JP2002190019A - Method of preparing mask for color image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily obtain mask data without using three color vectors. SOLUTION: Only two first and second color bands CB1, CB2 are set as color bands (classification basic colors) for classifying the color of a mask object region and the color of a background region. The components of the first and second color bands CB1, CB2 are acquired with respect to each pixel of a color image, and mask data is originated according to the components a, b of the first and second color bands CB1, CB2 in each pixel.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for creating mask data used for processing a desired color area in a color image.

[0002]

2. Description of the Related Art In color image processing, mask data representing a color region is often used in order to selectively perform processing on a region of a desired color. As a technique for generating such mask data, there is a technique described in Japanese Patent No. 2968611 disclosed by the present applicant. In this technique, three color vectors are determined to create a mask, and mask data is generated by obtaining coefficients of the three color vectors for the color of each pixel.

[0003]

However, in the above-mentioned conventional technology, since three color vectors are used, there is a problem that the amount of data to be processed becomes enormous, and a large amount of memory and a long processing time are required. Was.

The present invention has been made to solve the above-mentioned conventional problems, and has as its object to provide a technique capable of obtaining mask data more easily without using three color vectors. And

[0005]

In order to achieve the above object, the present invention provides a mask for distinguishing a desired mask target area in a color image from a background area other than the mask target area. A method for creating data, comprising: (a) first and second two basic color separations for separating the color of the mask target area from the color of the background area;
Setting only the two distinction basic colors; (b) acquiring the components of the first and second distinction basic colors for each pixel of the color image; and (c) acquiring the first distinction color of each pixel. And a step of creating the mask data according to the component of the second classification basic color.

[0006] According to this configuration, mask data can be created by using the two distinct basic colors, so that mask data can be obtained more easily than in the past.

The step (a) includes: (a1) obtaining a spectral reflectance characteristic of the mask target area and the background area in the color image target; and (a2) obtaining the mask target area and the background. Determining the first and second distinction basic colors based on the spectral reflectance characteristics of the region;
May be included.

[0010] According to this configuration, since the two distinction basic colors are determined according to the spectral reflectance characteristics of the mask target area and the background area, it is possible to set an appropriate distinction basic color.

[0009] The step (a2) may include a step of observing a spectral reflectance of the mask target area and the background area by an operator to determine the first and second distinction basic colors. .

In this configuration, since the operator observes the spectral reflectance and determines the two basic colors for classification, it is possible to easily determine an appropriate basic color for classification.

The step (a2) includes a step of obtaining the spectral characteristics of an imaging optical system for imaging the object and obtaining the color image; and a step of using the imaging optical system to convert the object to a specific color condition. Estimating spectral luminance characteristics obtained in the mask target area and the background area, respectively, for a plurality of color conditions, assuming that imaging is performed in a plurality of color conditions, and the mask target area and the background area in the plurality of color conditions From the spectral luminance characteristics, a step of estimating the tone value of the mask target area and the background area in a plurality of specific color images obtained by imaging the object under the plurality of color conditions, respectively. Based on the tone values of the mask target area and the background area in the plurality of specific color images, the first and second colors are used among the colors used in the plurality of color conditions. A step of selecting a basic color, may include a.

In this configuration, two distinct basic colors are determined according to the estimated gradation values of the mask target area and the background area. Therefore, the two basic colors are determined from the image data of the color image actually obtained by the imaging optical system. It is possible to easily obtain an appropriate classification basic color for creating mask data.

[0013] The step (a) may include a step of an operator observing the color of the mask target area and the color of the background area and designating the first and second two distinct basic colors. Good.

In this configuration, since the basic color for sorting is determined based on the observation of the worker, it is possible to easily set an appropriate basic color for sorting.

Further, the step (a) further comprises the step of, when the designated first and second discrimination basic colors include color components overlapping each other, the first and second discrimination so as not to overlap each other. A step of adjusting the basic color may be included.

When the two basic colors for color separation include color components that overlap with each other, it is often difficult to discriminate between the mask target area and the background area. Therefore, in such a case, if the two sorting basic colors are adjusted so as not to include color components overlapping each other, a sorting basic color suitable for separating the mask target area and the background area is set. It is possible.

The step (a) comprises the steps of obtaining an RGB image as the color image, and a distribution of R, G, and B components for a plurality of pixels in the mask target area and a plurality of pixels in the background area. Determining the first and second distinction basic colors based on the first and second colors.

With this configuration, for example, when an RGB image exists in advance, it is possible to directly generate mask data from the RGB image without measuring the spectral reflectance of the object.

In the step (c), the color of each pixel in the color image is represented by a two-dimensional color vector composed of the first and second distinction basic color components. Mapping the two-dimensional color vector for a plurality of pixels to a point on a specific straight line in the two-dimensional color space;
From the histogram of the plurality of points mapped on the specific straight line, determining a threshold value for separating pixels in the mask extraction region and pixels in the background region on the specific straight line. And generating the mask data by dividing the plurality of points using the threshold value.

In this configuration, a plurality of points mapped on a specific straight line are distributed separately into pixel points in the mask target area and pixel points in the background area. Therefore, if a threshold value is determined from the histogram of these points and a plurality of points are classified based on the threshold value, it is possible to obtain mask data for discriminating a mask target region from a background region.

Alternatively, the step (c) includes a step of representing a color of each pixel in the color image by a two-dimensional color vector composed of the first and second distinction basic color components. Determining, from the two-dimensional color vector for a plurality of pixels in the image, a partition vector capable of partitioning a pixel in the mask extraction region and a pixel in the background region in a two-dimensional color space; Creating the mask data by processing the color image using the partition vector.

According to this configuration, it is also possible to obtain mask data for distinguishing a mask target area from a background area.

It should be noted that the above-described mask data creating method is as follows.
Further, before the step (b), a step of performing a blurring process on the color image may be provided.

If the blur processing is performed before the processing of the color image using the discrimination basic color, the number of specific pixels is reduced. Therefore, it is possible to suppress the mask data from including specific pixels. it can.

Further, the above-described mask data creating method is as follows.
Further, after the step (c), a step of performing a dust erasing process including a thinning process and a thickening process on the mask data may be provided.

By performing the dust erasing process, pinhole-shaped areas (dust) included in the mask data can be reduced.

The present invention can be realized in various modes. For example, a method and an apparatus for creating mask data, a method and an apparatus for creating a mask using mask data, and the functions of those methods and apparatuses , A recording medium on which the computer program is recorded, a data signal including the computer program and embodied in a carrier wave, and the like.

[0028]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, embodiments of the present invention will be described based on examples in the following order. A. First embodiment (color band setting by spectral reflectance measurement): Second embodiment (visual color band setting of target object): Third embodiment (color band setting by simulating imaging): Fourth embodiment (color band setting based on existing color image): Modification of mask data creation processing:

A. First Embodiment (Setting of Color Band by Spectral Reflectance Measurement): FIG. 1 is an explanatory diagram showing a configuration of a printed circuit board inspection apparatus as a first embodiment. This apparatus includes a stage 30 on which a printed circuit board PCB is mounted.
A driving device 40 for driving the stage 30, a light source 50 for illuminating the printed circuit board PCB, and a microscope unit 60 for obtaining an enlarged image of the printed circuit board PCB;
The apparatus includes a spectroscope 70 for measuring the spectral reflectance of the printed circuit board PCB, an imaging unit 80 for taking a color image of the printed circuit board PCB, and a computer 100 for controlling the entire apparatus. The spectroscope 70 and the imaging unit 80 are switched and selectively connected to the microscope unit 60.

The computer 100 includes a color band setting section 110, a preprocessing section 120, and a mask data creation section 1
30 and a post-processing unit 140. The functions of these units are realized by a computer program stored in a memory (not shown) of the computer 100.

FIG. 2 is an explanatory diagram showing a color image of a printed circuit board PCB on which a mask is to be made in this embodiment. The surface of the printed circuit board PCB includes a first green area A in which a resist is applied on a substrate base, a second green area B in which a resist is applied on copper wiring, and a gold area C in which gold is applied. , A brown region D of the substrate base. Since the substrate base, which is the base of the first green area A, is brown and the copper wiring, which is the base of the second green area B, is copper, the colors of these two areas A and B are slightly different. It is still green. In this embodiment, mask data indicating two green areas A and B to which a resist is applied is created. That is, the green area A,
B is a mask target area, and the other areas C and D are background areas.

FIG. 3 is a flow chart showing a procedure for producing a mask in the embodiment. In step S1, two color bands are set. Here, "color band"
Means a separation basic color used to separate the color of the mask target area from the color of the background area.

FIG. 4 is a flowchart showing details of the color band setting process. In step S11, the spectral reflectance of each area on the printed circuit board PCB is measured using the spectroscope 70 (FIG. 1). FIG. 5 is a graph showing the spectral reflectance characteristics of the four regions A, B, C, and D shown in FIG. In this example, the reflectance of the first green area A is the lowest, and then the second green area B, the brown area D, and the gold area C
The reflectance gradually increases in the order of. However, specularly reflected light is generated by the unevenness of the surface of the printed circuit board PCB.
When the light is incident on 0, a phenomenon occurs in which the apparent reflectance increases. The “green area A” shown in the graph of FIG.
The graph of “(specular reflection)” is an example of the distribution of the reflectance by such specular reflection light.

The spectral reflectance distribution as shown in FIG. 5 is displayed on the display unit of the computer 100. The operator observes these spectral reflectance distributions and sets the first and second color bands. That is, in step S12 (FIG. 4), the first color band CB1 is set. The first color band CB1 is in a wavelength range in which the spectral reflectances of the background regions C and D have a peak and the difference between the spectral reflectances of the background regions C and D and the mask target regions A and B is large. The color represented. In the example of FIG. 5, the first color band CB1 is 6
The wavelength is set in the range of 00 nm to 700 nm. In the first color band CB1 set as described above, the difference in contrast between the background areas C and D and the mask target areas A and B is large. Therefore, the first color band CB1
Is used, it is easy to distinguish the background areas C and D from the mask target areas A and B.

As can be understood from the example of FIG. 5, even in the first color band CB1, the reflectance of the brown area D in the background area is not much different from the reflectance of the mask target areas A and B. not big. Therefore, the first color band CB
With only 1, it may be difficult to distinguish the brown region D from the mask target regions A and B.

In step S13, the operator sets the second color band CB2. The second color band CB2 includes a background region D and a mask target region B having reflectances close to each other in the first color band CB1.
Are selected in a wavelength range that has a greater reflectance difference from each other. In the example of FIG. 5, the second color band CB
2 is set in a wavelength range of 450 nm to 550 nm. The second color band CB2 thus set
Is used, it is possible to more reliably separate the background region D and the mask target region B that are difficult to separate in the first color band CB1.

The two color bands CB1 and CB2 are set in the color band setting section 110 (FIG. 1). Specifically, for example, the color band setting unit 110 displays a user interface (dialog box) for setting the color band on the display unit of the computer 100, and the operator operates the two color bands CB on the user interface.
1 and CB2 are input. That is, the color band setting unit 110 provides the operator with the color bands CB1, C2.
It has a function of allowing the setting of B2 and receiving information of the color bands CB1 and CB2 set by the operator.

Thus, the two color bands CB1, CB
When the setting of 2 is completed, in step S2 of FIG.
The image of the printed circuit board PCB is captured using the image capturing section 80. At this time, the color condition at the time of imaging is determined according to the first and second color bands CB1 and CB2. Here, the “color condition at the time of imaging” means a wavelength range to be imaged. The color conditions at the time of imaging include the spectral characteristics of the camera in the imaging unit 80, the spectral characteristics of the color filters (when color filters are used), and the spectral characteristics of the light source 50.

FIGS. 6A and 6B are explanatory diagrams showing examples in which color conditions at the time of image pickup are set using color filters. Here, the imaging unit 80 is configured by providing a color filter in front of the monochrome CCD camera, and the color condition at the time of imaging is set by the spectral characteristics of the color filter. The first color filter F1 is the first color filter F1.
Is a filter that transmits light of the color band CB1 and blocks light of other wavelengths. Second color filter F
Reference numeral 2 denotes a filter that transmits light of the second color band CB2 and blocks light of other wavelengths. FIG. 6 (A),
(B) shows two examples of the color filter. However, the color filter has two color bands CB1, C
What is necessary is just to acquire the image of B2, and those having various other characteristics can be used.

When capturing an image, the printed circuit board PC is used by using these two color filters F1 and F2.
The two multi-gradation image data B are acquired. The two multi-tone image data thus obtained represent different color components for the same object (printed circuit board PCB).
It can be considered that one color image is constituted by these two color components.

As can be understood from this example, a “color image” in this specification means an image having the same object and having two or more different color components.

In step S3 (FIG. 3), the pre-processing unit 120 (FIG. 1) performs a blurring process on the image data of the two color components corresponding to the two color bands CB1 and CB2 obtained in this manner. . By performing this blurring process, peculiar pixels existing in the image data of each color component can be removed, so that mask data with less dust (noise component) can be obtained. Note that the pre-processing can be omitted.

In step S4, the mask data creation unit 1
30 creates mask data from the image data of the two color components obtained in step S3. FIG. 7 is an explanatory diagram showing a method of creating mask data in step S4. The horizontal axis of FIG. 7A is the component a for the first color band CB1, and the vertical axis is the second color band CB2.
Is the component b. The black point is the color component (a,
b) is a point having coordinate values.

The point (a, b) of each pixel from the coordinate origin
Is equivalent to a color vector in a two-channel color space. Therefore,
It can be considered that the color component (a, b) of each pixel is represented by such a two-dimensional color vector.

In the process of creating mask data, FIG.
, A point of each pixel on the two-channel color space (that is, a two-dimensional color vector) is defined
Project onto L. Specifically, an intersection between the two-dimensional color vector of each pixel and the projection line PL is a point after projection of each pixel.

Here, the projection straight line PL is preferably represented by the following equation (1).

(Equation 1)

At this time, the coordinates (a ′,
b ′) is expressed by the following equations (2a) and (2b), respectively.

(Equation 2)

The projected straight line PL of this embodiment is a straight line passing through the points (1, 0) and (0, 1) where the coordinate values on the two coordinate axes are 1, so that the coordinate values a 'and b' after the projection are , Coordinate values a and b before the projection. Therefore, in this specification,
The coordinate values a ′ and b ′ after projection are also called “chromaticity values”.

The two color bands CB1 and CB2
For a pixel in which both components a and b are zero, the results of the above equations (2a) and (2b) become indefinite.
Therefore, as shown in FIG. 7B, it is preferable that the point SP (0,0) of the peculiar pixel at the coordinate origin is projected onto a predetermined point SP ′ on the projection line PL. This point S
P ′ can be set at an arbitrary position on the straight line PL, but in the example of FIG. 7B, the point (0.5, 0.5)
Is set to

Next, as shown in FIG. 7C, a histogram HG of the chromaticity value points (a ', b') is created. In this example, the histogram HG is created by calculating the frequency of pixels for each point on the projection line PL. Instead, a histogram may be created on one of the two coordinate axes a and b. Specifically, only the histogram related to the first chromaticity value a ′ may be created, or the second chromaticity value b ′ may be created.
It is also possible to create only a histogram relating to.

As described above, the two color bands CB
1 and CB2 are set to colors suitable for separating the mask target area and the background area, and thus the histogram HG
, A peak P1 corresponding to the mask target area and a peak P2 corresponding to the background area appear. Further, a minimum value of the appearance frequency exists between these peaks P1 and P2. Therefore, a point Th indicating the minimum value of the appearance frequency is obtained, and the chromaticity value of each pixel is binarized using the chromaticity value a ′ (or b ′) of the point Th as a threshold value. Mask data representing the target area can be created.

After the mask data is created in this way, the post-processing unit 140 executes post-processing in step S5 in FIG. This post-processing is dust elimination processing including thinning processing (shrinking processing) and thickening processing (expansion processing). For example, first, for a pixel having a binary mask data value of 1 (a pixel in a mask target area), a predetermined pixel width narrowing process is performed, and then the same pixel width widening process is performed. As a result, a pinhole-shaped mask target region smaller than the pixel width is removed. When the thinning process and the fattening process are similarly performed on the pixels in the background region, the pinhole-shaped background region is removed. As a result,
Fine dust-like regions can be removed.

In the post-processing, the mask data to be finally used may be obtained by further enlarging or reducing the mask data. Note that post-processing may be performed as needed, and may be omitted.

As described above, in the first embodiment, two color bands CB1 and CB2 (basic classification colors) suitable for separating the mask target region from the background region are set, and the color of each pixel is set. It is obtained as components a and b of two color bands CB1 and CB2. Since the mask data is created based on the sizes of these components a and b, it is possible to easily create the mask data only with two color bands without setting three or more color bands. is there.

In setting the color band, as shown in FIG. 5, the first color band CB1 is divided into a mask target area (green areas A and B) and a background area (gold area C).
And a brown region) are set in such a wavelength range that the contrast with the second color band CB2 is increased.
Is set to a wavelength range in which the reflectance difference between the mask target area and the background area in the color band CB1 is small and the reflectance difference between the background area and the mask target area is large. It is possible to

B. Second embodiment (color band setting by visual observation of an object): FIG. 8 is a flowchart showing the procedure of color band setting (step S1 in FIG. 1) in the second embodiment. The second embodiment is the same as FIG. 3 of the first embodiment except for the procedure for setting the color band.

In step S21, the operator observes the colors of the mask target area and the background area on the object (printed circuit board PCB) with the naked eye and designates the first color band candidate CC1. As the first color band candidate CC1,
A color close to the background area is selected from the colors of the mask target area. In the example of FIG. 2, the mask target areas A and B are both green, and the background areas C and D are gold and brown. However,
Since the gold area C is yellow on the captured image,
Recognize the area as yellow. On the other hand, the second mask target area B is formed by applying a resist on the copper wiring, and has a slightly more yellow background area C than the first mask target area A. Therefore, the color (light green) of the second mask target area B close to the background area C is selected as the first color band candidate CC1.

In step S22, among the colors of the background area, a color close to the mask target area is selected as the second color band candidate CC2. In the example of FIG. 2, the yellow background area C is selected as the second color band candidate CC2. As described above, in the second embodiment, among the colors of the mask target area and the background area, colors close to each other are the first and second colors.
Are selected as color band candidates CC1 and CC2.
The reason is that such two color band candidates CC
This is because if the mask target area and the background area can be distinguished by using 1, CC2, pixels having colors other than these can also be distinguished well.

In step S23, it is determined whether the first and second color band candidates CC1 and CC2 have independent colors. FIG. 9A shows an example of two color band candidates CC1 and CC2. The first color band candidate CC1 is green (G) and has a wavelength of about 500 nm to about 6 nm.
It has an intensity peak in the wavelength range of 00 nm. The second color band candidate CC2 is yellow (Y), which is the sum of the green and blue wavelength ranges.
It has high intensity in the wavelength range of about 700 nm.

Such two color band candidates CC
Whether CC1 and CC2 are independent of each other is determined by whether or not the wavelength ranges having a predetermined intensity value (for example, 50%) or more overlap each other. In the example of FIG. 9A, the second color band candidate CC2 includes the wavelength range of the first color band candidate CC1. At this time, it is determined that the two color band candidates CC1 and CC2 are not independent, and step S25 in FIG. 8 is executed.

In step S25, of the two color band candidates CC1 and CC2, the one with a narrower wavelength band is adopted as the first color band CB1. In the example of FIG. 9A, since the wavelength range of the first color band candidate CC1 is narrower, this is adopted as the first color band CB1. In step S26, the wavelength range obtained by taking the exclusive OR (EXOR) of the two color band candidates CC1 and CC2 is adopted as the second color band CB2.
In other words, the two color band candidates CC1, CC2
However, when color components overlapping each other are included, final color bands CB1 and CB2 (that is, classification basic colors) are reset so as not to overlap each other.

As can be understood from the description of FIG. 9,
The worker observes the target object and selects two color band candidates CC
When designating 1, CC2, the spectral characteristics (spectrum) of each candidate are substantially designated. The specification of the spectral characteristics of the color band candidates CC1 and CC2 is performed using the color band setting unit 110. For example, a relation between a plurality of selectable colors and their spectral characteristics is stored in advance in a data storage unit in the computer 100, and when setting a color band, the color band setting unit 110 transmits a plurality of color patches to the computer 100 on the screen. Then, when the operator selects the colors of the two color band candidates CC1 and CC2 from the plurality of color patches, the spectral characteristics corresponding to the selected colors are read from the data storage unit. However, the wavelength ranges of the color band candidates CC1 and CC2 can be specified by various other methods.

Thus, the two color bands CB1, C
When B2 is set, steps S2 to S5 of the first embodiment are performed.
Mask data is created according to the same procedure as in FIG.

As described above, in the second embodiment, since the operator observes the object and sets the two color bands CB1 and CB2, two similar colors of the mask target area and the background area are set. To two color bands CB1, CB2
Can be set as Therefore, two color bands CB1, which can distinguish the mask target area and the background area,
CB2 can be easily set. Further, there is an advantage that it is not necessary to measure the spectral reflectance of an object (mask target area or background area) in the color band setting step.

Instead of directly observing the object, two color bands CB1 and CB2 may be set by observing the full-color image of the object.

C. Third Embodiment (Color Band Setting by Simulating Imaging): In the setting of the color band in the third embodiment, the color band is set by simulating the imaging of an object (printed circuit board PCB). Hereinafter, before describing the specific processing procedure of the third embodiment, the spectral characteristics related to imaging by the imaging unit 80 will be described.

FIG. 10 shows spectral characteristics related to imaging of an object. FIG. 10A shows a spectral distribution of the light source 50, and FIG. 10B shows an object (printed circuit board PCB).
Of the plurality of regions R1 to R3. When the light source 50 illuminates the three regions R1 to R3, the intensity (spectral radiance) of the reflected light from these regions R1 to R3 is the product of the spectral distribution of the light source 50 and the spectral reflectance of each region. Given. Therefore, the spectral radiance of each area is as shown in FIG. When such a reflected light is captured by a camera (imaging unit 80) having the spectral sensitivity shown in FIG. 10D, the spectral distribution of light input to the camera indicates the spectral radiance of each region (FIG. 10D).
(C)) is multiplied by the spectral sensitivity of the camera (FIG. 10 (d)). FIG. 10E shows the spectral characteristics of the light input to the camera. The brightness gradation value of each area in the image of the target object corresponds to a value obtained by integrating the spectral distribution shown in FIG. 10E (the area of the hatched portion).

The three regions R1 to R3 of the object shown in FIG. 10 are for convenience of explanation, and in the example of the printed circuit board PCB of FIG. 2, four regions A, B, C, and D are illustrated. 10 (b), (c) and (e) are obtained respectively.

Considering such an imaging process, FIG.
If the spectral characteristics shown in FIGS. 10A to 10D are known, the gradation values of the images of the respective regions acquired by the imaging unit 80 (FIG.
(E)) can be simulated (predicted).

The data required to simulate this imaging process is as follows. (1) Spectral reflectance of each region of the inspection object; (2) Spectral distribution of illumination light source 50 (in the case where light source 50 is configured by combining white light with a color filter, spectral distribution after transmission through the filter); 3) Spectral sensitivity characteristics of the imaging unit 80 (for example, a CCD camera) (when a camera filter is used, the spectral sensitivity characteristics of the combination of the camera and the filter)

Therefore, in the third embodiment, as described below, these spectral characteristics are acquired, and the gradation values of the image in the mask target area and the background area are predicted.

FIG. 11 is a flowchart showing a procedure for setting a color band (step S1 in FIG. 1) in the third embodiment. The third embodiment is the same as FIG. 3 of the first embodiment except for the procedure for setting the color band.

In step S31, the spectral reflectance (FIG. 10B) of each area of the object is measured. This step is the same as step S4 in FIG. In step S32, the spectral characteristics (FIGS. 10A and 10D) of the light source 50 and the imaging unit 80 (FIG. 1) are obtained. In this embodiment, as the light source 50, three L colors of red (R), green (G), and blue (B) each having the spectral characteristics shown in FIG.
Use ED. The spectral characteristics of the light source 50 and the imaging unit 80 are:
For example, they are provided by manufacturers who market them.
Alternatively, it is also possible to measure each spectral characteristic by performing a spectral measurement. When a monochrome CCD camera is used as the imaging unit 80, the spectral sensitivity characteristic of the camera may be assumed to be constant in visible light for easy simulation.

In step S33, the spectral radiance (FIG. 10C) of each area of the object is calculated. Specifically, the spectral reflectance of each region of the object (FIG. 10B)
By multiplying the spectral distributions of the three color LEDs (FIG. 12),
The spectral radiance of each region of the object (FIG. 10C) is calculated. FIG. 13 shows that LEDs of three colors are individually used as the light source 5.
It shows the spectral radiance of each region of the object when used as 0.

In step S34, the image gradation value of each area of the object is estimated. This image gradation value is obtained by multiplying the spectral radiance of each area (FIGS. 10C and 13) by the spectral sensitivity of the camera (FIG. 10D) (FIG. 1).
0 (e)) is obtained by integration. However,
In this embodiment, as a simple method, it is assumed that the spectral sensitivity of the camera is constant (100%) in the visible light range. At this time, the image gradation value is given by the integral value of the spectral radiance (FIGS. 10C and 13). Here,
It is assumed that the relative relationship between image tone values does not shift due to A / D conversion or the like in the imaging unit 80.

FIG. 14 shows an example of the image gradation value of each area obtained in step S34. Each area is a three-color LE
It is presumed that such image gradation values are obtained by taking images with the D light source, respectively. In the example of FIG. 14,
It is assumed that the dynamic range of the image signal obtained by the imaging unit 80 is infinite. In reality, a tone value having a large value is limited to the upper limit of the dynamic range.

In step S35 of FIG. 11, the estimated value of the image gradation value of each area is plotted on a two-channel color space (two-dimensional color space). FIGS. 15 to 17 are graphs in which the estimated values of the image gradation values of each area are plotted on three sets of two-channel color spaces. For example, in FIG. 15, the horizontal axis represents the image gradation value of the red LED, and the vertical axis represents the gradation value of the green LED. Similarly, FIG. 16 shows the image gradation value of the red LED on the horizontal axis and the blue LED on the vertical axis, and FIG. 17 shows the image gradation value of the green LED on the horizontal axis and the blue LED on the vertical axis.

In this embodiment, since the image gradation value of the representative pixel of each area is estimated, only one point indicating each area is plotted in each graph of FIGS. Have been. However, when an object is actually imaged, an image gradation value is obtained for each of a plurality of pixels in each region. If the image gradation values of the plurality of pixels obtained in this way are plotted in a two-channel color space as shown in FIGS. 15 to 17, the graph shown in FIG. 7A should be obtained.

When creating mask data from the distribution of points in a two-channel color space as shown in FIG. 7A, the points of the pixels in the mask target area and the points of the pixels in the background area are separated as much as possible. Is preferred. For this purpose, when the gradation prediction values (simulation values) are plotted in a two-channel color space as shown in FIGS. 15 to 17, the color vectors Va and V from the origin to the points of the mask target areas A and B are obtained.
It is sufficient to select a two-channel color space in which the angle between b and the color vectors Vc and Vd reaching the background areas C and D is as large as possible. For example, in FIG. 15, the color vector Va ′ of the green area A (specular reflection) and the color vector V of the gold area C are shown.
The angle with c is quite narrow. On the other hand, in FIG. 16, the angle between these two vectors Va ′ and Vc is relatively large. Such a difference is considered to be due to a high reflectance of the green LED C in the wavelength region of the green LED. Also,
In FIG. 17, the color vector Va of the mask target area A substantially overlaps with the color vector Vd of the brown area D. Therefore, in FIGS. 15 to 17, in FIG. 16, the angle difference between the color vectors Va and Vb of the mask target area and the color vectors Vc and Vd of the background area is the largest.

In step S36 of FIG. 11, the first and second color bands CB1 and CB1 are set so that the angle difference between the color vector of the background area and the area to be masked is maximized from the graphs of FIGS. CB2 is determined. In FIGS. 15 to 17, in the case of FIG. 16, the angle difference between the color vectors of the background region and the mask target region is maximized. Therefore, as the first and second color bands, the colors of the red LED and the blue LED constituting the two-channel color space of FIG. 16 are respectively selected. This selection may be performed by data processing by the computer 100, or may be performed by an operator by observing graphs as shown in FIGS.

Thus, the two color bands CB1, C
When B2 is set, steps S2 to S5 of the first embodiment are performed.
Mask data is created according to the procedure shown in FIG.

As described above, in the third embodiment, based on the spectral characteristics of the mask target region and the background region, the spectral characteristics of the light source 50, and the spectral characteristics of the imaging unit 80, the image of each region at the time of imaging is obtained. A tone value is estimated, and two color bands CB1 and CB1 suitable for creating mask data are estimated based on the estimated value.
CB2 (sorting basic color) is selected. Therefore, the two color bands CB1 and CB2 suitable for creating mask data are taken into consideration while taking into account the imaging conditions of the mask target area and the background area.
Can be set.

D. Fourth Embodiment (Color Band Setting Based on Existing Color Image): In the fourth embodiment, when a color image of an object for which a mask is to be created exists, two color bands are determined from the existing color image. The fourth
The color image in the embodiment means an image having three color components, such as an RGB color image.

FIG. 18 is a flowchart showing a procedure for setting a color band in the fourth embodiment. Step S
At 41, existing color image data of the object is acquired. In step S42, the three color components of the color image data are plotted in three sets of two-channel color spaces.

FIGS. 19 to 21 are graphs in which the distribution of the color components of each pixel of the RGB color image data is plotted on three two-channel color spaces. These figures respectively correspond to FIGS. 15 to 17 in the third embodiment described above. Also in step S43 of the fourth embodiment, as in the third embodiment, the angle between the color vector of each pixel in the mask target areas A and B and the color vector of each pixel in the background areas C and D is maximized. A channel color space is selected, and two basic colors constituting the color space are designated by a color band CB.
1, CB2.

In the example of FIG. 19, there is a difference between the color vectors of the gold area C and the color vectors of the green areas A and B. Therefore, the combination of red (R) and green (G) that make up the color space of FIG.
It is considered preferable as B2. On the other hand, in the example of FIG. 20, the color vector of the high gradation portion (the portion where the B component is large) of the gold region C is quite close to the color vectors of the green regions A and B. Therefore, the combination of red (R) and blue (B) constituting the color space of FIG. 20 is not suitable as the two color bands CB1 and CB2. Further, in FIG. 21, a brown area D, a gold area C, a green area A,
Green (G) and blue (B) because C overlaps
Are not suitable as the color bands CB1 and CB2. As a result, in this object, two color bands CB1 and CB2 suitable for mask creation can be determined to be red and green. Note that this determination may be made by the operator observing FIGS. 19 to 20 or automatically by the color band setting unit 110 analyzing the RGB image.

Thus, the two color bands CB1, C
When B2 is set, steps S3 to S5 of the first embodiment are performed.
Mask data is created according to the procedure shown in FIG.
In the fourth embodiment, since the color image of the object has already been acquired, the image is not read in step S2 in FIG.

As described above, in the fourth embodiment, two color bands C suitable for creating mask data are created based on the existing color images for the mask target area and the background area.
B1 and CB2 (basic classification colors) are selected. Therefore,
Even when the target object cannot be used directly, such as when the target object is not at hand, it is possible to set two color bands CB1 and CB2 suitable for creating mask data. Further, since it is not necessary to measure the spectral characteristics of the object, it is possible to set the color band more easily than in the first and third embodiments.

E. Modified example of mask data creation processing: FIG.
As the mask data creation processing in step S4, the following method can be adopted.

FIG. 22 is an explanatory diagram showing a process of creating mask data using a distinguished color vector. FIG. 22 includes the same points as the graph shown in FIG. Also,
FIG. 22 shows a representative color vector V1 representing a mask target area and a representative color vector V2 representing a background area. It is assumed that these representative color vectors V1 and V2 are expressed by the following equations (3a) and (3b), respectively.

(Equation 3)

These two representative color vectors V1, V2
Is automatically determined by the mask data creation unit 130 analyzing a set of points in the two-channel color space. Alternatively, the operator may observe and set the distribution of points as shown in FIG.

FIG. 23 shows how an arbitrary color E on the two-channel color space is represented by two representative color vectors V1 and V2. It is assumed that the color E is represented by the following equation (4).

(Equation 4)

[0093] In this case, any color _E (a _e, b _e) is
As represented by the following equations (5a) and (5b), it can be represented by a linear combination of two representative color vectors V1 and V2.

(Equation 5)

Here, k1 and k2 are coefficients of the respective representative color vectors V1 and V2. The coefficient k1 indicates the weight of the first representative color vector V1, and the color closer to the first representative color vector V1 has a larger value of the coefficient k1. Conversely, for a color close to the second representative color vector V2, the value of the coefficient k1 becomes smaller.

Thus, coefficients k1 and k2 are calculated for each pixel in the color image, and coefficient data composed of the value of the first coefficient k1 of each pixel is created. The coefficient data is multi-tone data (hereinafter, referred to as a “shade mask” or “multi-tone mask”) reflecting the degree of color (direction) of the representative color vector V1 in the mask target area. By binarizing this shade mask, binary mask data can be obtained.

The above equations (5a) and (5b) are expressed as the following equations (6a) to (6c) using a matrix k of coefficients and a matrix M of representative color vectors V1 and V2. it can.

(Equation 6)

Therefore, the coefficient matrix k is given by the following (7)
As shown in the equation, the ^value can be calculated by multiplying an arbitrary color E vector by an inverse matrix M ⁻¹ of the representative color vector.

(Equation 7)

FIG. 22 shows the relationship between the threshold value u for binarizing the gray scale mask and the mask target area and the background area. Here, a straight line TL representing the threshold value u
Are parallel to the second representative color vector V2 because an arbitrary color is represented by two representative color vectors V1 and V2, and the threshold value u is different from that of the first representative color vector V1. This is because it corresponds to the value of the coefficient k1. That is, when a specific value u of the coefficient k1 of the first representative color vector V1 is used as a threshold value and the density mask represented by the coefficient k1 is binarized, each pixel is divided by a straight line TL shown in FIG. Is done. As a result, as can be easily understood from FIG. 22, the mask target region and the background region are clearly separated.

As described above, in this modification, the representative color vector V1 representing the mask target area and the representative color vector V2 representing the background area are determined, and the color of each pixel is linearly combined with these vectors. The mask data is created based on the coefficient k1 of the first representative color vector V1 when represented by. Therefore, as shown in FIG. 22, even if the mask target area has a highlight portion separated from the representative color vector, the mask target area and the background area can be distinguished well.

Various other methods can be used for the mask data creation processing. For example, the following method disclosed by the present applicant can be adopted.

(1) Paragraph 005 of Japanese Patent No. 2,896,311
3 to 0057, a method for determining a threshold value for binarization.
This method includes analyzing the histogram of the chromaticity values (a ′, b ′) (FIG. 7) and analyzing the representative color vectors V1 and V2.
Can be applied to both of the histogram analysis processes of the coefficient.

(2) A method of selecting n target colors and m non-target colors and creating a mask disclosed in Japanese Patent No. 2896310. This method has been described for the case where the coefficients of the representative color vectors V1 and V2 are obtained. When this method is applied to the case of chromaticity values (a ′, b ′), FIG.
As shown in FIG. 4, it can be used to narrow down the analysis range of the histogram. That is, in each of the n target color vectors and the m non-target color vectors,
A combination that minimizes the angle between the vectors is selected, and a histogram is created within the range of chromaticity between the two vectors. Then, the chromaticity value that minimizes the histogram within the range is adopted as the threshold value.

(3) A method of improving mask accuracy by repeatedly producing a mask disclosed in Japanese Patent No. 2896317. This method uses the above-described representative color vector V
1, V2 and a method using chromaticity values (a ′, b ′).

(4) A method disclosed in Japanese Patent No. 2896319 in which a representative color of an image is determined, and a target color and a non-target color are designated from the representative colors. This method also uses the representative color vector V
The case where the coefficients of 1, V2 are obtained is described.
When this method is applied to a method using chromaticity values (a ′, b ′), as shown in FIG. 25, the chromaticity sandwiched between the selected target color and the non-target color vector A histogram is created and analyzed within the range.

(5) A method disclosed in Japanese Patent No. 2896320, in which an arbitrary color is displayed and a target color and a non-target color are designated from among the colors. This method also uses the representative color vectors V1, V
The case of obtaining the coefficient of 2 has been described. Even when this method is applied to a method using chromaticity values (a ′, b ′), as shown in FIG. 25, a color sandwiched between color vectors for a selected target color and a non-target color is used. Create and analyze histograms within degrees.

The present invention is not limited to the above examples and embodiments, but can be implemented in various modes without departing from the gist of the invention.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing a configuration of a printed circuit board inspection apparatus as a first embodiment.

FIG. 2 is an explanatory diagram showing a color image of a printed circuit board PCB on which a mask is to be created in the embodiment.

FIG. 3 is a flowchart illustrating a procedure for creating a mask in the embodiment.

FIG. 4 is a flowchart illustrating details of a color band setting process.

FIG. 5 is a graph showing spectral reflectance characteristics of four regions A, B, C, and D in FIG. 2;

FIG. 6 is an explanatory diagram illustrating an example in which color conditions at the time of imaging are set using a color filter.

FIG. 7 is an explanatory diagram showing a method of creating mask data in step S4.

FIG. 8 is a flowchart illustrating a procedure for setting a color band in the second embodiment.

FIG. 9 shows two color band candidates C in the second embodiment.
FIG. 4 is an explanatory diagram showing a relationship between C1 and CC2 and color bands CB1 and CB2.

FIG. 10 is an explanatory diagram showing spectral characteristics related to reading of an image of an object.

FIG. 11 is a flowchart showing a procedure for setting a color band in the third embodiment.

FIG. 12 is a graph showing spectral characteristics of LEDs of three colors.

FIG. 13 is a graph showing the spectral radiance of each region of an object when three color light sources are used alone.

FIG. 14 is an explanatory diagram showing image gradation values in each region obtained in step S34 of FIG. 11;

FIG. 15 is a graph in which image tone values of respective areas are plotted on a two-channel color space of red and green.

FIG. 16 is a graph in which image tone values of respective areas are plotted on a two-channel color space of red and blue.

FIG. 17 is a graph in which image tone values of respective areas are plotted on a two-channel color space of green and blue.

FIG. 18 is a flowchart showing a procedure for setting a color band in the fourth embodiment.

FIG. 19 is a graph in which color components of each pixel of an RGB color image are plotted on a two-channel color space.

FIG. 20 is a graph in which the color components of each pixel of an RGB color image are plotted on a two-channel color space.

FIG. 21 is a graph in which color components of each pixel of an RGB color image are plotted on a two-channel color space.

FIG. 22 is an explanatory diagram showing a process of creating mask data using a distinguished color vector.

FIG. 23 is an explanatory diagram showing a process of creating mask data using a distinguished color vector.

FIG. 24 is an explanatory view showing another method used for the mask data creation processing.

FIG. 25 is an explanatory diagram showing another method used for the mask data creation processing.

[Explanation of symbols]

 Reference Signs List 30 stage 40 driving device 50 light source 60 microscope unit 70 spectroscope 80 imaging unit 100 computer 110 color band setting unit 120 pre-processing unit 130 mask data creation unit 140 post-processing unit

Continued on the front page (72) Inventor Hiroshi Sano 4-chome Tenjin Kitamachi 1-chome, Honkawadori-Teranouchi, Kamigyo-ku, Kyoto Dainippon Screen Mfg. Co., Ltd. (72) Inventor Junichi Shiomi, Horikawa-dori, Kamigyo-ku, Kyoto-shi 4-chome Tenjin, Teranouchi 1-1-1 Kitamachi 1 Dainippon Screen Mfg. Co., Ltd. F-term (reference) 2G020 AA08 DA02 DA03 DA12 DA22 DA31 DA34 DA52 5B057 AA03 BA02 CA01 CA08 CA12 CA16 DA08 DB02 DB06 DB09 DC25 5L096 AA02 EA35 GA10

Claims (11)

[Claims] 1. A method for creating mask data for distinguishing a desired mask target area in a color image from a background area other than the mask target area, the method comprising: (a) selecting a color of the mask target area and the background Setting only first and second two distinction basic colors as distinction basic colors for distinguishing the color of the region; and (b) the first and second distinction colors for each pixel of the color image. Acquiring a component of the classification basic color; and (c) generating the mask data according to the first and second classification basic color components of each pixel. A method for creating mask data for color images. 2. The mask data creating method according to claim 1, wherein in the step (a), (a1) obtaining a spectral reflectance characteristic regarding the mask target area and the background area in the target of the color image. And (a2) determining the first and second distinction basic colors based on the spectral reflectance characteristics of the mask target area and the background area. 3. The mask data creating method according to claim 2, wherein the step (a2) comprises: observing a spectral reflectance of the mask target area and the background area by an operator; A mask data creation method, comprising the step of determining a basic color for sorting. 4. The mask data creating method according to claim 2, wherein the step (a2) comprises: obtaining a spectral characteristic of an imaging optical system for obtaining the color image by imaging the object; Assuming that the object is imaged under a specific color condition using the imaging optical system, a step of estimating spectral luminance characteristics respectively obtained in the mask target region and the background region for a plurality of color conditions. The mask target area in a plurality of specific color images obtained by imaging the target under the plurality of color conditions from the spectral luminance characteristics of the mask target area and the background area under the plurality of color conditions, respectively. And estimating the tone values of the background area, respectively, based on the tone values of the mask target area and the background area in the plurality of specific color images, Selecting the first and second distinction basic colors from the colors used in the color condition. 5. The method according to claim 1, wherein in the step (a), an operator observes a color of the mask target area and a color of the background area by an operator. 2. A method of creating mask data, comprising the step of designating two distinct basic colors. 6. The mask data creating method according to claim 5, wherein the step (a) further comprises the step of: when the designated first and second classified basic colors include color components overlapping each other. A method of creating mask data, comprising the step of adjusting the first and second distinction basic colors so as not to overlap each other. 7. The method according to claim 1, wherein the step (a) comprises: obtaining an RGB image as the color image; and a plurality of pixels in the mask target area and in the background area. Determining the first and second distinction basic colors based on the distribution of the R component, the G component, and the B component for the plurality of pixels,
And a mask data creation method. 8. The mask data creating method according to claim 1, wherein said step (c) comprises: separating the color of each pixel in said color image into said first and second colors. Representing a two-dimensional color vector composed of basic color components; and mapping the two-dimensional color vector for a plurality of pixels in the color image to a point on a specific straight line in the two-dimensional color space. And determining, from the histograms of the plurality of points mapped on the specific straight line, a threshold value for classifying pixels in the mask extraction region and pixels in the background region on the specific straight line. And a step of creating the mask data by dividing the plurality of points using the threshold value. 9. The mask data creating method according to claim 1, wherein said step (c) comprises: separating the color of each pixel in said color image from said first and second colors. A step of representing a two-dimensional color vector composed of components of a basic color; and a two-dimensional color vector relating to a plurality of pixels in the color image, a pixel in the mask extraction area and the background area in a two-dimensional color space. Determining a partition vector capable of partitioning a pixel within the mask, and generating the mask data by processing the color image using the partition vector. . 10. The mask data creating method according to claim 1, further comprising a step of performing a blurring process on the color image before the step (b). Data creation method. 11. The mask data creating method according to claim 1, further comprising: a dusting process and a fattening process for the mask data after the step (c). A method for creating mask data, comprising a step of performing an erasing process.