JPH09304181A - Color discriminator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow a color of a region with regions reflecting colors of objects other than an object whose color is to be identified by setting a region with high chroma or a region with high concentration appearing in all regions of an object image as a color extraction selected region. SOLUTION: Image data of an object of interest is acquired in an image acquisition part 101 from an image input source such as a camera or a scanner, and an input image is processed in a color discrimination object region extraction part 102 thereby extracting a color discrimination object region of the object image, and the region is set as a color extraction selected region. Thereafter, a color extraction part 103 color-extracts image data in the color extraction selected region with YUV signals of brightness and chrominance, and further distribution data of a Y-brightness value, a U-concentration value on a U-space and a V-concentration value is produced based on an extracted color signal. A color identification part 104 utilizes the color data to discrimination colors.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an area of high color saturation or an area of high density in a color identification object image by subjecting an image of the object captured by a color camera to image processing. The present invention relates to a color identification device that recognizes and identifies the color of a color identification target object by performing color extraction on the color identification target area.

For example, regarding a waste glass bottle color selection device, since the glass bottle has a relatively high color saturation, the area with high saturation is subjected to color extraction as a color identification target area, so that stains and illumination can be eliminated. It is possible to extract the original color of the glass bottle excluding the reflection portion and the shadow portion caused by the influence of, and as a result, it is possible to perform appropriate color recognition and color identification of the glass bottle.

[0003]

2. Description of the Related Art In color extraction and color identification using conventional color image processing, in color extraction, by irradiating light to an area excluding a portion that may cause trouble to a target object, A method of extracting the color of the target object (JP-A-7-159242), a method of extracting the color with the highest saturation in the target object (JP-A-1-204186), and setting a selection reference color range in advance. Method of extracting color range area (Journal of Powder Engineering Vol.32, No.6 PAGE.39)
2-395 1995 Regarding waste glass bottle color sorting device, there was Shosuke Otake) (Japanese Patent Laid-Open No. 7-282263). On the other hand, in the color identification processing, a color identification method is performed by using a color signal of polar coordinates (XYZ color system) in the XYZ color system (Powder Engineering Journal VOL.32, No.6 PAGE.39).
2-3951995 About waste glass bottle color sorter
Shosuke Otake) and a method of converting the Lab value of the Lab color system into four values consisting of lightness, hue, saturation, and tone, and performing color identification (instrumentation VOL.37, No.12 PAGE.37-40).
1994 There was a merit of applying color image processing technology in the glass bottle recycling field Kei Fujimura).

[0004]

There are the following two problems in the conventional color identification apparatus.

(1) Weak to changes in the state of the target object due to dirt, etc. (2) Weak to changes in luminance due to lighting equipment, etc. Further, as a solution to (1), the conventional method uses light irradiation as an obstacle. Select an appropriate area by excluding the part that causes the average color of the target object, extract the color with the highest saturation, set the selection reference color range in advance, and set the reference color range area. However, this method is effective only when the target object is composed of one color. Therefore, in order to appropriately perform color extraction even when the target object is composed of a plurality of colors, the present invention provides a highly saturated area in the object image area of the target object captured from an image input source such as a color camera or a scanner. , Or a method of extracting an area having a high density frequency as a color identification target area and extracting a color signal of the color identification target area to appropriately extract the original color of the color identification target object. Is.

Further, as described in (2), the conventional color identification method uses the color signals of the XYZ color system in the RGB color system, and thus is weak in luminance change. As a countermeasure, the XYZ color system is used. There is a method to convert the Lab value of the converted Lab color system into lightness, hue, saturation, and color tone, but this is also because the spatial distribution is different for each color, so always set quantitative identification conditions. There is a problem that you cannot do it.

It is an object of the present invention to use a color signal of the YUV color system in which the same color changes linearly with a change in luminance in the UV space, so that it is resistant to a change in luminance and further has a quantified standard color setting condition. It is to provide a color identification device for use in.

[0008]

A color identification apparatus and an application system according to the present invention include an image acquisition unit for acquiring image data of a color identification object captured by an image input source such as a color camera or a scanner, and an input image of an input image. In the object image,
A color identification target area extraction unit having means for setting a color extraction selection area to be a color identification target area, a means for performing color extraction on the color extraction selection area, and for recognizing and identifying from color extracted color signals And a color identification unit having a unit for identifying a color based on the color data. As a color extraction selection area setting method in the color identification target area extraction unit, by adopting a method of setting a high saturation area appearing in the entire object image area or a high density frequency area as a color extraction selection area,
The color of the area excluding the area other than the color identification target object of the object image due to dirt or reflected light is extracted.

Further, the color signal Y luminance value, the U density value in the YUV color system is used as a color signal handled by the color identification target area extraction section, the color extraction section, the color identification section and any means including them. A method of using the V density value is provided, and the distribution state of the U density value and the V density value in the UV space in the YUV color system is added to the YUV signal as color data, and the color data is used to perform color extraction and color identification. As a result, a color identification device that is resistant to changes in brightness is provided.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of a color identification apparatus showing an embodiment of the present invention. In the present invention, the image acquisition unit 101 acquires image data of a target object from an image input source such as a color camera or a scanner, and the color identification target area extraction unit 102 performs image processing on the input image to perform color identification on the object image. The target area is extracted, and the extracted color identification target area is set as the color extraction selection area. After that, in the color extraction unit 103, the image data in the color extraction selection area is color-extracted by the YUV signal of the luminance and the color difference, and the Y luminance value and the UV space in the UV space are obtained as color data based on the color extracted color signal. The distribution data of the U concentration value and the V concentration value of is created. Then, the color identification unit 104 performs color identification using the color data. The result is reflected on the output unit 105.

The processing performed in this embodiment will be described with reference to FIG. 1 for explaining the overall flow and FIG. 2 and subsequent figures for explaining the details of the processing.

When color extraction and identification start, the input image of the color identification object is acquired from the image acquisition unit 101 in FIG. 1, and at the same time, the object area is determined. Here, a color camera is used to acquire an image of the target object, and a polarizing filter is attached to the lens of the color camera to avoid the input of reflected light. In addition, transmitted light is used as the illumination method.

In the previous embodiment, the polarizing filter was attached, but there are cases where an image is acquired without attaching the polarizing filter. In the glass bottle color identification device and system, which is one of the applied systems, it is effective to mount a polarizing filter as a means for blocking the reflected light input from the glass bottle.
In addition, an image of the target object may be acquired using an image input source such as a scanner.

FIG. 2 shows the details of the processing contents of the color identification target area extraction unit 102 in FIG. The acquired input image is divided into a Y image and a UV image in the input image acquisition 202 in FIG. 2, and the UV image is further divided into a U image reflecting the U density value and a V image reflecting the V density value. Next, in the UV distance value image creation 203, the U image and the V image are subjected to image processing to create a UV distance value image. Where U
The V distance value has a point (128, 128) 501 indicating monochrome in the UV space as a starting point in FIG. 5, and the U density value and the V density value obtained from the U image and the V image are reflected in the UV space. The length of the vector 503 whose end point is the point (u, v) 502 is shown. Therefore, if the UV distance value is expressed as UV_DIST,

[0015]

[Equation 1]

It can be calculated by the following equation. An image in which the UV distance value obtained by this conversion formula is converted into a density value to newly express the UV distance value becomes a UV distance value image. In this way, histogram processing is performed in the UV distance value image created in the UV distance value image creation 203 in FIG. 2 in the histogram processing execution 204, and the histogram graph 601 in FIG. 6 is created. In FIG. 6, a graph 604 in which the horizontal axis 602 of the histogram graph 601 is the UV distance value and the vertical axis 603 is the UV distance frequency value is created.

Next, in the color discrimination target area setting calculation processing 205 in FIG. 2, the histogram graph 60 in FIG.
1 is used to set the UV distance threshold for extraction. The maximum UV distance value appearing on the horizontal axis 602 in the graph 604 is set as the maximum threshold value 605, and the minimum threshold value 607 is set so that the area of the shaded portion 606 corresponds to the area that is arbitrarily set. (In the present embodiment, as an example, the minimum threshold value 607 is set in the area corresponding to 10% with respect to the target object area.) The UV distance set in the color identification target area setting calculation processing 205 in FIG. The UV distance value image is binarized by the binarization processing execution 206 by using the threshold value as a binarization threshold value to create a binarized image. Next, in the logical product processing execution 207 of the binary image and the input image, the binary processing execution 206
The input image 701 and the binarized image 7 in FIG.
A logical product process is performed with No. 02 to create a logical product image 703. The logical product image 703 of the input image 701 and the binarized image 702 is used as the color identification target image, and the color extraction selection region setting 208 in FIG. Selection area 704
Set as

FIG. 3 shows details of the processing contents of the color extraction unit 103 in FIG. First, in the color extraction 302, the average density value of the color extraction selection area 704 in FIG. 7 is extracted. This average density value is set to the color signal setting 303
Is defined as the color signal Y luminance value, U density value, and V density value of the target object.

In the previous embodiment, the color extraction method using the distribution of the UV space is adopted. However, in the color extraction method using the density frequency, the color identification target area extraction unit 102 in FIG. Histogram processing is performed on each of the Y, U, and V images obtained by the input image acquisition 202 in 2 and the frequency range of high frequency is examined for each of Y, U, and V from the processing result. After that, the color extraction selection area is set by using the binarized image obtained by binarizing each of the Y, U and V images with the density range as a threshold and the mask image of the input image as the color identification target image. This color extraction method using the density frequency is effective, for example, in a glass bottle color identification device and system, which is one of the application systems, in the case of extracting a glass bottle of a low saturation color such as transparent or black. is there. In the method of extracting color for each block, color extraction is performed for each block for the entire area of the color identification target object image, and the color signal with the highest saturation is defined as the color signal of the color identification target object. There is. In addition, the YUV of the luminance and the color difference in claim 2
In the color extraction method of extracting the color by the signal, the area where the Y luminance value is extremely small or large is an area that appears as a result of the reflection portion or the shadow portion of the color identification target object due to the influence of the illumination or the like. Since the color of the color identification target object is not reflected, a range (for example, from Y brightness range 20 to 200) is given to the Y brightness value in advance, and an area having a color signal within the Y brightness range is set. Treatment only,
There is also means for setting the color identification target area. Furthermore, by converting the color-extracted signal with the YUV signal of luminance and color difference, R
In the color extraction method using the Lab signal in the method of extracting the color with the RGB signal of the GB color system, first, the RGB signal is converted into the Lab signal, and then the area in which the Lab signal having high saturation appears appears as the color extraction selection area. Color extraction. Further, in the color extraction method utilizing the density frequency in the method of color extraction with the color signals of the RGB color system,
An area in which RGB signals having a high distribution frequency appear is set as a color extraction selection area and color extraction is performed.

Next, the Y luminance value, the U density value, which are defined as the color signal of the target object in the color signal setting 303 in FIG.
Color data for color recognition and identification is created based on the V density value. The UV vector orientation setting 304 in FIG. 3 sets the UV vector orientation. Here, the direction of the UV vector is defined as θ504 in FIG. However, UV
Vector orientation

[0021]

[Equation 2]

It can be calculated by the following equation. FIG.
In the UV distance value setting 305 in (3), the UV distance value is obtained by the calculation formula of (Equation 1). The direction of the UV vector thus obtained, the UV distance value and the Y luminance value are set as color data setting 30.
6 defines the color data of the target object.

FIG. 4 shows details of the processing contents of the color identification section 104 in FIG. Color identification unit 10
In 4, the color of the target object is identified. here,
Some identification target reference colors are defined in advance. When one of the identification target reference colors is the reference color X, the hatched portion 801 in (a) UV space and the hatched portion 805 in the (b) Y luminance value graph in FIG. 8 are defined as the reference color X. Further, in FIG. 8, X_minDIST802 is the shortest UV distance value of the reference color X, and X_minθ803 is the UV of the reference color X.
The minimum inclination value of the vector direction, X_maxθ804 is the maximum inclination value of the UV vector direction of the reference color X, Y_min806 is the Y luminance minimum value of the reference color X, and Y_max807 is Y of the reference color X.
It is defined as the maximum brightness value.

Color identification processing start 40 in FIG.
Color identification is started by 1. The Y luminance value is included in the hatched portion 805 of the (b) Y luminance graph in FIG. 8 by determining whether the Y luminance value of the target object satisfies Y_min806 or more and Y_max807 or less by the Y luminance value judgment 402 in FIG. If so, the determination result is true. Next, in the direction determination 403 of the UV vector, U of the target object
V vector direction θ is X_minθ 803 or more and X_maxθ
804 or less is determined, and the UV distance value determination 404 determines that the UV distance value of the target object is X_minDIST8.
By determining whether or not 02 or more is satisfied, the point (u, v) 502 in (a) UV space in FIG.
If it is included in 1, the determination result is true. Y luminance value determination 402, UV vector direction determination 403, UV
If the result of the distance value determination 404 is true, the identification color determination 40
In 5 it is determined that the color of the target object is X, and the identification color determination 4
At 06, the identification color of the target object is defined to be X. Further, in the previous embodiment, as the method of defining the reference color X, the shaded area 801 in FIG. 8 is set using the range depending on the direction of the UV vector, but the reference UV of the reference color X in FIG. 9 is set.
Vector 901 is defined, and reference U is shown as shaded area 901.
There is also a method of defining the reference color X by using the distance to the V vector. The output unit 105 in FIG. 1 outputs the result of color identification of the target object as described above.

In the above embodiment, one representative color for the input image is color-extracted and identified.
It is also possible to perform multi-color extraction and color identification.
First, the color identification target area extraction unit 102 in FIG. 1 extracts a plurality of highly saturated areas that appear in the object area at one time. At that time, the same color region is set with the direction θ of the UV vector as a threshold value. Next, the color extraction unit 103 performs color extraction on each of the same color areas to create arbitrary color data. Then, the color identification unit 104 performs color identification for each arbitrary color data. As an application system, it can be used for various systems relating to color identification of fish, color identification of vegetables, color extraction and color identification.

[0026]

According to the color identification apparatus of the present invention, the color identification target area extraction unit sets the color identification target area in the input image of the color identification target object, and the color extraction unit extracts color from the color identification target area. Therefore, it is possible to extract only the original color of the target object without extracting the dirt adhering to the target object or the impure color due to the reflection of light, etc., and extract the color of the target object appropriately. Is possible.

Further, since the color identifying section of the present invention utilizes the fact that the same color changes linearly with the change in luminance in the UV space of the YUV color system, it is possible to perform strong discrimination against the change in luminance. .

[Brief description of drawings]

FIG. 1 is a block diagram of a color identification apparatus according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram of a flow of processing in a color extraction target area extraction unit.

FIG. 3 is an explanatory diagram of a processing flow in a color extraction unit.

FIG. 4 is an explanatory diagram of a processing flow in a color identification unit.

FIG. 5 is a conceptual diagram of color data determining means in the UV space in the YUV color system.

FIG. 6 is an explanatory diagram of a graph by histogram processing.

FIG. 7 is an explanatory diagram of a binarization method.

FIG. 8 is a conceptual diagram that defines a color identification method.

FIG. 9 is a conceptual diagram that defines a color identification method.

[Explanation of symbols]

101 ... Image acquisition unit, 102 ... Color identification target region extraction unit,
103 ... Color extraction unit, 104 ... Color identification unit, 105
... Output section.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Minoru Yanagibashi 3-2-2 Sachimachi, Hitachi, Ibaraki Prefecture Hitachi Engineering Service Co., Ltd.

Claims (10)

[Claims] 1. A color identification target area extraction unit for extracting a highly saturated area as a color identification target area from an image of a color identification target object captured by an image input source such as a color camera or a scanner, and the color. A color identification device comprising: a color extraction unit that extracts a color from an identification target region; and a color identification unit that identifies a color in the color identification target region. 2. The color identification target area according to claim 1, wherein an image input section for extracting a target object image with a YUV signal of luminance and color difference, and an area distributed in a portion of the target object image with high saturation in the UV space. Or a color extraction method for each block in which a target object image is divided into blocks and a block showing a YUV signal with the highest saturation is used as a color identification target area, or a target object 2. The color identification device according to claim 1, further comprising a color extraction unit characterized by a color extraction method using a density frequency in which an area having a high Y, U, V density frequency in an image is a color identification target area. 3. The method according to claim 1, wherein the color signal extracted by the YUV signal of the luminance and the color difference in the image input section is converted into the RGB signal of the RGB color system, and then the color signal is extracted, the RGB signal is converted into the Lab signal. A color extraction method using a Lab signal that is converted to a region in which a Lab signal showing high saturation appears in a target object image as a color identification target region, and a R, G, B density that is high in the target object image The color identification device according to claim 1, further comprising a color extraction unit that uses a density frequency in which an area where a value appears is a color identification target area. 4. The method for color extraction using color signals of the YUV color system according to claim 2, wherein a region having U density values and V density values distributed at points apart from a point indicating monochrome in UV space is colored. A color identification device including a color identification target area extraction unit, which extracts an area as a highly saturated area. 5. The method of color extraction using the distribution of UV space according to claim 2, wherein when a color identification target area is set, the distribution of U density value and V density value in UV space in the image area of the target object. The value of the distance from a point indicating monochrome in the UV space is set as a histogram, and the distribution of the histogram is used to set a highly saturated area as a color identification target area. The color identification device described. 6. The color identification apparatus according to claim 1, further comprising a method of setting an area having a specified saturation as a color identification target area. 7. A color extraction method according to claim 2, wherein a color signal is extracted using a color signal of a YUV color system, wherein an area where the Y luminance value is extremely small or large is reflected by an object to be color identified due to the influence of illumination or the like. By utilizing the fact that the color of the color identification target object is not reflected because it is an area that appears as a result of a portion or a shaded portion, a range is given to the Y luminance value in advance, and a color signal within the Y luminance range is provided. A color identification device characterized in that a color identification target area is set by performing processing only on the area. 8. A distribution of U density value and V density value in UV space, wherein the same color is distributed along an arbitrary radiation centering on a point indicating monochrome in UV space, and further linearly with respect to luminance change. 2. The color identification device according to claim 1, further comprising a color identification unit using means for identifying the color by utilizing the change. 9. A distribution of U density values and V density values in the UV space, in which the same color is distributed along an arbitrary radiation centering on a point indicating monochrome in the UV space, the number is calculated at a time. The color identification apparatus according to claim 1, further comprising a multi-compatible color identification method characterized by means for extracting and identifying the colors of colors. 10. The apparatus according to claim 1, further comprising means for removing the reflected light by attaching a polarizing filter to the lens of the color camera when the image of the color identification target object is acquired by the color camera. Characterized color identification device.