JP3135731B2 - Pattern recognition method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern recognition method and an apparatus for automatically recognizing an image of a pattern (for example, characters, patterns, signs, special shapes, etc.) formed on a surface to be inspected.

[0002]

2. Description of the Related Art A two-dimensional pattern formed on a surface to be inspected is imaged by a camera, and a two-dimensional pattern signal obtained is
A partial pattern signal belonging to each of the plurality of partial pattern regions is cut out, and when the number of the cut out partial pattern signals having picture element signals indicating that there is a pattern is equal to or greater than a first threshold value, the pattern is extracted. Generating a first signal indicating that there is no pattern, and generating a second signal indicating that there is no pattern when the signal is equal to or less than a second threshold value smaller than the first threshold value, and A system in which a two-dimensional pattern is recognized from a combination of partial pattern regions having a second signal has already been proposed (see Japanese Patent Publication No. 61-9562).

[0003]

However, in the above-mentioned pattern recognition method, the generation of a "pattern present" signal or "no pattern present" signal is determined when the number of pixels having picture element signals exceeds a first threshold value. Alternatively, it is performed when there is a value equal to or smaller than the second threshold value (that is, binarization processing is performed). It is necessary that illumination during imaging is uniform. However, when images are actually taken by a camera, it is extremely difficult to make the illumination uniform, which may be a bottleneck in practical use. In addition, in the case of the method of the above-mentioned known example, there is also a problem that high-speed processing is difficult because the calculation required for the binarization processing is extremely complicated.

In addition, the surface to be inspected is imaged by a color camera,
Although it is conceivable to recognize the pattern from the obtained captured image, even in this case, it is not possible to uniformly irradiate the surface to be inspected under predetermined conditions, resulting in uneven illumination. In addition, there is a problem that the brightness of the image is uneven.

[0005] In the case where such uneven illumination exists, there is a possibility that erroneous recognition occurs frequently unless parameters such as a spatial filter for stabilizing the imaging luminance level are readjusted.

[0006] The present invention has been made in view of the above-mentioned point, and it is possible to remove unevenness in brightness of a picked-up image due to unevenness in illumination and then perform pattern recognition using a monochrome-converted image, thereby being affected by lighting conditions. It is an object of the present invention to enable high-speed pattern recognition without any problem.

[0007]

According to a first aspect of the present invention, as a means for solving the above-mentioned problems, a pattern on a surface to be inspected is imaged by a color camera, and a uniform luminance level is set for each of three primary color signals in the imaged image. After the illuminance change correction is performed, the corrected three primary color signals are converted into monochrome signals, and then the pattern is image-recognized.

According to a second aspect of the present invention, as a means for solving the above-mentioned problems, in a pattern recognition apparatus, an image of a pattern on a surface to be inspected is taken by a color camera and the pattern is automatically image-recognized. Illuminance change correction means for correcting each of the three primary color signals in the captured image to a uniform luminance level; monochrome signal conversion means for converting the three primary color signals corrected by the illuminance change correction means into monochrome signals; and the monochrome signal conversion means. And a pattern recognizing means for recognizing the pattern based on the monochrome signal obtained by the above.

[0009]

According to the method of the first aspect or the apparatus of the second aspect,
The following effects are obtained by the above means.

That is, the three primary color signals in an image picked up by a color camera for image recognition are respectively corrected to a uniform luminance level, and the corrected three primary color signals are converted into monochrome signals. Will be recognized. Therefore, even if illumination unevenness occurs on the surface to be inspected, the three primary color signals in the captured image are corrected to a uniform luminance level and converted to a monochrome signal, so that pattern recognition can be performed without being affected by brightness unevenness on the image. You can do it.

[0011]

According to the method of the first aspect or the apparatus of the second aspect, for each of the three primary color signals in an image captured by a color camera for image recognition, each of the three primary color signals is corrected to a uniform luminance level, and the corrected three primary colors are corrected. After converting the signal into a monochrome signal, the pattern is image-recognized. Therefore, even if illumination unevenness occurs on the surface to be inspected, pattern recognition can be performed at high speed without being affected by brightness unevenness on the image. Thus, there is an excellent effect that high-speed and stable pattern recognition can be performed without being influenced by the lighting conditions.

[0012]

Preferred embodiments of the present invention will be described below with reference to the accompanying drawings.

In this embodiment, as shown in FIG.
Pattern P on inspection surface W by color camera 1
(In the case of the present embodiment, ABC characters), and the three primary color video signals Vr, Vg, V in the captured image thus obtained.
b is to be input to the image processing controller 2 that performs various image analysis and the like. The controller 2
Includes an illuminance change correction unit 3 and an image processing unit 4.
Here, the pattern P is formed on the inspection surface W, and is, for example, a character, a pattern, a sign, a special shape, or the like.
Vr indicates a red video signal, Vg indicates a green video signal, and Vb indicates a blue video signal.

An image picked up by the color camera 1 is
The three primary color video signals Vr, Vg, Vb are taken into the illuminance change correction unit 3, and the illuminance change correction unit 3 performs a process of correcting each of the video signals Vr, Vg, Vb to a uniform luminance level. To be corrected video signals Vr ', Vg', Vb ',
And performs analysis processing, and performs processing such as conversion to monochrome signals and pattern recognition. The illuminance change correction section 3 functions as illuminance change correction means, and the image processing section 4 corrects the corrected video signals Vr ', Vg', Vb 'in a state corrected by the illuminance change correction means 3.
Signal conversion means 5 for converting the image into a monochrome signal Y
A pattern recognition unit 6 for recognizing the pattern P based on the monochrome signal Y obtained by the monochrome signal conversion unit 5; and an output unit 7 for outputting a result recognized by the pattern recognition unit 6. (See FIG. 2).

Next, the operation of the illuminance change correction unit 3 will be described with reference to the block diagram shown in FIG. The illuminance change correction unit 2 includes a plurality of calculation units 11 to 20, and configures calculation logic for correcting the video signals Vr, Vg, Vb of the three primary colors to a uniform luminance level.

First, the video signals Vr, Vg, Vb input from the color camera 1 are respectively taken into the normalization calculation units 11, 12, and 13, and the bias signals B are obtained from the video signals.
r, Bg, and Bb are respectively subtracted and removed, and gain signals Gr, G
g and Gb are respectively multiplied and output as normalized signals Sr, Sg and Sb, respectively.

Next, the above-mentioned normalized signals Sr, Sg, Sb are taken into the color signal correction calculating sections 14, 15, 16 respectively, and the correction reference values SBr, SBg, SBb relating to the color levels are converted into the normalized signals Sr, Sg, SB. respectively the division by and a predetermined value α each Sb power is taken ratio reference color level, i.e. the correction factor K ₁ of the color level is obtained. Then, the correction factor K ₁ of the normalized signal Sr, Sg, is outputted as the correction value for the color level SrK _1, SgK _2, SbK ₃ by multiplying each to Sb.

The normalized signals Sr, Sg, Sb output from the normalization calculation units 11, 12, 13 are taken into an illuminance change rate calculation unit 17, and are squared for each of the correction reference values SBr, SBg, SBb. , The sum of the correction reference values SBr, SBg, and SBb (SBr ² + SBg ² + SB
b ² ) is squared for each of the normalized signals Sr, Sg, Sb to obtain the total sum, that is, divided by the vector sum of the normalized signals Sr, Sg, Sb (Sr ² + Sg ² + Sb ² ). Illuminance change rate (S
Br ² + SBg ² + SBb ² ) / (Sr ² + Sg ² + Sb ² ) is obtained, and a power is raised to a predetermined value β to obtain a correction coefficient K ₂ at the illuminance level.
Is output as

Thereafter, the color signal correction calculation section 14,1
The correction values SrK ₁ , SgK ₂ , and SbK ₃ , which are outputs of 5 and 16, are taken into the illuminance correction calculation units 18, 19 and 20, and are multiplied by the correction coefficient K ₂ obtained by the illuminance change rate calculation unit 17. As a result, the illuminance is corrected, and corrected video signals Vr ', Vg', and Vb 'are obtained.

The corrected video signal V thus obtained is
r ′, Vg ′, and Vb ′ are optimal for use as signals for pattern recognition because the rate of change in illuminance has been corrected (that is, the influence of uneven illumination has been removed). It has become.

The corrected video signals Vr ', Vg', Vb 'obtained as described above are taken into the monochrome conversion means 5, as shown in FIG.

Y = 0.30Vr ′ + 0.59Vg ′ + 0.11Vb ′ Next, the operation of the pattern recognition apparatus according to the present embodiment will be described with reference to the flowchart shown in FIG. The subscript i in the flowchart indicates red (r), green (g), and blue (b) sequentially.

Each of the three primary color video signals Vi (i = r, g, b) in the image of the pattern P picked up by the color camera 1 is normalized on a reference scale and normalized signals Si (i = r, g). , b) (step S ₁ ), and then performs a color signal correction calculation and an illuminance change rate calculation to obtain a correction value Si at a color level.
(i = r, g, b) and the correction coefficient K ₂ are obtained (steps S ₂ ,
S ₃ ), and an illuminance correction operation is performed based on them to obtain a corrected video signal Vi ′ (i = r, g, b) (step S ₄ ). After a while
The correction video signal in step _{S 5 Vi '(i = r} , g,
Convert b) a monochrome signal Y, after an image recognition pattern P on the basis of the monochrome signal Y in step S _6, it has decided to output the result in step S _7.

As described above, according to this embodiment, each of the three primary color video signals Vr, Vg, Vb in the pattern P picked up by the color camera 1 for image recognition is corrected to a uniform luminance level. After converting the corrected video signals Vr ', Vg', Vb 'in the
Since the pattern P is image-recognized, even if illumination unevenness occurs on the inspection surface W, pattern recognition can be performed at high speed without being affected by brightness unevenness on the image. Therefore, high-speed and stable pattern recognition can be performed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a main part of a pattern recognition device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing contents of a controller in the pattern recognition device according to the embodiment of the present invention.

FIG. 3 is a block diagram showing a calculation logic of an illuminance change correction unit in the controller of FIG. 2;

FIG. 4 is a block diagram showing the contents of a monochrome signal conversion means in the controller of FIG. 2;

FIG. 5 is a flowchart for explaining the operation of the pattern recognition device according to the embodiment of the present invention.

[Explanation of symbols]

1 is a color camera, 2 is a controller, 3 is an illuminance change correction unit (illuminance change correction unit), 4 is an image processing unit, 5 is a monochrome signal conversion unit, 6 is a pattern recognition unit, 7 is an output unit, 11, 12, 13 is a normalization operation unit, 14, 15, and 16 are color signal correction operation units, 17 is an illuminance change rate correction operation unit, 1
8, 19, and 20 are illuminance correction calculation units, W is a surface to be inspected, P is a pattern, Vr, Vg, and Vb are three primary color signals (video signals), Sr,
Sg, Sb normalized signal, SBr, SBg, SBb the correction reference value, K _1, K ₂ is a correction coefficient, Vr ', Vg', Vb ' are corrected three primary color signals (correction video signal), Y monochrome signal.

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields surveyed (Int. Cl. ⁷ , DB name) G01B 11/00-11/30 G06T 1/00 G06T 7/00-7/60

Claims (2)

(57) [Claims] 1. A pattern on a surface to be inspected is imaged by a color camera, and illuminance change correction for correcting each of three primary color signals in the imaged image to a uniform luminance level is performed.
A pattern recognition method comprising: converting a corrected three primary color signal into a monochrome signal; and then performing image recognition of the pattern. 2. A pattern recognition apparatus for picking up a pattern on a surface to be inspected by a color camera and automatically recognizing the pattern, wherein each of the three primary color signals in the image picked up by the color camera is corrected to a uniform luminance level. Illuminance change correction means, a monochrome signal conversion means for converting the three primary color signals corrected by the illuminance change correction means into a monochrome signal, and an image of the pattern based on the monochrome signal obtained by the monochrome signal conversion means. A pattern recognition device, comprising: a pattern recognition unit for recognizing the pattern.