D (a _j , b _j ) is obtained by subtracting the minimum value V (k _min ) from the _{perpendicularity} evaluation value V (a _j ) of the end point candidate a _j and the end point candidate b as shown in FIG. _It is the sum of the squareness evaluation value V (b _j ) of _j and the value obtained by subtracting the minimum value V (k _min ). Therefore, even if the perpendicularity of the end point candidates formed by burrs, dust, etc. is high, the point located between these end point candidates a _j and b _j does not have high flatness (that is, the perpendicularity is low Must not). Therefore, by using the endpoint pair evaluation function D (a _j , b _j ) in the above formula (5), it is possible to prevent the selection of the endpoint candidate having a high squareness as the correct endpoint combination. Can do.

In the case of using the evaluation function V (k) shown in the above equation (4), for example, the evaluation function _{D (a} j, _{b j)} shown in the following equation (6) can be used.

In the evaluation function V (k) shown in the above equation (4), the perpendicularity is the highest when the minimum value is reached. Therefore, in the evaluation function D (a _j , b _j ) of Expression (6), a value obtained by subtracting the _squareness evaluation value V (a _j ) of the end point candidate a _j from the maximum value V (k _max ), and the maximum value The sum of the value obtained by subtracting the _{perpendicularity} evaluation value V (b _j ) of the end point candidate b _j from V (k _max ) is used as the evaluation value.

In addition to the above, the evaluation function for evaluating the endpoint pair candidates can take various forms. For example, when the evaluation function V (k) shown in the above equation (3) is used, D (a _j , b _j ) = (V (a _j ) −V (k _min )) ² + (V (b _j ) −V (k _min )) ² and D (a _j , b _j ) = | V (a _j ) −V (k _min ) | + | V (b _j ) −V (k _min ) | Also good. In addition, also when using evaluation function V (k) shown in said Formula (4), it can arrange similarly to the above.

In step S52 of FIG. 4, it is determined whether or not the processes of steps S46 to S50 have been executed for all the end point pair candidates created in step S44. If the processing of steps S46 to S50 has not been executed for all end point pair candidates (NO in step S52), the process returns to step S46 and the processing from step S46 is executed. Thereby, for each of all the end point pair candidates created in step S44, the determination based on the distance in step S48 and the numerical value based on the evaluation function (a _j , b _j ) are calculated.

On the other hand, when the process of steps S46 to S50 is executed for all the end point pair candidates (YES in step S52), the end point pair candidate whose numerical value by the evaluation function (a _j , b _j ) of step S50 is maximized. Is selected as the end point of the spiral spring 30. Thereby, the end point of the spiral spring 30 is specified, and the subsequent processing is executed. Examples of the subsequent process include a process of creating a core image of the spiral spring 30 by thinning a contour image obtained from a captured image. In this case, the created cored image of the spiral spring 30 is displayed on the display 20. Further, the end point extraction technique disclosed in the present specification can be applied to, for example, a spiral spring inspection apparatus disclosed in Japanese Patent Application Laid-Open No. 2009-257950. The size of each part of the spiral spring can be calculated using the end point information extracted by the technique disclosed in this specification.

As described above, in the image processing apparatus 10 according to the present embodiment, the contour line series (x _i , y _i ) (i = 0 to n) of the spiral spring 30 is extracted from the captured image, and the extracted contour line is extracted. Extract endpoint candidates from the sequence. Then, end point pair candidates are created from the extracted end point candidates, each of the created end point pair candidates is comprehensively evaluated, and the end points of the spiral spring 30 are specified. For this reason, even if burrs are generated on the end face of the spiral spring 30 or dust or the like adheres to the surface of the spiral spring 30, the end point (ie, end face) of the spiral spring 30 can be appropriately specified. . For this reason, the shape of the spiral spring 30 can be accurately measured.

  Specific examples of the present invention have been described in detail above, but these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and changes of the specific examples illustrated above.

For example, in the above-described embodiment, the trigonometric function having θ _k as a variable is used as the evaluation function V (k) for evaluating the perpendicularity, but the technique disclosed in the present specification is not limited to such an example. . For example, as shown in FIG. 15, an evaluation function obtained by approximating an evaluation function V (k) for evaluating a squareness with a Gaussian function may be introduced. That is, as shown in FIG. 16, the two peaks of the evaluation function V (k) shown in the above figure may be approximated by a Gaussian function, and the sum of these two Gaussian functions may be used as an evaluation function for evaluating the squareness. If an evaluation function for evaluating perpendicularity is obtained, the subsequent processing can be performed in the same manner as in the above-described embodiment.

  In addition, the technical elements described in the present specification or the drawings exhibit technical usefulness alone or in various combinations, and are not limited to the combinations described in the claims at the time of filing. In addition, the technology exemplified in this specification or the drawings achieves a plurality of objects at the same time, and has technical utility by achieving one of the objects.

10: Image processing device 12: Stage 14: Illuminator 16: CCD camera 18: Communication line 20: Display 22: Computer 30: Spiral spring

Claims (7)

An image processing device for measuring the shape of a spiral spring formed in a spiral shape,
An image input means for inputting a photographed image of the spiral spring;
An outline extraction means for extracting the outline of the spiral spring from the input photographed image;
End point candidate extracting means for extracting a plurality of end point candidates of at least one end face of the spiral spring from the extracted contour line;
End point pair candidate creating means for creating a plurality of end point pair candidates constituted by two end point candidates selected from the plurality of extracted end point candidates;
End point pair selecting means for selecting one end point pair candidate as two end points of at least one end face of the spiral spring from among the plurality of end point pair candidates created;
An image processing apparatus.   For each of the created end point pair candidates, the end point pair selecting means uses the perpendicularity of the two end point candidates constituting the end point pair candidate and the perpendicularity of at least one point on the contour line connecting these end point candidates. The image processing apparatus according to claim 1, wherein the endpoint pair candidates are evaluated to select one endpoint pair candidate.   The end point pair selection means evaluates the end point pair candidate using the perpendicularity of the two endpoint candidates and the perpendicularity of the point with the lowest perpendicularity on the contour line connecting these endpoint candidates. Image processing device.   The end point pair selecting means further selects one end point pair candidate from among end point pair candidates in which the perpendicularity of each point on the contour line connecting the two end point candidates is equal to or less than a set value. An image processing apparatus according to claim 1.   The end point pair selecting means further selects one end point pair candidate from end point pair candidates whose distance between the two end point candidates is within the setting range. Image processing device. An image processing method for measuring the shape of a spiral spring formed into a spiral shape.
Image input processing for inputting a captured image of the spiral spring;
An outline extraction process for extracting the outline of the spiral spring from the input photographed image;
An end point candidate extraction process for extracting a plurality of end point candidates of at least one end face of the spiral spring from the extracted contour line;
End point pair candidate creation processing for creating a plurality of end point pair candidates constituted by two end point candidates selected from a plurality of extracted end point candidates;
An end point pair selection process for selecting one end point pair candidate as two end points of at least one end face of the spiral spring from among the plurality of end point pair candidates created;
An image processing method for executing This is a program for measuring the shape of a spiral spring formed into a spiral shape.
Image input processing for inputting a captured image of the spiral spring;
An outline extraction process for extracting the outline of the spiral spring from the input photographed image;
An end point candidate extraction process for extracting a plurality of end point candidates of at least one end face of the spiral spring from the extracted contour line;
End point pair candidate creation processing for creating a plurality of end point pair candidates constituted by two end point candidates selected from a plurality of extracted end point candidates;
An end point pair selection process for selecting one end point pair candidate as two end points of at least one end face of the spiral spring from among the plurality of end point pair candidates created;
A program that executes

Images