JP4921526B2 - Register mark - Google Patents

Description

  The present invention relates to a register mark used for registration control.

  Conventionally, register marks are used for registration control in a printing process. This register mark is required to be sensitive to printing misalignment in the vertical and horizontal directions.

  For this reason, various figures that satisfy this requirement have been proposed as register marks, and some of them have been put into practical use.

  Registration control is performed by measuring the width of the characteristic part of the register mark (hereinafter referred to as “first registration control”) and the position of the center (center or center of gravity) of the register mark. This is roughly classified into what is performed (hereinafter referred to as “second registration control”).

  In the first registration control, for example, a right triangle is used as the register mark. For example, Patent Document 1 proposes a web alignment control device that uses right-angle diamonds as register marks.

  In the second registration control, for example, a circle is used as a register mark.

Japanese Patent No. 2538913 (Japanese Patent Laid-Open No. 63-22651)

  The first registration control and the second registration control are based on the premise that the graphic to be controlled, that is, the register mark is a complete graphic (that is, a graphic with no missing or surplus).

  However, various register noises are included in an actual register mark, and imperfections of the register mark as a pattern are generated due to the graphic noise. For example, the degree of unevenness of the surface of the sheet on which the register mark is printed, insufficient resolution when resolving the register mark, distortion of the lens that captures the register mark, various processes from register mark imaging to resolution Various graphic noises are generated due to electrical noise and the like.

  Therefore, theoretically, as a problem before the performance of the registration control device that performs the registration control, it is impossible to perform the registration control as intended.

  Specifically, in the first and second registration controls described above, the following control failure occurs.

  In the first registration control, the register mark is analyzed based on data obtained by scanning performed once on one position of the register mark. For this reason, if high frequency graphical noise is included in the obtained data, it is directly connected to a measurement error.

  In general, in calculating the center of gravity and the center, it is assumed that the weight distribution of a target graphic is isotropic. For this reason, in the second registration control, a chip in the outer shape of the register mark and the presence of a hole (void) inside the register mark cause an error.

  If there is a chip in the outer shape of the register mark, it is conceivable that the chip is complemented by figure fitting (a method for predicting the chipped part by another part other than the chip). Because the register mark outline is incomplete in the first place, the figure fitting does not function as desired.

  As described above, any of the currently used registration controls is immune to graphic noise (incompleteness as a register mark figure) that is always included in the actual image of the register mark. Not.

  The present invention has been made in view of such problems in conventional registration control, and a register mark having sufficient immunity against graphical noise that is necessarily included in an actual image of the register mark, Furthermore, it aims at providing the register mark position identification method which can exhibit the performance of the said register mark effectively.

In order to achieve the above object, the present invention provides a register mark position identifying method for reading the position of a register mark printed on a sheet, wherein each direction of a plurality of sides constituting the register mark is predetermined. The register mark position identification method includes a first process of printing the star-shaped register mark on the sheet, and a second process of imaging the register mark on the fed sheet. The image data of the register mark imaged in the first process is scanned in parallel with the respective sides constituting the register mark, and the direction in which the scanning is performed is a direction orthogonal to the respective sides. And a position of each side of the register mark based on a result of the scanning and a third process of moving the register mark to the register Providing a fourth step of locating the over click, the register mark position identification method comprising.
The register mark is composed of, for example, a five-legged star shape.

  The register mark position identification method according to the present invention preferably further comprises a step of storing the image data of the register mark imaged in the second step in an image memory.

  The position of each side is specified by, for example, determining the peak value of the number of integrated pixels obtained by each scan.

  According to the register mark or the register mark position identification method according to the present invention, the following effects can be obtained.

  First, great immunity is guaranteed against missing or surplus register mark shapes.

  Conventionally, if a register mark is missing, it is almost impossible to accurately specify the position of the register mark. On the other hand, according to the register mark or the register mark position identifying method according to the present invention, since the scanning is performed along each side constituting the register mark, the register mark is missing a shape (for example, a hole). Or dents) or extra shapes added (for example, bots (blots)), edge location is not affected by them and can be performed accurately . That is, even if there is a hole, a chip, or a void in the register mark, it does not cause an error in specifying the register mark position.

  Secondly, the noise on each side of the register mark is averaged over the entire side, thus substantially reducing the error.

  Third, an error in register mark position can be reduced.

  For example, when the register mark is a five-legged star shape, the position of the register mark can be specified by specifying the positions of the five sides. For this reason, even if the register mark itself lacks the perfect shape as a star, the error of the register mark position is averaged over the entire figure, and the error can be reduced.

FIG. 1 is a plan view showing a register mark according to the first embodiment of the present invention. FIG. 2 is a plan view showing another example of the register mark according to the first embodiment of the present invention. It is a top view which shows the number of five sides of a star shape. FIG. 4 (A) shows combinations in which three of the five sides constituting the star are obtained (combination numbers 7, 11, 13, 14, 19, 21, 22, 25, 26, 28). 4 (B) shows a combination in which four of the five sides constituting the star are obtained (combination numbers 15, 23, 27, 29, 30). FIG. 5 is a block diagram of a register mark position identification device used for register mark position identification according to the first embodiment of the present invention. 6 is a block diagram of a scanning device which is a component of the register mark position identification device shown in FIG. FIG. 7 is a block diagram of an analysis device which is a component of the register mark position identification device shown in FIG. FIG. 8 is a schematic view showing a state of scanning for one side of the register mark. FIG. 9 is a schematic diagram showing the result of analysis of register mark image data by the analyzer. FIG. 10 is a schematic diagram showing a result of analysis of register mark image data by the analysis apparatus using a method other than that shown in FIG. FIG. 11 is a schematic diagram showing a state of scanning for one side of the register mark. FIG. 12 is a schematic diagram showing a scanning implementation state for one side of the register mark. FIG. 13 is a schematic diagram showing a state of scanning in the conventional registration control. FIG. 6 is a block diagram of an example of a registration control device including the register mark position identification device shown in FIG. 5. It is a top view which shows the area division | segmentation of a star-shaped register mark. FIG. 16A illustrates the boundary expression in simulation 1, and FIG. 16B plots the determination of the inside and outside of the boundary on a matrix. It is a top view which shows the determination process of the definition formula of each side of a star-shaped register mark. It is a top view which shows the determination process of the definition formula of each side of a star-shaped register mark. It is a figure which shows the verification result of the star-shaped register mark used as the object of simulation 1, and simulation. It is a figure which shows the verification result of the star-shaped register mark used as the object of simulation 1, and simulation. It is a figure which shows the verification result of the star-shaped register mark used as the object of simulation 1, and simulation. It is a top view which shows the star-shaped register mark used as the object of the simulation 2. FIG. It is a figure which shows the verification result of the star-shaped register mark used as the object of simulation 2, and simulation. It is a figure which shows the star-shaped register mark used as the object of Example 1 of simulation 3, and the verification result of simulation. It is a figure which shows the verification result of the star-shaped register mark used as the object of Example 2 of the simulation 3, and a simulation. It is a figure which shows the star-shaped register mark used as the object of Example 1 of simulation 4, and the verification result of simulation. It is a figure which shows the star-shaped register mark used as the object of Example 2 of simulation 4, and the verification result of simulation.

  As described above, the register mark according to the present invention has a star shape.

  FIG. 1 is a plan view showing a register mark according to the first embodiment of the present invention. As shown in FIG. 1, the register mark according to the first embodiment of the present invention has a five-leg star shape.

  The number of legs in a star-shaped figure is not limited to five. Any odd number of 5 or more can be selected.

  FIG. 2 is a plan view showing another example of the register mark according to the first embodiment of the present invention. As shown in FIG. 2, for example, the register mark according to the first embodiment of the present invention can be configured from a seven-legged star shape.

  The reason why the star shape is selected as the register mark shape will be described below.

  For example, assume an N-gon having a predetermined positional relationship between N sides. In such an N-gon, if one line segment meaning at least one side can be completely defined and which side of the N-gon can be defined, This N-gon is theoretically identifiable. This means that (N−1) of the N line segments forming the N-gon are redundant in identifying the N-gon.

  However, this is in a mathematical sense. In reality, a line segment that constitutes a certain figure, that is, two end points (both ends of the line segment) are directly defined by physical observation. Is impossible. This is because the line segment is originally defined from two end points, but due to the resolution limit or blur of the collected image, the exact coordinates of the N-shaped vertex to be the end point cannot be obtained.

  For this reason, instead of the end point, a straight line that can be defined as an extension of each side is used. This is expressed as a linear expression. In addition, two straight lines define one intersection by viewing them as simultaneous equations. This intersection represents one vertex of the N-gon. To define an edge using this method, it is necessary to define two intersection points, so to define an end point (ie two points), at least three straight lines are required. To do. This means that, for example, when a triangle is used as a basic figure, there is no redundancy at all.

For this reason, the maximum redundancy R in the N-gon is given by
R = (N−3)

  As will be described below, for example, when there is a missing part on any side, immunity against the missing part can be improved by assigning this redundancy to that side.

  From the viewpoint of providing redundancy, the characteristics required for a figure suitable for use as a register mark are listed as follows.

(1) Having a sufficient number of sides to obtain redundancy, that is, R = N−3 is 1 or more, that is, N is 4 or more (a quadrilateral or more).

(2) N is an odd number Since an odd-numbered square does not have a pair of sides that are parallel to each other, the method of identifying the position of a register mark (described later) does not identify the side twice (this point) Will be described later).

(3) The length of one side is as large as possible as compared with the pinching diameter (representative diameter of the figure). It is narrowed down to three of a heptagon and a nine-sided shape. When the condition (3) is evaluated for these three N-gons, the following Table 1 is obtained.

The pinching diameter D in Table 1 is as shown in FIGS. The side length E is defined by the following equation.
E = LS

  The pinching diameter D and the side length E in Table 1 are calculated values when the distance from the center of gravity to the apex, that is, the representative radius is 0.5.

  V indicates the “number of cases” where each side is detectable / impossible.

  From a practical standpoint, nine-sided polygons with poor side length efficiency and an excessive number of cases are excluded from the candidates, and only pentagons and heptagons are considered below.

  Next, for pentagons and heptagons, the star shape, which is a modification thereof, is evaluated. An N-leg star shape can be obtained by extending each side of a regular N-angle to both sides.

  Table 2 is a table in which image efficiency, that is, E / D is calculated for pentagons and heptagons.

  From Table 2, it can be concluded that the figure having the longest effective side length with respect to the representative outer dimension, that is, the figure having the highest image utilization efficiency is a five-legged star (hereinafter referred to as “five-legged star”). Simply called "star").

There are 2 ⁵ = 32 possible combinations of edges in the star shape. The number of intersection points (that is, the number of convex vertices and concave vertices in the star shape) is determined according to the detected sides. Table 3 shows the combinations of detectable edges in the star shape.

  In Table 3, the meaning of each symbol is as follows.

P: Combination number af: Five sides of star (see FIG. 3)
S: Number of detected edges D: “○” when the minimum intersection point for estimating the star shape can be defined, and “×” when it cannot.

  The five sides af of the star shape are defined as shown in FIG. However, each side of the pentagon inscribed in the star shape is virtual (that is, the pentagon is composed of line segments that do not appear in the actual figure (and therefore cannot be detected)), and the length of the side Not included.

  FIG. 4 shows all the combinations of D-TRUEs using the results defined as described above.

  FIG. 4 (A) shows combinations in which three of the five sides constituting the star are obtained (combination numbers 7, 11, 13, 14, 19, 21, 22, 25, 26, 28). 4 (B) shows a combination in which four of the five sides constituting the star are obtained (combination numbers 15, 23, 27, 29, 30). In addition, the combination (combination number 31) from which all five sides constituting the star shape were obtained was omitted.

  In FIG. 4, “◯” indicated by A represents a convex end point, and “◯” indicated by B represents a concave end point.

[Register mark position identification device]
An example of an apparatus and method for identifying the position of a star-shaped register mark is shown below.

  First, FIG. 14 shows an example of a registration control device that performs registration control using register marks.

  FIG. 14 is a block diagram of the registration control apparatus 1000.

  The registration control apparatus 1000 includes a register mark position identification apparatus 100 (described later) for identifying the position of a register mark according to an embodiment of the present invention, a comparator 200, a memory 300, and a control apparatus 400. Yes.

  The memory 300 stores the position where the register mark 600 should be on the sheet 500, that is, the coordinates of the reference position of the register mark 600.

  A signal indicating the position of the register mark 600 identified by the register mark position identifying apparatus 100 is transmitted to the comparator 200.

  The comparator 200 receives a signal indicating the reference position from the memory 300 and compares the position of the register mark 600 received from the register mark position identifying apparatus 100 with the reference position.

  The comparator 200 transmits a comparison signal indicating the comparison result to the control device 400. The control device 400 corrects the position of the sheet 500 according to the shift of the position of the register mark 600 from the reference position indicated by the comparison signal.

  As described above, the register mark position identification device 100 is a constituent element of the registration control device 1000.

  Hereinafter, a register mark position identification device 100 for identifying the position of a register mark according to an embodiment of the present invention will be described.

  FIG. 5 is a schematic diagram showing a register mark position identification apparatus 100 for identifying the position of a register mark according to the embodiment of the present invention.

  As shown in FIG. 5, the sheet 500 to be registered is fed in the direction of arrow A. A plurality of register marks 600 are printed on the sheet 500 at regular intervals, and each register mark 600 is printed on the sheet 500 in the same direction. That is, the directions of the plurality of sides constituting the register mark 600 are determined in advance, and the positional relationship between the plurality of sides constituting the register mark 600 is known.

  The register mark position identification device 100 includes an imaging device 120 that images the register mark 600 on the fed sheet 500, an image memory 130 that stores image data of the register mark 600 captured by the imaging device 120, and an image memory 130. The scanning device 110 that scans in parallel with each side constituting the register mark 600 on the image data stored in the image data and moves the scanning direction in a direction orthogonal to each side, and based on the scanning result And an analysis device 140 that identifies the position of each side of the register mark 600 and thereby identifies the position of the register mark 600.

  FIG. 6 is a block diagram of the scanning device 110.

  As shown in FIG. 6, the scanning device 110 includes a normal scan execution device 111 and a sub scan execution device 112.

  The normal scan execution device 111 performs a scan in parallel with each side constituting the register mark 600. The sub-scan execution device 112 moves the scan performed by the normal scan execution device 111 at regular intervals in a direction orthogonal to each side constituting the register mark 600 and then performs the scan. In other words, the normal scan execution device 111 and the sub scan execution device 112 perform a plurality of scans in parallel with each side constituting the register mark 600 and at a certain distance from each side.

  FIG. 7 is a block diagram of the analysis device 140.

  The analysis device 140 includes a central processing unit (CPU) 1401, a first memory 1402, a second memory 1403, an input interface 1404 for inputting various instructions and data to the central processing unit 1401, and a central processing unit. An output interface 1405 for outputting the result of the processing executed by 1401; and a bus 1406 for connecting the central processing unit 1401 to the first memory 1402, the second memory 1403, the input interface 1404, and the output interface 1405. Has been.

  Each of the first memory 1402 and the second memory 1403 includes a read-only memory (ROM), a random access memory (RAM), a semiconductor storage device such as an IC memory card, a storage medium such as a flexible disk, and a hard disk Or an optical magnetic disk or the like. In the present embodiment, the first memory 1402 is a ROM, and the second memory 1403 is a RAM.

  The first memory 1402 stores various control programs to be executed by the central processing unit 1401 and other fixed data. The second memory 1403 stores various data and parameters, and provides an operation area for the central processing unit 1401, that is, data temporarily required for the central processing unit 1401 to execute a program. Is stored.

  The central processing unit 1401 reads a program from the first memory 1402 and executes the program in the operation area of the second memory 1403. That is, the central processing unit 1201 operates according to a program stored in the first memory 1402.

  To simplify the explanation, it is assumed that the register mark 600 is a right triangle as shown in FIG. That is, the register mark 600 includes three sides 601, 602, and 603, two sides 601 and 602 form a right angle, and the other side 603 forms a hypotenuse.

  The register mark position identification device 100 identifies the position of the register mark 600 as follows.

  First, an arbitrary point inside the register mark 600 is defined as the reference point 630. The coordinates of the reference point 630 are represented by (0, 0).

  First, attention is paid to, for example, the side 601 among the three sides 601, 602, and 603.

  A straight line (a straight line parallel to the side 601) S having a vector equal to the vector of the side 601 is translated from the reference point 630 (0, 0) in the direction of the perpendicular 601A to the side 601. Record the distance r.

  Next, the length m (described later) at which each straight line S intersects the side 601 is recorded for each distance r.

  Among the distances r thus obtained, a distance r (a) (described later) that maximizes the length m is determined. This distance r (a) is a parameter indicating the position of the side 601 of the register mark 600.

  The above processing is specifically performed as follows.

  FIG. 8 is a schematic diagram showing a situation in which scanning is performed on the side 601 of the register mark 600.

  When the sheet 500 is fed in the direction of arrow A, the imaging device 120 sequentially captures each of the plurality of register marks 600 printed on the sheet 500 and the image data of each register mark 600 is stored in the image memory 130. Send. The image memory 130 stores the image data of the register mark 600 received from the imaging device 120.

  Next, as described below, scanning by the scanning device 110 is performed for each of the plurality of register marks 600 continuously moving in the direction of the arrow A as if each register mark 600 is stationary. Is called. As is clear from the above, the scanning of the register mark 600 means scanning of the image data of the register mark 600 stored in the image memory 130.

  The normal scanning execution device 111 of the scanning device 110 is parallel to the side 601 of the register mark 600 with respect to the image data of the register mark 600 stored in the image memory 130 (that is, orthogonal to the direction in which the sheet 500 is fed). Scanning is performed at regular intervals.

  For example, as shown in FIG. 8, the normal scan execution device 111 sequentially emits scanning lines 111 a, 111 b, 111 c, 111 d, and 111 e (corresponding to the above-described straight line S) parallel to the side 601 of the register mark 600. In this case, the sub-scan execution device 112 maintains the intervals between the scanning lines 111a, 111b, 111c, 111d, and 111e to be constant.

  The analysis device 140 accesses the scanning device 110 and analyzes the result of scanning the register mark 600 by the scanning device 110.

  FIG. 9 is a schematic diagram showing a result of analysis performed by the analysis device 140 on the image data of the scanned register mark 600.

  As described above, when viewed at the pixel level, as shown in FIG. 9A, each side 601, 602, 603 of the register mark 600 is actually not a straight line but a waveform line.

  As shown in FIG. 9A, the scanning device 110 sequentially runs the scanning lines 111a, 111b, 111c, 111d, and 111e in the scanning direction Y (direction parallel to the side 601) at a constant interval X.

  By scanning the register mark 600, the number of pixels corresponding to the length of the region where the scanning lines 111a, 111b, 111c, 111d, and 111e intersect the register mark 600 is detected.

  FIG. 9B is a graph showing the number of detected pixels on the vertical axis and the position on the horizontal axis in the direction orthogonal to the side 601 of the register mark 600 (direction parallel to the side 602).

  When the number of detected pixels is plotted according to the position of the detected pixel, a graph 121 as shown in FIG. 9B is obtained. This graph 121 has a peak (maximum value) 122. That is, the number of detected pixels shows a peak 122 at a position closest to the side 601.

  FIG. 9C is a graph 121A obtained by differentiating the values of the number of pixels shown in FIG. 9B and plotting the obtained differential values.

  By differentiating the number of pixels, the peak 122A is more clearly shown as shown in FIG. 9C. The position of the peak 122 or the peak 122A (the position of the peak 122 and the position of the peak 122A is substantially the same) (that is, the distance r (a) at which the length m is maximum) indicates the position of the side 601 of the register mark 600. ing.

  FIG. 10 is a schematic diagram showing a result of analysis performed by the analysis device 140 on the image data of the scanned register mark 600 by a method different from the method shown in FIG.

  By applying Laplacian, for example, to the register mark 600, an outline figure 161 of the register mark 600 is obtained as shown in FIG.

  FIG. 10B is a graph obtained when scanning is performed on the contour figure 161 of the register mark 600 shown in FIG. The graph shown in FIG. 10B is equivalent to the graph shown in FIG.

  Next, as illustrated in FIG. 11, the scanning device 110 scans the image data of the side 602 in the same manner as in the case of the side 601. The analysis device 140 analyzes the result of scanning by the scanning device 110 in the same manner as in the case of the side 601 and identifies the position of the side 602.

  Further, as illustrated in FIG. 12, the scanning device 110 scans the image data of the side 603 in the same manner as in the case of the side 601. The analysis device 140 analyzes the result of scanning by the scanning device 110 in the same manner as the case of the side 601 and identifies the position of the side 603.

  As described above, the analysis device 140 analyzes the result of scanning performed by the scanning device 110 on the image data captured by the imaging device 120, whereby the three sides 601, 602, and 603 constituting the register mark 600 are analyzed. Is determined.

When the three sides 601, 602, and 603 are obtained in this way, three intersection points 701, 702, and 703 (see FIG. 8 ) at which these three sides 601, 602, and 603 intersect can be defined. . These three intersections 701, 702, and 703 indicate the three vertices of the register mark 600.

  Next, the analysis device 140 virtually draws a perpendicular line from the center of two adjacent vertices, that is, the vertices at both ends of one side. For example, when vertices 701 and 702 are selected as two vertices, a perpendicular line of the side 601 is drawn from the midpoint of the side 601 having the vertices 701 and 702 as both ends.

  Similarly, with respect to the remaining two sides, a perpendicular line is virtually drawn from each middle point.

  A total of three perpendicular lines are drawn. The intersection of these three perpendiculars is the center of the register mark 600 (in this case, the center is synonymous with the center of gravity).

  As described above, the positions of the three sides 601, 602, and 603 constituting the register mark 600 are specified, then the position of each vertex of the register mark 600 is specified, and the center of the register mark 600 is specified. . This means that the position of the register mark 600 on the sheet 500 has been identified.

  According to the register mark position identification device 100, the following advantages can be ensured.

  First, great immunity is guaranteed against the lack of shape of the register mark 600.

  FIG. 13 is a schematic diagram showing the state of register mark scanning in the conventional registration control.

  In the conventional registration control, for example, as shown in FIG. 13A, the position of the register mark 600 is specified only by one scanning line 610 passing through the register mark 600. When the scanning line 610 intersects the register mark 600, a binary pulse 614 (pulse width is B) corresponding to two points 611 and 612 where the scanning line 610 and the register mark 600 intersect is created.

  For example, as shown in FIG. 13B, it is assumed that a chip 620 is generated in the outer shape of the register mark 600.

  When the scan line 610 passes through the chip 620, the scan line 610 intersects the register mark 600 at two points 613 and 612. One of the two points 613 is a point located on the chip 620. Therefore, a binary pulse 615 (with a pulse width A) corresponding to two points 613 and 612 where the scanning line 610 and the register mark 600 intersect is created.

  Due to the chipping 620 of the register mark 600, the width A of the binary pulse 615 is smaller than the width B of the binary pulse 614. For this reason, the position of the register mark 600 specified based on the binary pulse 615 is not accurate.

  On the other hand, according to the register mark position identification device 100, since scanning is performed along each side constituting the register mark 600, it is assumed that the outer shape of the register mark 600 has a missing shape such as a chip 620. However, the specification of the position of each side of the register mark 600 is not affected by the chip 620. That is, it is possible to accurately specify the positions of the sides 601, 602, and 603 regardless of the presence or absence of the chip 620.

  As described above, in the register mark position identification device 100, even if a void (hole) or a dent (chip) exists in the register mark 600, they do not cause an error in specifying the position of the register mark 600.

  Second, since the noise at the sides of the register mark 600 is averaged over the entire side, the error is substantially reduced.

  Third, an error in the position of the register mark 600 can be reduced.

  In the register mark position identification device 100, the position of the register mark 600 is specified by specifying the positions of the three sides 601, 602, and 603. For this reason, even if the register mark 600 itself lacks the completeness of the shape as a triangle (that is, even if the shape as a triangle is missing or redundant), the position error is averaged over the entire figure. Is reduced.

  In general, the position identification of the register mark 600 when the register mark 600 is an N-gon is performed according to the following steps.

(1) First, one test plane expressed in four quadrants is prepared. The coordinates of an arbitrary point on this plane are global coordinates. Therefore, when a figure (register mark 600) whose position is to be determined is overlaid on this plane, the global coordinates of the reference position (for example, the center of the figure) of this figure become a parameter indicating the position of the figure. The displacement from the reference position indicated by this parameter represents a position error.

  First, an arbitrary point inside the N-gon is defined as a reference point. Let the coordinates of this reference point be (0, 0).

  Next, a straight line S having a vector equal to the vector of each side is translated from the reference point (0,0) in the direction of the perpendicular to each of the N sides of the N-gon, and the reference point (0,0) The distance r from each to the straight line S is recorded.

(2) Next, the length m at which each straight line S intersects each side of the N-gon for each distance r is recorded.

(3) The above processes (1) and (2) are repeated within the range of 0 <r <R (R is a physical limit of the global coordinate plane).

(4) Of the distances r thus obtained, the distance r (a) at which the length m is maximum is determined. This distance r (a) is a parameter indicating the position on the global coordinate of each side of the N-gon.

(5) N distances r (a) are obtained by repeating the processes (1) to (4) for all the perpendiculars.

(6) N intersection points can be defined by the N distances r (a) thus determined. These intersection points indicate N vertices of the N-gon.

(7) A perpendicular line is drawn from the center of two vertices adjacent to each other, that is, from the center of the vertices at both ends of one side. Since the N-gon has N sides, N vertical lines are formed.

(8) Any two vertical lines out of _N vertical lines form _N C ₂ intersections. For example, the number of intersection points is 3 when N = 3 (triangle).

(9) If the N-gon is perfect as a figure and the sampling of the distance r is dense, the coordinates of the _N C _two intersections match. However, the above calculated coordinates and the actual figure coordinates generally do not match. In such a case, the accuracy of each coordinate can be improved by performing a weighted average of each coordinate after performing a reliability evaluation depending on the process of obtaining each coordinate.

  As described above, in the register mark position identification apparatus 100, whatever data is used as the register mark 600, the data to be used is only data on a line indicating the outer shape (outline) of the register mark 600. Therefore, even if the register mark 600 is a solid figure or a contour (hollow) figure, the operation of the register mark position identification apparatus 100 does not change. For this reason, even if a void exists in the register mark 600, the void does not cause an error in the position identification of the register mark 600.

[Simulation 1]
A simulation was conducted to show the effectiveness as a star-shaped register mark.

  When performing a simulation, it is necessary to accurately define a graphic area for a star-shaped register mark.

  A graphic region definition is a kind of function. The input of this function is a coordinate, and the output is indicated by IN or OUT. IN means that the coordinates are inside the boundary, and OUT is outside the boundary.

(1) Region division Star-shaped division can be performed in any way as long as each region can be expressed in monomial form. The basis for representing each region as a monomial is that each region is represented by an N-gon. Here, in order to reduce the number of terms and because all regions can be expressed by three straight lines, four divisions that divide the star shape into four triangles L, M, N, and P are adopted as shown in FIG.

(2) Segmented area definition Graphic standard: Center of gravity circumscribed circle radius: 1
Angular unit: xπ radians Input coordinates: [x, y]
Straight line a: ya = −sin (1/10)
Straight line b: yb = −1 + tan (2/5) × (−x)
Straight line c: yc = (3/10) cos (1/5) -tan (1/5) × (−x)
Straight line d: yd = (3/10) cos (1/5) -tan (1/5) × x
Straight line e: ye = −1 + tan (2/5) × x

At this time, if ((y> ya) && (y <yc) && (y <yd)) L = true; else L = false;
if ((y <ya) &&(y> yb) &&(y> ye)) M = true; else M = false;
if ((x <0) &&(y> yb) && (y <yd)) N = true; else N = false;
if ((x> 0) && (y <yc) &&(y> ye)) P = true; else P = false;
if (L || M || N || P) prot (1); else prot (0);

(3) Verification FIG. 16A illustrates the boundary expression, and FIG. 16B plots the determination of the inside / outside of the boundary on a matrix.

  Since the matrix size is 201 × 201, the coordinates of x and y are −100 to +100. The radius of the star (the radius of the circumscribed circle) was set to 100. The star-shaped center of gravity is given an offset of 5 pixels downward from the center of the matrix. Therefore, the coordinates of the star-shaped heart finally determined are (0, 5).

  In this way, it was confirmed that the region defining formula was correct.

  Hereinafter, a simulation performed for the star-shaped register mark according to the present embodiment will be described.

  Here, a binary image is treated as a figure.

The steps to identify a star-shaped heart are as follows.
(1) Detect each edge as an image (2) Sweep with a scanning line in a certain direction to determine an estimated boundary position as a figure (3) Express a straight line indicating the estimated boundary position as a linear expression (4) Estimate the coordinates of each vertex of a virtual inscribed pentagon from the intersection of two straight lines expressed by a linear expression. (5) Define a perpendicular to the side from the midpoint between two adjacent vertices. (6) Calculate the intersection coordinates (maximum 10) of any two perpendiculars (7) Calculate the average value of the intersection coordinates and use this as the heart

  Define Laplacian for binary images to substitute edges for boundaries. This is a pixel present at the boundary between the white area and the black area in FIG.

  Next, a straight line with a predefined slope is moved from the center of the test plane along a vector indicating the orientation of each star foot, and pixels that have been previously found to have positive Laplacian (ie, The degree of coincidence between the white region and the pixel existing at the boundary of the black region in FIG. The position where the matched pixel is the maximum within the entire movement range is stored.

  Since the coordinates of the plane expressed by the pixels are represented by integers, the vector is represented by a real number, and therefore, not all pixels indicating edges hit the mathematical formula.

Essentially necessary data obtained by this processing is the distance from the center of the pixel plane to each side. Mathematically, this value is sin (π / 10) × 100 = 30.90
It becomes. The actual detection results are distributed between 25 and 35 as shown in Table 4 below.

  For confirmation, a linear expression is defined again using these data and plotted on a plane.

  The distances from the plane center obtained for the straight lines a to e are ta to tb. If the figure is an isotropic star, the number of coefficients used to define the line is finite. They are defined in advance as follows using trigonometric functions. The angle unit is (× π radians).

sin (n / 10) = ksn For example, sin18 ° = sin (1 × π / 10) = ks1
cos (n / 10) = kcn For example, sin 36 ° = sin (2 × π / 10) = kc 2
tan (n / 10) = ktn For example, tan 72 ° = sin (4 × π / 10) = kt 4

  At this time, the following expressions are determined for each side of the star shape.

(Side a)
Basic formula y = 0
Correction formula y = ta

(Side b) (See FIG. 17)
Basic formula y = −kt4 × x
ib = tb / ks1
Correction formula y = tb / ks1-kt4 × x

(Side c) (See FIG. 18)
Basic formula y = kt2 × x
ic = − (1 / kc2) × tc
Correction formula y = −tc / kc2 + kt2 × x

(Side d)
Since side d is a mirror image of side c, the following equation is obtained.

Basic formula y = −kt2 × x
id = − (1 / kc2) × td
Correction formula y = −td / kc2−kt2 × x

(Side e)
Since side e is a mirror image of side b, the following equation is obtained.

Basic formula y = kt4 × x
ie = te / ks1
Correction formula y = te / ks1 + kt4 × x

  Next, a midpoint between two adjacent vertices is detected.

  First, the coordinates of the star-shaped concave portion, that is, the vertex of the inscribed regular pentagon are obtained. It is equivalent to obtaining the intersection of two adjacent straight lines of a linear expression representing each side.

  Let Xab be the intersection of the straight lines indicating side a and side b. Similarly, Xbc, Xcd, Xde, and Xea are defined. Finding the coordinates of each vertex is equivalent to solving the following simultaneous equations.

(1) About Xab y = ta
y = tb / ks1-kt4 * x

From these two equations, ta = tb / ks1-kt4 × x
0 = tb / ks1-kt4 * x-ta
kt4 * x = tb / ks1-ta
x = (tb / ks1-ta) / kt4

(2) About Xbc y = tb / ks1-kt4 × x
y = −tc / kc2 + kt2 × x

From these two formulas, tb / ks1-kt4 × x = −tc / kc2 + kt2 × x
-Kt4 * x-kt2 * x = -tc / kc2-tb / ks1
kt4 * x + kt2 * x = tc / kc2 + tb / ks1
x (kt4 + kt2) = (tc / kc2 + tb / ks1)
x = (tc / kc2 + tb / ks1) / (kt4 + kt2)

(3) About Xcd y = −tc / kc2 + kt2 × x
y = td / kc2-kt2 * x

From these two expressions, −tc / kc2 + kt2 × x = −td / kc2−kt2 × x
2 × kt2 × x = (− td + tc) / kc2
x = (− td + tc) / (kc2 / 2) × kt2

(4) About Xde y = −td / kc2−kt2 × x
y = te / ks1 + kt4 × x

From these two expressions, −td / kc2−kt2 × x = te / ks1 + kt4 × x
-Kt2 * x-kt4 * x = te / ks1 + td / kc2
x =-(te / ks1 + td / kc2) / (kt2 + kt4)

(5) About Xea Since Xea is a mapping of Xab,
y = ta
x =-(te / ks1-ta) / kt4

  Next, the midpoint coordinates Ma to Me are obtained from the two sets of vertex coordinates thus obtained. They are calculated as the arithmetic mean of the coordinates of the two vertices that sandwich the midpoint.

  The coordinates of the determined midpoint are shown by cross plots C1-C5 in FIG. Since the coordinates of the midpoint are expressed as correct, the vertex coordinates of the inscribed pentagon in the previous stage may be calculated as correct. In FIG. 19, the coordinates of each vertex are indicated as Pentagonal_Cros.

  Next, from each midpoint Ma-Me, a perpendicular to the side where it exists is defined. Here, the characters x, y, xm, and ym are used in common to all the perpendiculars for each perpendicular parameter.

(Perpendicular A)
Basic x = 0
Correction x = xm

(Vertical B)
Basic y = kt1 × x
Correction y = kt1 × (x−xm) + ym

(Vertical C)
Basic y = kt3 × x
Correction y = kt3 × (−x + xm) + ym

(Vertical D)
Basic y = kt3 × x
Correction y = kt3 × (x−xm) + ym

(Vertical E)
Basic y = kt1 × (−x)
Correction y = kt1 × (−x + xm) + ym

  FIG. 20 shows the five obtained vertical lines. These perpendiculars were almost proven to be defined correctly.

  Five perpendicular lines define ten intersections in principle. Ideally these 10 intersections have exactly the same coordinates. Therefore, ideally, a star-shaped heart can be defined with only two perpendicular lines.

  The calculation of the heart is performed again by solving the linear simultaneous equations. These equations use the slope and midpoint coordinates that are predetermined for each perpendicular. Here, the coordinates of each midpoint are expressed as cv [p].

  c is a fixed character, and v indicates each side a to e. p indicates a coordinate component, and 0 = x and 1 = y.

(1) XAB (intersection defined by perpendicular A and perpendicular B)
vx = ca [0]
vy = kt1 (vx−cb [0]) + cb [1]
fx = vx
fy = kt1 (ca [0] -cb [0]) + cb [1]

(2) XBC (intersection defined by perpendicular B and perpendicular C)
vy = kt1 (vx−cb [0]) + cb [1]
vy = −kt3 (−fx + cc [0]) + cc [1]
kt1 (vx−cb [0]) + cb [1] = kt3 (−vx + cc [0]) + cc [1]
kt1 * vx-kt1 * cb [0] + cb [1] =-kt3 * vx + kt3 * cc [0] + cc [1]
kt1 * vx + kt3 * vx = kt3 * cc [0] + cc [1] + kt1 * cb [0] -cb [1]
fx = (kt3 × cc [0] + cc [1] + kt1 × cb [0] −cb [1]) / (kt1 + kt3)
fy = kt1 (fx−cb [0]) + cb [1]

(3) XCD (intersection defined by perpendicular C and perpendicular D)
vy = kt3 ・ (-vx + cc [0]) + cc [1]
vy = kt3 ・ (fx−cd [0]) + cd [1]
kt3 ・ fx−kt3 ・ cd [0] + cd [1] = -kt3 ・ fx + kt3 ・ cc [0] + cc [1]
kt3 ・ fx + kt3 ・ fx = kt3 ・ cc [0] + cc [1] + kt3 ・ cd [0] −cd [1]
fx = cc [0] + cd [0] + (cc [1] −cd [1]) / (2 ・ kt3)
fy = kt3 ・ (-fx + cc [0]) + cc [1]

(4) XDE (intersection defined by perpendicular D and perpendicular E)
vy = kt3 ・ (vx−cd [0]) + cd [1]
vy = kt1 ・ (-vx + ce [0]) + ce [1]
kt3 ・ vx−kt3 ・ cd [0] + cd [1] = -kt1 ・ vx + kt1 ・ ce [0] + ce [1]
vx (kt3 + kt1) = kt3 ・ cd [0] −cd [1] + kt1 ・ ce [0] + ce [1]
fx = (kt3 ・ cd [0] −cd [1] + kt1 ・ ce [0] + ce [1]) / (kt3 + kt1)
fy = kt3 (fx−cd [0]) + cd [1]

(5) XEA (intersection defined by perpendicular E and perpendicular A)
vy = kt1 ・ (-vx + ce [0]) + ce [1]
vx = ca [0]
fx = vx
fy = kt1 ・ (-ca [0] + ce [0]) + ce [1]
Hereinafter, calculation is performed in the same manner, and only the result is shown.

(6) XAC (intersection defined by perpendicular A and perpendicular C)
fx = ca [0]
fy = kt3 ・ (-fx + cc [0]) + cc [1]

(7) XCE (intersection defined by perpendicular C and perpendicular E)
fx = (kt1 ・ ce [0] + ce [1] −kt3 ・ cc [0] −cc [1]) / (kt1−kt3)
fy = kt3 ・ (-fx + cc [0]) + cc [1]

(8) XEB (intersection defined by perpendicular E and perpendicular B)
fx = (-kt1 ・ cb [0] + cb [1] −kt1 ・ ce [0] −ce [1]) / (− 2 ・ kt1)
fy = kt1 ・ (-fx + ce [0]) + ce [1]

(9) XBD (intersection defined by perpendicular B and perpendicular D)
fx = (-kt3 ・ cd [0] + cd [1] + kt1 ・ cb [0] −cb [1]) / (kt1−kt3)
fy = kt1 ・ (fx−cb [0]) + cb [1]

(10) XDA (intersection defined by perpendicular D and perpendicular A)
fx = ca [0]
fy = kt3 ・ (fx−cd [0]) + cd [1]

  In FIG. 21, the final coordinates as the calculation result are displayed as “Determ'd Co_ord”.

  As described above, the expected coordinate value of the heart in this simulation is (0, 5). On the other hand, the evaluation result of this time is almost perfectly matched when the discretization error is taken into consideration.

  What is particularly noteworthy is that any combination of two perpendiculars alone satisfies the expected value. This means that the star-shaped register mark theoretically includes redundancy.

  Hereinafter, a simulation in the case of identifying a heart for a star shape that is not perfect as a figure will be exemplified.

[Simulation 2]
In the simulation 2, as shown in FIG. 22, the case where a defect occurs in two of the five star-shaped legs is taken up.

  Such a graphic defect causes a large error in a conventional heart detection mechanism, for example, a detection mechanism for obtaining the center of gravity or the center.

  Thereafter, the same analysis as in the simulation 1 is performed on the star-shaped register mark having a defect shown in FIG.

  When the star-shaped figure shown in FIG. 22 is compared with the star-shaped figure shown in FIG. 16B, the number of pixels that match the expected tangent is reduced by the amount of shortening the predetermined side length. The distance from the center is recognizable in the same manner as in simulation 1 (this indicates that no further study is necessary in the original case, but for clarity of explanation, Continue to show simulation results).

  FIG. 23 shows the result of detecting the midpoint between the two vertices, setting and confirming the perpendicular definition formula, and calculating the final intersection as in the case of the simulation 1.

  As is clear from the result shown in “Determ'd Co_ord” in FIG. 23, it was confirmed that a deficit of about two feet as shown in FIG. 22 has no influence on the heart identification of the star-shaped register mark. .

[Simulation 3]
(Example 1)
In Example 1 of the simulation 3, as shown in FIG. 24, a star shape in which partial collapse occurs is targeted.

  For example, a star-shaped figure as shown in FIG. 24 is generated by generating a random size and position, performing erosion + dilation processing on the original binary image a considerable number of times, and intentionally performing partial destruction. Is obtained.

  Also for the star shape shown in FIG. 24, the reference position (heart) is detected according to the same procedure as in simulation 1.

  FIG. 24 shows the results of detecting the midpoint between two vertices, setting and confirming a perpendicular definition formula, and calculating the final intersection as in the case of simulation 1.

  As shown in “Determ'd Co_ord” in FIG. 24, the final error of the reference position is less than one pixel even for a star shape that is partially collapsed.

(Example 2)
In the second example of the simulation 3, as shown in FIG. 25, a star shape whose parameters are adjusted so that a void and a remote island different from the first example can be formed is targeted. The overall collapse is even greater than in Example 1. Example 2 assumes a figure irregularity when actual printing is performed on paper with poor smoothness and then enlarged.

  FIG. 25 shows the result of detecting the midpoint between the two vertices, setting and confirming the perpendicular definition formula, and calculating the final intersection as in the case of the simulation 1.

  As shown in “Determ'd Co_ord” in FIG. 25, the detection error is less than one pixel in both the X and Y directions.

[Simulation 4]
In the simulation 4, a star shape that is highly collapsed is targeted.

  Examples 1 and 2 below are examples of detecting a reference position in a star shape that has collapsed to the extent considered to be a limit.

(Example 1)
In Example 1 of the simulation 4, as shown in FIG. 26, the star-shaped basic figure is not only extremely deformed, but also involves a division of the figure.

  As in the case of simulation 1, FIG. 26 shows the results of detecting the midpoint between two vertices, setting and confirming a perpendicular definition formula, and calculating the final intersection.

  As shown in “Determ'd Co_ord” in FIG. 26, the final error of the reference position is less than one pixel even for a star shape that is highly collapsed.

(Example 2)
In Example 2 of the simulation 4, as shown in FIG. 27, the star shape is largely collapsed to such an extent that it is hardly recognized as a star shape.

  As in the case of simulation 1, FIG. 26 shows the results of detecting the midpoint between two vertices, setting and confirming a perpendicular definition formula, and calculating the final intersection.

  As shown in “Determ'd Co_ord” in FIG. 27, even if the star shape has collapsed to such an extent that it is hardly recognized as a star shape, the final error of the reference position is about 1 pixel.

  As described above, according to the simulation 1-4, it was shown that the reference position (heart) can be identified very accurately by using the star-shaped register mark.

100 Register Mark Position Identification Device 110 Used for Register Marks According to Embodiment of the Present Invention 110 Scanning Device 111 Normal Scan Execution Device 112 Sub Scan Execution Device 120 Imaging Device 121 Graph 130 Image Memory 140 Analysis Device 500 Sheet 600 Register Mark 601, 602, 603 Register mark three sides 601A, 602A, 603A Register mark three sides perpendicular 630 Reference point 1000 Registration control device 200 Comparator 300 Memory 400 Control device

Claims (4)

A register mark position identification method for reading the position of a register mark printed on a sheet,
The direction of each of the plurality of sides constituting the register mark is predetermined,
The register mark position identification method is:
A first step of printing a star-shaped register mark on the sheet;
A second step of imaging the register mark on the sheet that has been fed;
The image data of the register mark imaged in the first process is scanned in parallel with the respective sides constituting the register mark, and the scanning direction is set in a direction orthogonal to the respective sides. A third process to move,
A fourth step of identifying the position of each side of the register mark based on the result of the scan, thereby identifying the position of the register mark;
A register mark position identifying method comprising: 2. The register mark position identifying method according to claim 1, wherein the register mark is a five-leg star shape. 3. The register mark position identification method according to claim 1, further comprising a step of storing the image data of the register mark imaged in the second step in an image memory. The register mark position identification according to any one of claims 1 to 3, wherein the position of each side is specified by determining a peak value of the number of integrated pixels obtained by each scan. Method.