KR20090124335A - Color Factor Determination Method for Fruit Color Selection and Fruit Color Grading Method - Google Patents

Abstract

The present invention, in order to be able to best distinguish the color of the fruit changes the color of the skin according to the maturity, (i) a plurality of the color (surface color) and ground color (surface color) of the epidermis of the fruit samples to be screened Obtaining coordinate values of a color coordinate system, (ii) calculating a Mahalanobis distance for coloration and ground color by a plurality of preset color factors using the obtained coordinate values, and (iii) the calculated Mahal It provides a fruit color selection color factor determination method and a fruit color grade determination method comprising the ;; setting the color factor having the largest value of the Lanobis distance as the color factor for fruit color selection.

Description

Methods of determining color index for color grading of fruits and Method of grading color of fruits}

The present invention relates to a method for determining a color factor for fruit color selection and a method for determining a fruit color grade. In the case of discriminating a fruit grade such as an apple by color, the present invention determines a color factor that is significant for color distinction and uses the determined color factor. It relates to a color factor determination method for fruit color selection and fruit color grade determination method capable of determining the color grade of the fruit.

In the selection of harvested fruits, the color of the fruit epidermis is a very important quality factor representing the character of the variety. In the past, the human eye was mainly selected, but this requires a lot of labor and inaccuracy based on subjective standards. In order to solve this problem, a method of mechanically selecting fruits has recently emerged.

As the mechanical screening method, there are methods such as weight screening, color screening, and non-destructive sugar measurement screening. In the case of checkweighing, the sorting criteria are clear and easy to measure, but there is a problem in that the grade cannot be determined only by the weight of the fruit. Color classification is divided into a method using a black and white image, and a method using a color image. In the case of the screening method using black and white image processing, since the color is evaluated only in the light and dark extent of the skin color, accurate color determination is limited. In particular, the pale green area and the yellow area of the apple have a similar contrast, there is a problem to recognize the same color.

To overcome this, a method using a color visual device is used. As such a method, the method of using the average pixel value of fruit individuals was proposed conventionally. However, the method of using the average pixel value does not properly represent the grade of the fruit because a high pixel value (for example, when the red color is dark in some areas of the apple) has a large influence on the overall average. There is this. In addition, in order to investigate the pixel value and color grade section for each grade, the average pixel value for each grade of fruit should be obtained through preliminary experiments. Therefore, when changing the system, the grade standard should be readjusted according to the changed system. There is a problem.

Color factor determination method and fruit color grading method according to the present invention determines the fruit color selection method suitable for determining the grade of the fruit having the characteristic that the color of the skin is changed according to the maturity, and machine visual It is an object of the present invention to provide a method for determining color factors for fruit color selection and a method for discriminating fruit color grades, which do not require preliminary experiments for setting the standard for each grade, and facilitate color adjustment for each grade.

In order to solve the above problems, according to an aspect of the present invention, (i) obtaining coordinate values of a plurality of color coordinate system for the surface color and the ground color of the skin of the fruit samples to be screened; (Ii) calculating a Mahalanobis distance for coloring and ground color for each of a plurality of preset color factors using the obtained coordinate values, and (iii) having the largest value among the calculated Mahalanobis distances; A color factor determination method for fruit color selection is disclosed, comprising: setting a color factor as a color factor for fruit color selection.

The plurality of color coordinate systems may be at least one selected from the group consisting of RGB, rgb, Yxy, HSI, YIQ, and L * a * b * or a combination of two or more.

The color factors are R, G, B, G / (G + B), r, g, b, Y, x, y, y / x, Hue, Saturation, Intensity, ΔE, In-Phase, Quadrature It may be a combination of two or more selected from the group consisting of Q / I, L *, a *, b *, b * / a *.

The method may further include verifying whether the set color factor is significant in distinguishing coloration and color of the fruit to be selected.

According to another aspect of the present invention, (i) determining whether the fruit to be selected is present, (ii) calculating the value of the color factor set in claim 1 for each individual pixel constituting the region of the fruit determined to exist; (I) calculating the number of colored pixels whose color factor value calculated for the number of individual pixels is greater than a reference color value; (iii) a ratio of the number of colored pixels to the total number of the individual pixels; Disclosed is a fruit color grading method comprising calculating a coloring ratio and (iii) determining a color grade of the fruit to be selected based on the coloring ratio.

The determining of the presence may include (i) acquiring an image of the fruit to be selected, (ii) binarizing data of all pixels constituting the image, and (i) using a predetermined value among the binarized data. Labeling the binarized pixel data sequentially, labeling adjacent pixels with the same number, (i) calculating the largest number of label numbers among the label numbers, and (i) calculating the number of label numbers calculated as a reference size. And determining whether there is a fruit to be selected in comparison with the present invention.

When the fruit to be selected is an apple, the color factor may be b * / a *.

The grade may also be adjusted by changing a reference color value, which is a reference for determining whether the colored pixel is present.

According to the color factor determination method for fruit color selection according to an embodiment of the present invention, since the color grade is determined by the coloring ratio in consideration of the characteristics of the fruit maturing, the color of the fruit is well represented. The color factor that best distinguishes the coloration and the coloration is determined according to the fruit, thus providing accuracy in color screening.

In addition, according to the fruit color grade determination method according to an embodiment of the present invention, since the grade is determined by the color ratio, that is, the number of colored pixels / the total number of individual pixels through the measurement of the coloration and the color of the ground, setting the criteria for each grade through preliminary experiments The process is not necessary, and color adjustment for each grade can be easily performed only by changing the reference of the colored pixel.

Hereinafter, with reference to the embodiments shown in the accompanying drawings, the present invention will be described in detail.

1 is a flowchart illustrating a method of determining color factors for fruit color selection according to an embodiment of the present invention. Fruits to be screened may vary from apples, tomatoes, peaches, etc. Hereinafter, a method for determining color factors for screening apples will be described by taking apples as screens as an example. In order to determine the color factor for apple color selection, first, the coordinate values of various color coordinate systems for the surface color and the ground color of the apple skin are obtained.

Here, the ground color is the color of the epidermis of the fruit in the chlorophyll-rich state before the fruit matures, and in the case of apples, the coloring is red as the color changes as the fruit enters maturity. In order to obtain various color coordinates for the coloring and the color of the apples, the skin color of ripe apples with relatively even color characteristics can be divided into the coloring (red areas) and the skin color of the less ripe apples to green (green areas).

As a color coordinate system for color selection of apples, at least one or more color coordinate systems of RGB, rgb, Yxy, HSI, YIQ, and L * a * b * may be used. The RGB coordinate system means R (red), G (green), and B (blue), which are the three primary colors of light. The rgb coordinate system is a color component obtained by measuring R, G, and B for colors and normalizing them by a predetermined equation.

Experimentally, it is known that color cannot be displayed by the additive operation based on RGB coordinates. Complementary to this, the theoretical XYZ coordinate system was created, and based on this, conversion to all color coordinate systems is possible. The HSI coordinate system is based on the principle that the human eye senses objects, and expresses colors using Hue, saturation, and intensity, which are concepts in art. The YIQ coordinate system is a color coordinate system designed for color TV broadcasting, and uses the characteristic that human vision is more sensitive to contrast information than color information. Y (Luminance) represents contrast information, and I (In-phase) and Q (Quadrant) Indicates color information. The L * a * b * coordinate system converts the RGB coordinate system for standard white color into the XYZ coordinate system by a predetermined equation to obtain Xo, Yo, Zo, and the RGB coordinate system for the sample image by the predetermined expression X, Y, Z. Find and calculate by using them. Since the color difference can be calculated in this coordinate system, the relationship between the two colors can be explained according to the value.

In an embodiment of the present invention, color coordinate values in other color coordinate systems are calculated by performing color coordinate system transformation using L * a * b * values of coloring and ground color measured with a color system.

In the present invention, as a plurality of color coordinate systems, only RGB, rgb, Yxy, HSI, YIQ, and L * a * b * are exemplified, but the protection scope of the present invention is not limited thereto, and other color coordinate systems not illustrated are added. Can be.

The coloring of the apple peel and the measurement of the color of the green color can be carried out, for example, by measuring 30 colored apples and 30 colored apples using a colorimeter.

Now, in order to find a color factor that can separate the colored and colored groups well, R, G, B, G / (G + B), r, g, b, Y, x, y, y / x, Hue, Saturation, Intensity, ΔE, In-Phase, Quadrature, Q / I, L *, a *, b *, b * / a * color factors were investigated for coloration and color. ΔE is a separate value calculated in the L * a * b * coordinate system. The respective color factor values are calculated using the obtained color coordinate values. That is, color factor values for coloring and color factor values for ground color are calculated. The Mahalanobis distance for coloring and ground color is calculated for each calculated color factor. (S110) The Mahalanobis distance is a measure for statistically calculating the distance between two groups, and the farther the Mahalanobis distance is, the better. It can be said to be significant.

In the present invention, as a color factor, R, G, B, G / (G + B), r, g, b, Y, x, y, y / x, Hue, Saturation, Intensity, ΔE, In-Phase, Quadrature Although only Q / I, L *, a *, b *, b * / a * are exemplified, the protection scope of the present invention is not limited thereto, and other color factors not illustrated may be added.

Now, the calculated Mahalanobis distance for each color factor is compared with each other, and the farthest color factor is determined as the color factor for apple color selection (S120). This can be interpreted as having a significance that can distinguish between the coloring and the land groups.

Table 1 shows the Mahalanobis distance between the coloration and the coloration of apples by color factors.

Color factor Mahalanobis Street Color factor Mahalanobis Street RGB R 17.8 L ^* a ^* b ^* L ^* 96.2 G 125.0 a ^* 123.2 B 18.2 b ^* 69.5 HSI H 137.9 rgb r 74.2 S 26.1 g 130.7 I 58.9 b 0.7 YIQ Y 83.1 Yxy Y 83.1 I 2.8 x 57.4 Q 168.7 y 113.4 Q / I 148.8 y / x 121.5 b ^* / a ^* 493.2 G / (G + B) 108.4

In the present embodiment using an apple, it was found that the b * / a * value in the L * a * b * coordinate system has the largest distance difference between the coloring and the ground color, and then the Quadrature, Quadrature and In of the YIQ coordinate system. The difference in the Mahalanobis distance was shown in the order of -phase ratio, Hue representing color, g normalizing green, and a * value. That is, b * / a * was determined as the most significant color factor for apple color selection. Therefore, the color factor b * / a * value can be used to distinguish whether the apple is colored.

In an embodiment of the present invention taking an apple as an example, b * / a * was determined as the most significant color factor for color selection, but the protection scope of the present invention is not limited thereto, and the type or variety of fruit is changed. It should be understood that other color factors may be set as the most significant color factor.

Finally, it is verified whether the set color factor (b * / a * in the present embodiment) is significant in distinguishing the coloring of the apple from the green color (S130). In the case of determining the coloring part only by the coordinate system value, it is difficult to understand it realistically. Therefore, the color paper that has the color closest to the value of the color coordinate system suggested for the coloring and the ground color is selected, and whether or not the color factor b * / a * set previously can be distinguished from the coloring and the ground color using the selected color paper. It is necessary to set the threshold value for class determination).

2 is a graph showing color factor values for coloration and coloration of apples and color factor threshold values for distinguishing coloration and coloration.

As shown in the figure, the color factor values for coloring and ground color were clearly divided into two groups, and when b * / a * is expressed in degrees, a b * / a * value of about 71 degrees is associated with the coloring of the apple. It can be seen as a threshold for distinguishing color. That is, when the value of the color factor b * / a * is set to 71 degrees, if the color factor value of any pixel is less than 71 degrees, this pixel is determined as a colored pixel, and if it is larger than 71 degrees, this pixel is determined as a color pixel. . As described above, according to the color factor determination method for fruit color sorting according to an embodiment of the present invention, color adjustment for each grade is easy because only the threshold value of the color factor is changed.

Figure 3 shows a fruit color grading method according to an embodiment of the present invention. With reference to FIG. 3, the method to determine the color grade of the fruit used as a grade determination object, for example, an apple, is demonstrated.

First, it is determined whether there are apples to be screened.

To this end, an image of where the apple is located is obtained.

Then, all the pixel data in the image are binarized (S210). For example, the pixel data is divided into light and dark color parts based on a predetermined value and binarized to 1 and 0 correspondingly. Can be.

After the binarization step, all of the binarized pixel data are sequentially searched from the upper left to the lower right of the image, for example, and the binarized pixel data is sequentially labeled (S220). 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0... Only the pixel data having the value of S is sequentially numbered, but the same number is given to adjacent pixel data. In this case, therefore, 1, 0, 2, 2, 2, 0, 3, 3, 3, 3, 3, 0, ... are labeled corresponding to the pixel data.

Thereafter, the label number having the largest number among the label numbers is calculated (S230). In the above example, the label number 3 is calculated because the label number 3 is the most as five. Since the image of the part where the apple is located is photographed in a lighter color than the background, the part where the apple is located will be continuously binarized to 1. Therefore, the label number in the portion is given the same number, and the size of the area corresponding to the label number corresponds to the area occupied by the apple.

Therefore, after determining whether the number of the largest number of label numbers is greater than or equal to the reference value (S240), if it is greater than or equal to the reference value, it may be determined that an apple exists in the photographed image. If the value is less than the reference value, the apple has not yet been transported or positioned to the shooting position. Therefore, the image is taken after a predetermined time without performing image processing for color grading, and the above process is repeated to determine whether the apple exists. To judge.

When it is determined that an apple exists in the photographed image, the value of the color factor determined above is calculated for each individual pixel representing the individual in the fruit area (S250). The pixels corresponding to the label number considered in operation S240, that is, the pixels in the region in which the apple is recognized as present.

It is compared whether or not the calculated color factor value in each individual pixel is greater than the reference color value (S270). That is, it is determined whether each individual pixel is a color pixel or a color pixel. The reference color value will be the color factor threshold described above.

If the color factor value in an individual pixel is smaller than the threshold of the set color factor (b * / a *), the individual pixel is determined to be a colored pixel, and the total number of colored pixels is increased by one and the total number of individual pixels is increased. Increase one (S280).

On the other hand, if the color factor value is larger than the threshold value of the set color factor, the object pixel is determined to be a color pixel, and in this case, the total number of individual pixels is increased by one (S260).

In this manner, it is determined whether all individual pixels are colored pixels or colored pixels (S290).

When this determination is completed for all individual pixels, the coloring ratio is calculated (S300). The coloring ratio is a percentage obtained by dividing the total number of colored pixels by the total number of individual pixels.

Finally, the color grade of the fruit is determined based on the coloring ratio (S310). For example, if the coloring ratio is 80% or more, the grade 1 is determined. Can be.

Thus, the color grade is determined by the coloring ratio which is the ratio of the coloring and the ground color in consideration of the characteristic that the fruit matures, so it can be said that it represents the color of the fruit well.

In addition, the coloration ratio through measurement of coloration and ground color, that is, selection by the number of colored pixels / total individual pixels does not require a process for setting a standard for each grade through preliminary experiments, and it is easy to adjust the color for each grade simply by changing the standard of the color pixels. can do.

The invention can be used in the post-harvest processing industry of fruits.

1 is a flowchart illustrating a method of determining color factors for fruit color selection according to an embodiment of the present invention.

2 is a graph showing color factor values for coloration and coloration of apples and color factor threshold values for distinguishing coloration and coloration.

3A and 3B are flowcharts illustrating a method for determining fruit color grades according to an embodiment of the present invention.

Claims (8)

Acquiring coordinate values of a plurality of color coordinate systems for the surface color and the ground color of the epidermis of the fruit sample to be selected; Calculating a Mahalanobis distance for coloration and ground color for each of a plurality of preset color factors using the obtained coordinate values; And And setting a color factor having the largest value among the calculated Mahalanobis distances as a color factor for fruit color selection. According to claim 1, The plurality of color coordinate systems are at least one selected from the group consisting of RGB, rgb, Yxy, HSI, YIQ, and L * a * b *, or a combination of two or more colors. According to claim 1, The color factors are R, G, B, G / (G + B), r, g, b, Y, x, y, y / x, Hue, Saturation, Intensity, ΔE, In-Phase, Quadrature , Color factor determination method for fruit color selection, which is a combination of two or more selected from the group consisting of Q / I, L *, a *, b *, b * / a *. According to claim 1, And determining whether the set color factor is significant in distinguishing the coloration and the color of the fruit to be selected. Determining whether a fruit to be selected is present; Calculating a value of the color factor set in claim 1 for each individual pixel constituting an area of fruit judged to exist; Calculating the number of colored pixels whose color factor values calculated for the individual pixel numbers are larger than a reference color value; Calculating a coloring ratio that is a ratio of the number of the colored pixels to the total number of the individual pixels; And And determining a color grade of the fruit to be selected based on the coloration ratio. The method of claim 5, The determining of the presence or absence, Obtaining an image of the fruit to be selected; Binarizing data of all pixels forming the image; Sequentially labeling pixel data binarized to a predetermined value among the binarized data, but labeling adjacent pixels with the same number; Calculating the largest number of label numbers among the label numbers; And And comparing the calculated number of label numbers with a reference size to determine whether or not the fruit to be selected is present. The method of claim 5, And the color factor is b * / a * when the fruit to be selected is apple. The method of claim 5, And the grade can be adjusted by changing a reference color value, which is a reference for determining whether or not the colored pixel.

Images