JP2006038549A - Grade determining method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a grade determining method for determining the grade of an inspection target such as a powder or a film on the basis of the number or ratio of feature points featured by a plurality of the feature quantities of the feature points present on the inspection target. <P>SOLUTION: Two kinds of feature quantities x and y are measured in relation to a plurality of the feature points present on the inspection target and a plurality of grade regions are set on a plane having the feature quantities x and y as axes. The feature points are plotted on the plane and, on the basis of the number of the feature points contained in the respective grade regions or the ratio of the feature points in the respective grade regions with respect to all of the feature points, the grade of the inspection target is judged. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a method for determining the grade of an inspection object. More specifically, the present invention relates to a method for determining the grade of an inspection object such as powder or film based on the number or ratio of feature points characterized by a plurality of feature amounts of feature points existing in the inspection object.

As a method for determining the grade of an inspection object, the feature amount of the object is calculated from a two-dimensional image pattern obtained by imaging the inspection object with a camera, and the feature amount is compared with a reference value to determine the grade of the inspection object. The method of doing is known. For example, Patent Document 1 discloses, as a specific example, determining an individual insulator grade from parameters and colors related to the insulator shape.
However, this method determines an individual inspection object as one feature point, and has, for example, a plurality of feature points such as a colored object in a white powder or a foreign object, a discolored portion in a film, and a scratch. The inspection object is not graded based on the number or ratio of feature points characterized by a plurality of feature quantities.
JP-A-6-347239

  An object of the present invention is to provide a method for determining the grade of an inspection object such as powder or film based on the number or ratio of feature points characterized by a plurality of feature quantities of feature points existing in the inspection object. There is to do.

  In order to solve such a problem, the present inventors have intensively studied a method for determining based on a plurality of feature points existing in an inspection object. As a result, (1) two types of feature amounts x for a plurality of feature points, respectively. , Y are measured, a plurality of grade regions are set on a plane having x and y as axes, and the feature points are plotted on the plane, or (2) three types of feature amounts x, Measure y, z, set a plurality of grade areas in the space with x, y, z as axes, plot the feature points in the space, and the number of feature points included in each grade area or all features By determining the grade based on the ratio of the feature points in each grade area to the points, it was found that the grade of the inspection object can be easily determined with high accuracy, and the present invention has been completed.

  That is, the present invention measures two types of feature amounts x and y for a plurality of feature points existing in the inspection object, sets a plurality of grade regions on a plane having x and y as axes, and determines the feature points. A grade determination method characterized by plotting on the plane and determining the grade of the inspection object based on the number of feature points included in each grade area or the ratio of the feature points in each grade area to all the feature points It is.

  Further, the present invention measures three types of feature amounts x, y, and z with respect to a plurality of feature points existing in the inspection object, and sets a plurality of grade regions in a space with x, y, z as axes, The feature points are plotted in the space, and the grade of the inspection object is determined based on the number of feature points included in each grade area or the ratio of the feature points in each grade area to the total feature points. It is a grade judgment method to do.

  According to the present invention, the grade of an inspection object such as a powder or a film can be easily determined with high accuracy.

Specific examples of the inspection object in the present invention include powders such as methionine, sheet-like or film-like products such as acrylic plates and optical films, but are not limited thereto.
White powder or the like is colored when dried, and may contain colored parts of yellow to brown to dark brown to black depending on the degree. Moreover, a film etc. may also have a foreign material and the part discolored partially. These colored portions, foreign matters, discolored portions, and the like are extracted as feature points. These feature points are usually defective areas of the inspection object.
As the two types of feature quantities x and y of the feature points, normally, intensity information and shape information are used, and the intensity information is the difference between the average brightness of the normal background of the inspection object and the brightness of the defect area or the average of the normal background. As the shape information, the difference between the chromaticity, the defect area, and the chromaticity represents an amount representing the size of the defect area.
Hereinafter, the present invention will be described in detail by taking powder as an example of the inspection object.

FIG. 1 is a diagram showing an example of an apparatus used in the present invention. The apparatus comprises a powder image measuring device 1 and an image processing device 7. The powder image measuring device has two fluorescent lamps 3 for illuminating the powder on the upper portion thereof, a line sensor 2 for photographing an image of the powder, and A uniaxial stage 6 for moving powder at a constant speed is provided at the lower part thereof, and these are isolated by a transparent glass plate 5 as a dust prevention measure.
A transparent glass plate having a thickness of about 10 mm is used, and two straight tube fluorescent lamps are arranged on the top thereof. By illuminating from two directions, imprints due to irregularities on the powder surface can be eliminated. The number of fluorescent lamps is not limited to two, and many fluorescent lamps may be arranged.

The powder is put in a tray, the surface is scraped off, flattened, loaded on a single axis stage, and moved at a constant speed.
A two-dimensional image is obtained by taking a one-dimensional image of the powder surface with a line sensor and moving the powder at a constant speed. The imaging area is not particularly limited, and is, for example, a visual field of 80 mm × a moving distance of 200 mm (160 cm ² ).
The video signal from the line sensor is processed by the image processing device 7 to determine the powder grade.

FIG. 2 is a diagram showing another example of an apparatus used in the present invention. Instead of the line sensor, two area sensors 2 and a plurality of fluorescent lamps 3 are arranged. A diffusion film 4 is disposed between the fluorescent lamp 3 and the transparent glass plate 5 to uniformly illuminate the entire field of view. Others are the same as in FIG.
In the area sensor, the powder surface is photographed in a stopped state, and sequentially photographed while moving one screen. The imaging area is not particularly limited. For example, the field of view is 30 mm × 28 mm × two cameras, and the uniaxial stage length is 240 mm (144 cm ² ).

As a sensor, a CCD camera is usually used, and a digital signal obtained by photographing is processed by an image processing apparatus.
A characteristic point in the case of this powder is usually a powder or a foreign substance having a color or shade different from that of a normal powder which is a defect region different from that of the normal powder.
Two types of feature amounts x and y are measured for the plurality of feature points. As the feature amount, the intensity difference between the average brightness of the normal area and the defect area, or the difference between the average chromaticity of the normal area and the defect area, and the shape information represents the size of the defect area. Is mentioned.
Specifically, the colored part in the white powder is measured. A colored part shows yellow-brown-dark brown-black according to a grade, and the grade changes with areas. Usually, the brightness and size of the colored part are measured.

FIG. 3 is a diagram illustrating an example of image processing. (A) shows a cross-sectional waveform of an original image obtained by photographing. The horizontal axis indicates the position, usually the position perpendicular to the moving direction of the powder, and the vertical axis indicates the brightness in 256 gradations.
Shading correction is performed in order to remove the non-uniformity of brightness existing in the normal part due to non-uniform illumination intensity, etc., and to keep the brightness of the normal area constant. For example, 128 gradations are added to the difference between the original image and the shading image (B) obtained as an average of normal regions, for example, to obtain a waveform (C) based on the 128 gradations.

In order to distinguish between the normal region and the colored portion, a threshold is set for the image after shading correction. Usually, a standard deviation σ of lightness is obtained for an image after shading correction, and 128−kσ is set as a threshold (E). Below this threshold is determined as a colored portion, that is, a feature point. Here, k is a coefficient that is appropriately determined in order to adjust the sensitivity at the time of color portion extraction, and a value of about 1 to 3 is usually used. As the value of k is smaller, even a slight brightness difference is extracted as a colored portion, but noise components are also extracted more. Conversely, the larger the value of k, the more difficult it is to extract the noise component, but the extraction sensitivity of the colored portion also decreases. The value of k should be determined experimentally while evaluating the actual product.
At that time, noise components other than the colored part may be extracted, but these usually have a smaller area than the colored part, so that the noise is eliminated by using a process that removes a minute region. Can do.

Next, labeling is performed for each colored portion. For each labeled defect area, the average brightness is obtained here as intensity information, and the area (number of pixels or a value obtained by converting it into mm ² ) is obtained here as dimension information. The area here is not the area of the shaded portion in (E) but the area of the plane of the image. Here, since black is extracted on a white background, using the average brightness itself means that the larger the value (closer to 128 gradations), the more normal. If the difference between the average brightness and 128 gradations is used as the intensity information, the closer the color is to black, the larger the numerical value, which makes it easier to understand intuitively. Therefore, it is preferable to use the difference from the 128 gradations as the defect density.

Grade determination is performed using the feature quantities measured in this way, that is, the brightness that is the intensity information of the colored portion that is the defect area and the area that is the shape information.
A plurality of grade regions are set on a feature plane having lightness and area as x and y axes. The setting method is performed by a plurality of circles or ellipses centered on the intersection of the x axis and the y axis, but is not limited thereto, and may be a rectangle or a triangle whose sides are the x axis and the y axis. FIG. 4 shows the result of the example. In this example, the grade region is set by multiple ellipses. The intervals between the ellipses need not be equal, and the curvature of the ellipses need not be predetermined. The setting method is selected based on the tendency of the feature that is known in advance for the feature point.

The measured feature points are plotted on this feature plane. Even when this operation is performed inside the computer and it is determined whether or not it is within the area by the computer, although it is not specifically plotted on a plane, it is equivalent to substantially plotting.
The number of feature points that may exist in each grade area is set in advance, and whether or not each grade area is satisfied is determined based on whether or not this threshold value is exceeded. A plurality of threshold values can be set, and it can be determined as ΔΔ ×.
Further, based on the ratio of the number of feature points existing in each grade area to the total number of feature points, it can be determined whether or not each grade area is satisfied.
In the case of these judgment methods, the lowest grade value that satisfies the grade condition is usually set as the grade of the inspection object. That is, if the condition of the grade B is satisfied but the condition of the grade C lower than the grade B is not satisfied and the condition of the lower grade D is satisfied, the grade of the inspection object is D. Note that the criterion is not limited to this.

Instead of the two types of feature values (x, y), three types of feature values (x, y, z) are measured for a plurality of feature points existing in the inspection object, and x, y, z are used as axes. Set multiple grade areas in the space to be plotted, plot the feature points in the space, and inspect based on the number of feature points included in each grade area or the ratio of feature points in each grade area to all feature points The grade of the object can also be determined.
For example, the inspection object is an object having a color, the feature point is an area having a different color or brightness, and the brightness difference, the color difference, and the area are measured as feature amounts.
In this case, except that the grade area is set in a space having x, y, and z as axes, the process is performed similarly to the case where the grade area is set in a plane having x, y as axes.

  EXAMPLES Hereinafter, although an Example is shown and this invention is shown concretely, this invention is not restrict | limited to the following Example.

Using the apparatus shown in FIG. 1, the grade of methionine powder was determined.
A transparent glass plate (5) having a thickness of 10 mm is arranged inside a powder image measuring device (1) having a width of 450 mm, a length of 500 mm, and a height of 500 mm, and two fluorescent lamps (3) and a powder are disposed above the transparent glass plate (5). The line sensor (2) for taking the image of is arranged. The powder was put in a tray, the surface was scraped off, flattened, loaded on a single axis stage, and moved at 30 mm / sec.
The imaging area was 160 cm ² with a visual field of 80 mm × movement distance of 200 mm. A single image of 2000 pixels × 5000 lines was obtained with a resolution of 40 μm / pixel.

This was processed by the image processing apparatus 7. As shown in FIG. 3, the shading correction is performed, a threshold value is set, a colored portion (defective region) is determined, labeling is performed for each defective region, and the brightness and the area (number of pixels) for each labeled region. Asked.
The equipment specifications used are shown below.
Fluorescent lamp: Twin fluorescent lamp Twin 27 watts 3 wavelength daylight FPL27EX-D National line sensor: CCD line black and white camera NS2048 NED
Lens: Micro-NIKKOR Nikon Corporation Image processing device (7): FHC330A (image input board), FVL (image processing library) First Corporation

A grade region is set as an ellipse on a plane in which the x-axis is the defect density (represented by a difference from 128 gradations of brightness) and the y-axis is the defect area (represented by the number of pixels). Substituting (50, 100), (70, 200), (90, 300), (110, 400) into x ₀ and y ₀ in the following formula (1), respectively, draws four ellipses, and 5 Two grade areas were set.

Measurement was performed on four samples (No. 1 to No. 4), and the results were plotted in the set grade region. The results are shown in Table 1 and FIG.
The number of defects allowed in each grade area was set, and it was determined whether or not this reference number was satisfied.
Based on the result, the grade of each sample was determined. The lowest grade value that satisfies the grade condition was taken as the grade of the inspection object.

It is a figure which shows the example of the apparatus used by this invention. It is a figure which shows the other example of the apparatus used by this invention. It is a figure explaining the example of image processing. It is the figure which plotted the feature point in the grade area | region in an Example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Powder image measuring apparatus 2 Line sensor 3 Fluorescent lamp 4 Diffusion film 5 Transparent glass plate 6 Single axis stage 7 Image processing apparatus

Claims (8)

  Two types of feature quantities x and y are measured for a plurality of feature points existing on the inspection object, a plurality of grade areas are set on a plane with x and y as axes, and the feature points are plotted on the plane. A grade determination method characterized by determining the grade of an inspection object based on the number of feature points included in each grade area or the ratio of the feature points in each grade area to all feature points.   An image is acquired by photographing an inspection object, feature points existing in the inspection object are extracted by image processing, and two types of feature amounts are measured for each extracted feature point. Item 1 grade evaluation method.   3. The method according to claim 1, wherein the feature point is a defect area of the inspection object, and the two kinds of feature amounts are intensity information and shape information of the defect area.   The intensity information is the difference between the average brightness of the normal area of the inspection object and the brightness of the defect area, or the difference between the average chromaticity of the normal area and the defect area and the chromaticity, and the shape information indicates the size of the defect area. 4. The method according to claim 3, wherein the grade is an amount to be expressed.   3. The grade determination method according to claim 1, wherein the inspection object is a powder, and the characteristic point is a powder or a foreign substance having a color or shade different from that of a normal powder.   The grade determination method according to claim 1, 2 or 5, wherein the object to be inspected is methionine powder, and the characteristic point is a powder or foreign matter having a color or shade different from that of normal powder.   The grade determination method according to claim 1 or 2, wherein the inspection object is a sheet-like or film-like product, and the feature point is a region, a foreign matter, a flaw, or a bubble that is different in color or shade from the normal region. Three types of feature quantities x, y, and z are measured for a plurality of feature points existing on the inspection object, a plurality of grade regions are set in a space with x, y, and z as axes, and the feature points are A grade determination method characterized by plotting in space and determining the grade of an inspection object based on the number of feature points included in each grade area or the ratio of feature points in each grade area to all feature points.

Images