JP6180889B2 - Printing inspection device - Google Patents

Description

  The present invention relates to a print inspection apparatus.

  As a background art in this technical field, there is JP-A-2008-97589 (Patent Document 1). This gazette states that “it is an object to provide a technique for accurately cutting out a character region included in a captured image even when noise and dirt in a relatively wide region are generated in the background image. For each pixel position in the character string direction A of the image 61 including the column, a pixel value integrated evaluation value obtained by integrating the pixel values in the character cutout direction B is obtained, and the waveform is represented by the waveform data 62. A first threshold value 63 and a second threshold value 64 are set, and a region where the waveform data 62 exceeds the first threshold value 63 is set as a character candidate region, and the pixel value integrated evaluation value is the second threshold value 64 in the character candidate region. If there is an area exceeding, a character is cut out with the character candidate area as a true character area ”(see summary).

JP 2008-97589 A

  Food and pharmaceutical packaging containers are legally required to display the date of manufacture (including year, month, hour, minute), expiration date (expiration date), and production lot. A printing device such as a printer. Packaging containers are diversified, and printing is also increasing on colored portions other than white background. A print inspection device is used to confirm that printing has been performed correctly by the printing device. The print inspection apparatus captures an image using an imaging device such as a camera, and automatically inspects printed characters for contamination and chipping by collation inspection with a normal pattern registered in advance.

  In this case, the character part must be accurately extracted. Finally, it is necessary to separate, that is, binarize from the captured image data so that the character portion becomes 1, the background portion becomes 0, and so on. Several methods have been proposed for accurately extracting only the character portion and binarizing it. For example, in Patent Document 1, there is a method using a multi-valued image as a method for extracting a character region.

  However, as shown in FIG. 5, when the character 21a is red, the background 21b is green, and there is not much difference in brightness, the character portion 21a and the background portion 21b are favorably separated by the binarization threshold value 21c as shown in FIG. Could not be separated.

  An object of the present invention is to improve the accuracy of print inspection.

In order to solve the above problems, for example, the configuration described in the claims is adopted.
The present application includes a plurality of means for solving the above-mentioned problems. For example, an imaging means for imaging an inspection object, an image data storage means for storing image data captured from the imaging means, and a registration in advance. A printing inspection apparatus comprising: data storage means for storing the normal character data that has been read; and inspection means for comparing the character information extracted from the image data storage means with the registered normal character data to determine whether normalization is defective. Thus, imaging is performed using at least two light components having different wavelengths.

  According to the present invention, the accuracy of print inspection can be improved.

Explanatory drawing of the synthetic lightness concerning the Example of this invention. The histogram of FIG. The inspection object figure of a prior art example. The histogram of FIG. The inspection object figure concerning the Example of this invention. The histogram of FIG. FIG. 6 is a color distribution diagram of FIG. 5. The histogram of FIG. BRIEF DESCRIPTION OF THE DRAWINGS The whole apparatus block diagram concerning the Example of this invention. The hardware block diagram concerning the Example of this invention. The whole processing flow concerning the Example of this invention. The filter adjustment processing flow concerning the Example of this invention. The example of a range designation | designated screen concerning the Example of this invention. The color filter calculation processing flow concerning the Example of this invention. The inspection character setting processing flow concerning the Example of this invention. The inspection processing flow concerning the Example of this invention.

Hereinafter, examples will be described with reference to the drawings.
FIG. 5 shows the inspection object 8, in which the character 21a is red and the background 21b is green. When this is imaged with a monochrome camera, the relationship between the brightness and the number of pixels (hereinafter referred to as a histogram) is as shown in FIG. The character 21a and the background 21b have little difference in brightness, and cannot be separated by the binarization threshold value 21c on the histogram.

  FIG. 7 is a distribution diagram of the result of imaging the inspection object 8 with a color camera. In the RGB three-dimensional space, the character 21a and the background 21b are separated.

  FIG. 8 is a histogram projected on each axis of RGB. This corresponds to a histogram obtained when a red, green and blue color filter is attached to a monochrome camera. In this case, it is understood that the character 21a and the background 21b cannot be separated only by simple information of each color. Therefore, the combination of the monochrome camera and the color filter cannot separate the character 21a and the background 21b.

  In FIG. 1, the axis connecting the center of gravity of the character 21a and the background 21b is set as the combined brightness. When projected onto this combined lightness axis, the histogram of FIG. 2 is obtained. In this case, the character 21a and the background 21b can be separated by the binarization threshold 21c.

  It is known that such an optimal composite axis (synthetic lightness axis) in multiple dimensions can be obtained statistically by multivariate analysis. In this embodiment, the composite brightness axis is obtained by applying principal component analysis.

  In the principal component analysis, a variance-covariance matrix of a portion including the character 21a and the background 21b is obtained, and the maximum eigenvalue is calculated. When the eigenvector for the maximum eigenvalue is obtained, the eigenvector becomes the combined lightness axis direction. If the combined brightness is defined by an eigenvector corresponding to the maximum eigenvalue, the variance of the combined brightness is maximized, and the best contrast can be obtained.

  In the general principal component analysis, the calculation is completed so far, but the following [Equation 1] is used in order to keep the combined brightness within a certain range.

  Where I (i, j) is the combined lightness of the coordinates (i, j), R (i, j), G (i, j), B (i, j) are the pixel coordinates (i, j), respectively. RGB components, fr, fg, and fb are constant coefficient multiples of eigenvectors, and fn is a constant term for adjusting the origin. The best contrast is always obtained by converting the captured color image using fr, fg, fb, and fn as filter coefficients.

  FIG. 9 is an overall configuration diagram of this embodiment. Reference numeral 1 denotes a print inspection device that inspects print characters. A color camera 2 captures an image. This color camera is equipped with filters having peaks at wavelengths of 610 nm, 540 nm, and 455 nm corresponding to red (R), green (G), and blue (B), respectively, and can capture images by separating light of three wavelengths. . A strobe illumination device 3 emits light in accordance with the imaging timing of the color camera in accordance with a light emission command from the print inspection device. Reference numeral 4 denotes a monitor with a touch panel, which displays captured image contents and an operation menu, and can input an operator's instruction through the touch panel. A photoelectric sensor 5 sends a signal to the print inspection device when an inspection object comes. Reference numeral 6 denotes a line control device that controls the conveyor 9 and the discharger 7. 7 is a discharger, which is excluded from the conveyor when the inspection object is defective printing. 8a, 8b, and 8c are inspection objects. Reference numeral 9 denotes a conveyor which conveys the inspection object from left to right at a constant speed.

  When the inspection object 8 conveyed on the conveyor 9 is detected by the photoelectric sensor 5, it is imaged by the color camera 2 in synchronization with the light emission of the strobe 3 after a certain time, and the print content is inspected by the print inspection device 1. . When normal, it flows on the conveyor as it is, and when printing is defective, it is removed from the conveyor by the discharger 7.

  Next, the inside of the print inspection apparatus 1 will be mainly described. FIG. 10 is a hardware configuration diagram. Reference numeral 10 denotes a bus, which connects each device. Reference numeral 11 denotes a CPU which performs all calculations in this apparatus. A ROM 12 stores an initialization program at power-on. Reference numeral 13 denotes a RAM which stores programs executed by this apparatus, intermediate results of operations, image data captured from a color camera, and the like. Reference numeral 14 denotes a serial I / O, through which the CPU 11 exchanges information with the touch panel of the monitor 4 with a touch panel. Reference numeral 15 denotes a camera interface. When an imaging command is issued from the CPU 11, the camera interface 15 performs imaging control of the color camera 2 and transfers the captured image data to the RAM 13 via the bus 10. Reference numeral 16 denotes an interface with an illumination strobe, and the strobe 3 is turned on and off in synchronization with the imaging timing of the color camera 2 according to an issuance command from the CPU 11. Reference numeral 17 denotes a DISK interface, which inputs and outputs data to and from the hard disk 18. The hard disk 18 stores a program executed by the apparatus, print character registration data, various control conditions, and the like. A program to be executed by this apparatus is transferred to the RAM 13 by a program stored in the ROM 12 when the power is turned on, and is executed by the CPU 11. A parallel I / O 19 performs signal input from the photoelectric sensor and exchange of data with the line control device.

  FIG. 11 is an overall process flow of a program executed by the CPU 11. When the power is turned on, initialization is performed at 101. In initialization, the program stored in the hard disk 18 and all print character registration data, control condition setting files, and the like are transferred to the RAM 14. Also, the I / O and interface are initialized, and when completed, an initial menu is displayed on the monitor 4 with a touch panel.

  Next, at 102, input from the touch panel is checked. When the end menu is selected, the process ends. When the adjustment menu is selected, the process proceeds to 103. When the inspection menu is selected, the process proceeds to 106. The processing content of 106 will be described in detail later.

  In 103, the type number is input. Depending on the type number, different setting data such as filter values, binarization threshold values, inspection character setting data, and the like are managed. Next, the routine proceeds to 104. The processing content of 104 will be described later in detail. Thereafter, the process proceeds to 105. The processing content of 105 will also be described in detail later.

FIG. 12 is a processing flow of 104 filter adjustment.
First, the image is taken by the color camera 2 at 201. The captured image information is stored in the RAM 13.

  Next, in 202, a range for calculating the filter is designated using the touch panel. FIG. 13 shows a display screen in a state where a range is designated. The range 110 is designated so as not to include extra prints and stains 102 other than the background portion. The specified range is stored as the coordinate values of the upper left point Ps (is, js) and the lower right point Pe (ie, je).

  At 203, the actual color filter value is calculated. Details of the calculation method will be described later.

  In 204, a composite brightness image is created according to Equation 1 using the filter value obtained in 203 and displayed. For fr, fg, fb, and fn, the filter values obtained in 203 are used.

  Next, in 205, the operator determines whether the displayed composite brightness image is appropriate. If the contrast between the character portion and the background is appropriate, the process proceeds to 206. If the contrast is not sufficient, the process is repeated from 202.

  In 206, a binarization threshold is set for the displayed composite brightness image, and a binary image is displayed.

  Next, in 207, the operator confirms that the character portion and the background are separated into black and white in the displayed binary image. If it is separated, the result of the filter value and the binarization threshold value is stored in correspondence with the product number in 208, and the process ends. If not, repeat 206.

  FIG. 14 is a processing flow of color filter calculation showing details of 203. The filter value is subjected to principal component analysis in three dimensions of RGB, and the filter coefficient is determined based on the first principal component.

  First, at 301, the variance-covariance matrix within the specified range 110 is calculated. The following formula is used in the calculation. First, the sum, square sum, and area of each pixel value are obtained by [Equation 2].

  Next, each element of the variance-covariance matrix is calculated from the obtained value using [Equation 3].

  [Equation 4] A having these elements is the variance-covariance matrix obtained.

  Next, the eigenvalues and eigenvectors of this variance-covariance matrix are obtained from [Equation 5].

  From this, λ is obtained. Here, E is a 3 × 3 unit matrix, and det is a function for obtaining a determinant. Since three solutions are obtained, the one with the maximum absolute value is defined as the final λ (first principal component).

  Next, an eigenvector for this λ is obtained.

  Solve these simultaneous equations to find Fr, Fg, and Fb. This is the eigenvector for the first principal component. When the composite value is obtained by the following [Equation 7], the dispersion within the specified range is maximized by the principle of principal component analysis, and the character and the background can be separated satisfactorily.

  When obtaining the combined lightness, it is necessary to limit the minimum value and the maximum value in order to handle them in a unified manner with other images. In this embodiment, the minimum value is 0 and the maximum value is 255. Therefore, the constant term fn and the magnification k are calculated.

  First, at 303, the maximum value Tmax and the minimum value Tmin of [Expression 7] within the specified range are obtained. Next, at 304, the constant term fn and the magnification k are obtained by the following equations.

  If the relationship between the character and the background black and white is to be exchanged, the constant term fn and the magnification k are obtained by the following equations.

  Finally, at 304, the filter coefficient is finally determined.

  The fr, fg, fb, and fn obtained here are filter values. The determined filter value is stored in the hard disk 18 corresponding to the product number.

  When the combined lightness is calculated by Equation 1 using this filter value, the variance within the specified range is maximized, so that a bright and dark image with the best contrast can be obtained.

  FIG. 15 is an inspection character setting process flow explaining the details of 105.

  First, in 301, an inspection character is input using a touch panel. For example, in the inspection target of FIG. 13, “2013.09.20” is designated on the first line and “Lot N0235” is designated on the second line. However, since “2013.09.20” is the manufacturing date and “N0235” is the manufacturing number, its attributes are also specified.

  Next, in 302, a reference inspection object is set, and the inspection object is imaged. At this time, the strobe 3 is turned on in synchronization.

  Based on the RGB image picked up at step 303, the combined lightness conversion is performed using equation (1). In this embodiment, the entire captured image is converted to create a composite brightness image. When Expression 1 is used, the combined lightness value may be outside the range of 0 to 255, or less than 0 is limited to 0, and 256 or more is limited to 255.

  In 304, the composite lightness image is binarized using the binarization threshold set in 206 to create a binary image. In 305, character segmentation is performed. In 306, it is normalized, and in 307, it is registered as a reference pattern. The registered contents are stored in the hard disk 18 corresponding to the product number.

FIG. 16 is an inspection process flow explaining the details of 106.
First, at 401, the input is confirmed. If end is touched on the touch panel, the inspection process is ended. If there is an input from the photoelectric sensor 5, the process proceeds to 402. If there is no input, this process is repeated.

  In 402, the model number is fetched. In this embodiment, the type number to be inspected is received from the line control device 6. Thereafter, the inspection is performed with the set value corresponding to the imported product number.

  Next, in 403, the filter calculation coefficient corresponding to the designated product number is switched to obtain the optimum filter calculation method for the designated product.

  After waiting for the inspection object to move from the detection position of the photoelectric sensor 5 to the imaging position in 404, the strobe 3 is turned on to take an image.

  In 405, the captured RGB image is converted into a combined brightness using Equation 1, and a combined brightness image is created. As a result, as shown in FIG. 2, a combined brightness image having a histogram in which the character portion and the background are separated is obtained.

  In 406, the combined brightness image is binarized with a threshold corresponding to the product number. A character is cut out at 407, and the character cut out at 408 is normalized.

  Compared to the standard character set in 105 in 409, it is determined that there is no defect if all characters match, and that there is a defect otherwise.

  In 410, it is determined whether there is a defect. If there is a defect, a discharge output is performed to the line control device 6 in 411. Thereby, the line control apparatus 6 drives the discharge machine 7 and excludes defective products from the conveyor.

  In this way, it is possible to inspect a printed character with an optimal contrast for a colored inspection object.

  Further, even when the inspection target is switched, the filter value is optimally set to the inspection target by designating the product number from the line control device 6, so that a good contrast is always obtained.

  In this embodiment, a color camera equipped with RGB filters is used. However, if principal component analysis is used, principal components can be obtained for a distribution of arbitrary dimensions, so that four wavelengths (ie, four colors) can be obtained. The camera can be easily extended to the above-described imaging. The wavelength need not be in the visible range.

  In this embodiment, the combined lightness is calculated from the RGB values using Equation 1, but it is also possible to create a relation table between RGB and the combined lightness in advance and perform conversion using the relation table. It is.

  According to this embodiment, there are the following effects. Since the range for calculating the filter coefficient is specified using the touch panel, it is easy to specify an area including only characters and background. In addition, after calculating the filter coefficients, the composite brightness can be displayed and checked, so it can be confirmed that the filter coefficients are not incorrect. Further, the filter coefficient can be determined so that the calculation result falls within an arbitrary range (0 to 255 in this embodiment), and can be adjusted according to the memory size for image processing.

DESCRIPTION OF SYMBOLS 1 Printing inspection apparatus 2 Color camera 3 Strobe illumination apparatus 4 Touchscreen monitor 5 Photoelectric sensor 6 Line control apparatus 8a, 8b, 8c Inspection object 21a Character 21b Background 21c Binary threshold 104 Filter adjustment process 403 Filter switching process 405 Composite brightness conversion processing

Claims (3)

An imaging means for imaging the inspection object using at least two light components having different wavelengths ;
Image data storage means for storing image data captured from the imaging means;
Data storage means for storing normal character data;
Inspection means for comparing the character data to be inspected, which is character information extracted from the image data storage means, with the registered normal character data to determine whether it is normal or defective;
A printing inspection apparatus comprising :
further,
Input means for receiving input from the user;
Filter adjustment means for determining a filter value and a binarization threshold for each product number input by the input means using the image data captured by the imaging means;
Character information extraction means for extracting character information from the image data captured by the imaging means using the filter value and binarization threshold value calculated by the filter adjustment means;
Having a control means for controlling the printing inspection apparatus;
The input means accepts an input of a product number from the user;
The data storage means associates the filter value and the binarization threshold value determined by the filter adjustment means using the image data picked up by the image pickup means with the product number received by the input means. Remember,
The imaging means images a reference inspection object serving as a reference on which characters to be the normal character data are printed,
The character information extraction unit extracts character information from the imaged image data of the reference inspection target using a filter value and a binarization threshold stored in the data storage device, and generates the normal character data And
The data storage means stores the generated normal character data in association with the product number received by the input means;
The character information extracting means receives an inspection object type number indicating an inspection object type number from the control means,
The imaging means images the inspection object corresponding to the inspection object type number,
The character information extraction unit extracts character information from the imaged image data of the inspection target using a filter value and a binarization threshold corresponding to the inspection target product number, and generates the inspection target character data A printing inspection apparatus characterized by: The print inspection apparatus according to claim 1 ,
Having designation means for designating a range of images stored in the image data storage means;
The filter adjusting means, the printing inspection apparatus characterized by having processing means for determining the filter value by statistically processing the two or more light components with different wavelengths within the range. The print inspection apparatus according to claim 2 ,
A print inspection apparatus, wherein only the character and the corresponding background portion are designated to determine the filter value .

Images