KR102625717B1 - Vision egg sorting system - Google Patents

Abstract

The present invention relates to a vision egg sorting system. More specifically, it provides a vision egg sorting system that determines whether the eggs are defective by performing a vision inspection of eggs passing through a specific section during the egg transport process.

Description

VISION EGG SORTING SYSTEM}

The present invention relates to a vision egg sorting system, and more specifically, to a vision egg sorting system that determines whether the eggs are defective by performing a vision inspection of eggs passing through a specific section during the egg transport process.

Farms or factories that ship eggs in large quantities select eggs that are defective as products, such as eggs with broken or cracked surfaces, blue eggs, eggs with foreign matter, eggs with lumps, and eggs with shell abnormalities. Although a considerable amount of labor and time was consumed in excluding eggs and packaging only eggs in good condition, this problem is gradually being resolved as egg sorters that automatically select defective eggs are developed and distributed.

This egg sorter is a sorting device that sorts eggs transported from a chicken house by a transport device such as a roller conveyor to a predetermined number of rows, aligns them in the same direction, and then performs one or more of crack detection or weight measurement. , it can be configured in various ways depending on the type, such as further including a process of washing and drying the transported eggs.

In this regard, prior inventions related to egg sorters include “Vision system including software control for detecting contaminants and other defects on the surface of eggs” in Korean Patent Publication No. 10-2010-0121638 and Korean Patent Publication No. Inventions such as "Egg sorting device for detecting and sorting defective eggs" in No. 10-2192613 have been proposed and disclosed.

First, the “Vision system including software control for detecting contaminants and other defects on the egg surface” in Korean Patent Publication No. 10-2010-0121638 uses a high-resolution camera and a non-linear diffuser for uniform illumination. By generating multiple high-resolution images for each of the eggs that are continuously transported along the egg processing device, and classifying eggs with defects detected through the images into acceptable and rejected, only eggs with no defects in quality as a product are selected. An invention has been proposed for a system that allows selective shipping.

In addition, the "egg sorting device for detecting and sorting defective eggs" of the Republic of Korea Patent Publication No. 10-2192613 is provided with a defect sorting unit to select defective eggs and a weight sorting part to sort the weight, thereby effectively detecting broken eggs. An invention has been proposed for a device that can quickly sort by weight and at the same time, has a fertilized egg reading unit that detects implantation, so that fertilized eggs can be distinguished from unfertilized eggs and individually packaged.

However, in the process of sorting the quality of eggs, the egg sorting machines so far have problems such as not being able to detect eggs with shell abnormalities such as eggs with chicken feces, eggs with foreign substances, eggs with lumps, or poor pigmentation. There is a problem with reliability.

Korean Patent Publication [10-2010-0121638] discloses a vision system including software control for detecting contaminants and other defects on the egg surface.

Korean registered patent [10-2192613] discloses an egg sorting device for detecting and sorting defective eggs.

Korean published patent [10-2010-0121638] (Publication date: 2010.11.18) Korean registered patent [10-2192613] (registration date: 2020.12.11)

Therefore, the present invention was created to solve the problems described above, and the purpose of the present invention is to provide a vision egg sorting system that determines whether the eggs are defective by performing a vision inspection of eggs passing through a specific section during the egg transport process. It is provided.

The purposes of the embodiments of the present invention are not limited to the purposes mentioned above, and other purposes not mentioned will be clearly understood by those skilled in the art from the description below. .

A vision egg sorting system according to an embodiment of the present invention for achieving the above-described object includes an egg transfer unit 100 including an egg transfer line including a conveyor; A vision camera unit 200 that acquires a color vision image of a specific area of the egg transfer unit 100; A lighting unit 300 that irradiates light to a specific area of the egg transfer unit 100, which acquires a color vision image with the vision camera unit 200; An egg detection unit 400 that determines whether an egg is present based on a first set value in the inspection area of the color vision image obtained from the vision camera unit 200; For the inspection area of the color vision image obtained from the vision camera unit 200, the pixels belonging to the RGB value or HSI value in the third set value range are within a predetermined distance between pixels. A defect determination unit 500 that determines defects based on the size of adjacent sets to which they belong; And controlling the vision camera unit 200, allocating a plurality of inspection areas among the color vision images obtained from the vision camera unit 200, and making decisions by the egg detection unit 400 and the defect determination unit 500. A control unit 900 that sets a standard, sequentially inspects each inspection area, inspects a plurality of vision images for one egg, and determines that an egg that has been judged defective at least once is finally defective. It is characterized by

In addition, the egg detection unit 400 is characterized in that it determines that an egg exists if it is greater than a first set value selected from 50 to 70% of the number of pixels occupying the entire inspection area.

In addition, the defect determination unit 500 is characterized in that it determines a plurality of defects selected from adhered foreign matter, chicken manure, stains, lumps, eggshell abnormalities, and cracks, according to the size of the pixel set group within the third set value range. .

In addition, the defect determination unit 500 is characterized in that it undergoes a pre-processing process for the color vision image before determination.

In addition, the defect determination unit 500 is characterized in that it performs the pre-processing process by leveling using an OCR tool.

In addition, the control unit 900 adjusts the sensitivity used in the pre-processing process of the defect determination unit 500 and the group size, which is the standard for determining defectiveness of the defect determination unit 500, to adjust egg test quality. It is characterized by being possible.

In addition, the control unit 900 is characterized by masking parts other than the egg image, which are background parts that have little influence on the egg detection unit 400's determination of whether an egg is present or defectiveness of the defect determination unit 500. .

According to the vision egg sorting system according to an embodiment of the present invention, by determining whether the chicken is defective based on the color vision image, eggs with chicken feces, eggs with foreign substances, eggs with lumps, and poor pigmentation are detected. It has the effect of improving the reliability of automatic egg inspection by detecting various defects in chicken, including eggs with abnormal egg shells and eggs with cracks.

In addition, by determining the presence or absence of an egg using the brightness value of the pixel, there is an effect of determining the presence or absence of an egg without using a dedicated sensor for detecting the egg.

Additionally, by going through the preprocessing process, there is an effect of reducing the amount of calculation required to determine whether or not an egg is defective.

In addition, by performing a leveling preprocessing process using an OCR tool, there is an effect of making it easier to perform the preprocessing process necessary to determine whether or not an egg is defective through a simple task using publicly available open source.

In addition, there is an effect of adjusting egg test quality by adjusting the sensitivity, which is the primary adjustment variable of test quality, and the group size, which is the secondary adjustment variable of test quality.

In addition, by masking parts other than the egg image, there is an effect of further reducing the amount of calculation required to determine whether or not the egg is defective.

1 is a block diagram of a vision egg sorting system according to an embodiment of the present invention.
Figure 2 is an exemplary diagram showing an operation of a vision egg sorting system according to an embodiment of the present invention.
Figure 3 is an exemplary diagram showing an example in which the egg detection unit of the vision egg sorting system according to an embodiment of the present invention is in operation.
Figure 4 is an example diagram showing an image without eggs in inspection area 1 of Figure 2.
Figure 5 is an exemplary diagram showing a color vision image obtained from a vision egg sorting system according to an embodiment of the present invention.
Figure 6 is an exemplary diagram showing an example in which the defect determination unit of the vision egg sorting system according to an embodiment of the present invention is in operation.
Figure 7 is an example diagram showing an image of surveillance station number 7 in Figure 6.
Figure 8 is an example diagram showing the process of inspecting a faint large foreign body after performing binarization based on the image of Figure 7.
Figure 9 is an example diagram showing a window that can adjust sensitivity, which is the first adjustment variable of inspection quality, and group size, which is the secondary adjustment variable of inspection quality.

Since the present invention can make various changes and have various embodiments, specific embodiments will be illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all changes, equivalents, and substitutes included in the spirit and technical scope of the present invention.

When a component is said to be "connected" or "connected" to another component, it is understood that it may be directly connected to or connected to the other component, but that other components may exist in between. It should be.

On the other hand, when it is mentioned that a component is “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between.

The terms used herein are merely used to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this application, terms such as “comprise” or “have” are intended to designate the presence of features, numbers, processes, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, processes, operations, components, parts, or combinations thereof.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by a person of ordinary skill in the technical field to which the present invention pertains. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and should not be interpreted as having an ideal or excessively formal meaning unless explicitly defined in the present application. No.

Hereinafter, the present invention will be described in more detail with reference to the attached drawings. Prior to this, the terms or words used in this specification and claims should not be construed as limited to their usual or dictionary meanings, and the inventor should appropriately define the concept of terms in order to explain his or her invention in the best way. Based on the principle of definability, it must be interpreted with meaning and concept consistent with the technical idea of the present invention. In addition, if there is no other definition in the technical and scientific terms used, they have meanings commonly understood by those skilled in the art to which this invention pertains, and the gist of the present invention is summarized in the following description and accompanying drawings. Descriptions of known functions and configurations that may be unnecessarily obscure are omitted. The drawings introduced below are provided as examples so that the idea of the present invention can be sufficiently conveyed to those skilled in the art. Accordingly, the present invention is not limited to the drawings presented below and may be embodied in other forms. Additionally, like reference numerals refer to like elements throughout the specification. It should be noted that like elements in the drawings are represented by like symbols wherever possible.

Figure 1 is a block diagram of a vision egg sorting system according to an embodiment of the present invention, Figure 2 is an exemplary diagram showing an embodiment of the vision egg sorting system in operation according to an embodiment of the present invention, and Figure 3 is a block diagram of the vision egg sorting system according to an embodiment of the present invention. It is an exemplary diagram showing an embodiment in which the egg detection unit of the vision egg sorting system according to an embodiment of the invention is in operation, and FIG. 4 is an exemplary diagram showing an image in a state where there are no eggs in inspection area 1 of FIG. 2, and FIG. 5 is an exemplary diagram showing a color vision image obtained from a vision egg sorting system according to an embodiment of the present invention, and Figure 6 shows an embodiment in which the defect determination unit of the vision egg sorting system according to an embodiment of the present invention is in operation. It is an example diagram, and Figure 7 is an example diagram showing the image of surveillance area number 7 in Figure 6, and Figure 8 is an example diagram showing the process of inspecting a faint large foreign object after performing binarization based on the image of Figure 7. , Figure 9 is an example diagram showing a window that can adjust the sensitivity, which is the first adjustment variable of inspection quality, and the group size, which is the secondary adjustment variable of inspection quality.

As shown in Figure 1, the vision egg sorting system according to an embodiment of the present invention includes an egg transport unit 100, a vision camera unit 200, a lighting unit 300, an egg detection unit 400, and a defect determination unit. It includes 500 and a control unit 900.

The egg transfer unit 100 includes an egg transfer line including a conveyor.

The egg transport unit 100 is a device installed for the purpose of shooting color vision images to inspect the condition of eggs before shipping. The filming location where the vision camera unit 200 and lighting unit 300, which will be described later, are installed is a dark room. It may be composed of a structure that allows this to be formed.

The rear end of the egg transfer unit 100 may be configured in various ways depending on the type, such as further including a process of selecting eggs judged to be defective and packaging only those judged to be normal. .

The vision camera unit 200 acquires a color vision image of a specific area of the egg transfer unit 100.

The purpose of using color vision images is to be able to detect defects that were difficult to detect in black-and-white vision images.

The color vision image may be an RGB vision image or an HSI vision image.

Computer vision is a technology that generates meaningful information by analyzing data about objects and their surroundings, such as objects, foregrounds, and backgrounds, from images or videos input through image input/output media such as cameras or scanners. , can be viewed as a field of artificial intelligence.

Cameras used in computer vision are called vision cameras.

The vision camera unit 200 may include multiple vision cameras.

The vision camera is seated on the top of the conveyor of the egg transport unit 100, and is rotated by a roller. It can generate multiple egg images by photographing the transported eggs several times in an internal dark room, and can generate multiple egg images. This configuration allows multiple egg images generated by the camera to be transmitted to an external sorting unit.

For example, if one vision camera can obtain a color vision image containing a total of 9 eggs in 3 rows and 3 columns, and the conveyor of the egg transfer unit 100 is composed of 12 columns, each is responsible for 3 columns. By arranging the four vision cameras so that the conveyor of the egg transfer unit 100 can inspect all 12 rows, the vision camera unit 200 can be configured.

If each vision camera acquires a color vision image every time an egg advances one row, 3 color vision images can be acquired for each egg.

In other words, it is possible to determine whether each egg is defective a total of three times based on three color vision images with different positions pointing upward.

In other words, the vision camera unit 200 can be installed at various positions on the conveyor of the egg transport unit 100, as needed, so as to obtain color vision images of all rows of the conveyor of the egg transport unit 100. there is.

In the above example, 4 vision cameras capable of taking pictures in 3 rows were placed in 12 rows. However, 7 vision cameras capable of taking pictures in 4 rows in 12 rows were placed at intervals of 2 rows, creating one Of course, various implementations are possible, such as arranging eggs so that they can be photographed by two different vision cameras.

If one egg is photographed from different angles with two different vision cameras, inspection blind spots can be further reduced.

The lighting unit 300 irradiates light to a specific area of the egg transfer unit 100, where a color vision image is acquired by the vision camera unit 200.

The lighting unit 300 is configured to help capture color vision images of eggs passing through a specific area of the egg transport unit 100, and irradiates light to eggs passing through a specific area of the egg transport unit 100 to provide clear images. It makes it possible to acquire color vision images.

The lighting unit 300 is preferably installed to illuminate a specific area of the egg transport unit 100 downward from the top of the conveyor of the egg transport unit 100.

This is because it is possible to obtain clearer color vision images only when light is illuminated from the vision camera.

At this time, the lighting unit 300 can be equipped with one or more lights, and their quantity and arrangement can be implemented in various ways.

Additionally, the lighting unit 300 can be adjusted to emit light only while the vision camera is taking pictures. In other words, it can be made to operate like a camera flash.

The egg detection unit 400 determines whether an egg exists based on the first setting value for the inspection area of the color vision image obtained from the vision camera unit 200.

The egg detection unit 400 is configured to determine the presence or absence of an egg, and determines whether there is an egg in the inspection area among the acquired color vision images.

The inspection area refers to an area used for determining the presence of eggs and determining defective eggs.

In order to store and manage the test results for each egg, the test results for each egg must be managed separately, and for this purpose, the test area can be set among the acquired color vision images.

Figure 2 shows an example in which nine inspection areas are formed in one vision image.

There are no eggs in areas 0 to 2 of Figure 2, and there are eggs in areas 3 to 9.

The obtained color vision image includes many eggs, and there are also many pixels unrelated to the eggs.

That is, in the obtained color vision image, there are parts where eggs can be filmed and parts where components other than eggs can be filmed, and the inspection area can be set with the part where eggs can be filmed as the main axis.

At this time, if the egg is large, the inspection area can be set to be large accordingly, and if the egg is small, the inspection area can be set to be small accordingly.

Of course, the first setting value can be a variety of values, such as RGB value, HSI value, and brightness value, as long as the presence or absence of an egg can be determined.

The defect determination unit 500 determines that pixels belonging to the RGB value or HSI value in the third set value range for the inspection area of the color vision image obtained from the vision camera unit 200 are pixels. Defects are judged based on the size of adjacent groups whose distances fall within a predetermined distance.

The defect determination unit 500 may determine the pixel to be defective if the size of the blob of pixels belonging to the RGB value or HSI value in the third set value range exceeds a set standard value.

In other words, even if a pixel belongs to the RGB value or HSI value used for defect determination, if it is a set of a certain size or less, it can be treated as noise.

The RGB value is a value composed of red, green, and blue.

HSI value is a value composed of Hue, Saturation, and Intensity.

Hue represents the primary color of that color. Saturation refers to the purity of a color and indicates how much white is mixed into the primary color. The HIS color model facilitates color conversion. For example, to turn red into pink, simply adjust the saturation. To make something dark brighter, simply increase the brightness. For this reason, image processing systems often use the HSI model when processing color images.

In some image processing, only brightness is used, and in such processing, it is more effective to use the HSI model than to use the RGB model. Additionally, it is effective to use the HSI model when classifying objects by color in computer vision.

In the HIS color model, color and brightness are expressed separately, but in RGB color, information about the brightness of light is not separated, and the color of an object is affected by the brightness of the light, so it is difficult to know whether it was the original color.

The HSI color model is expressed in a cone-shaped coordinate system. Colors are expressed as angles ranging from 0 to 360 degrees along the circumference of the cone. 0 degrees represents red, 120 degrees represents green, and 240 degrees represents blue. Saturation has a value from 0 to 1 and is expressed as the horizontal distance from the center of the cone.

The saturation value at the center of the cone is 0, making it 100% white, and at the edge of the cone, the saturation value is 1, making it a pure primary color with no white mixed in at all. Brightness corresponds to the vertical axis, with the lowest value being 0 and representing black, and the uppermost being 1 representing white.

The control unit 900 controls the vision camera unit 200, allocates a plurality of inspection areas among the color vision images obtained from the vision camera unit 200, and the egg detection unit 400 and the defect determination unit ( 500) judgment standards are set, each inspection area is inspected sequentially, and multiple vision images are inspected for one egg, and eggs that are judged defective at least once are finally judged defective.

The control unit 900 is responsible for overall setting and control of the vision egg sorting system according to an embodiment of the present invention, allocating inspection areas, setting judgment criteria, determining the order of progress, final defect determination, and setting up each vision camera. Functions such as adjustment (vision camera exposure time adjustment, etc.) can be performed.

The control unit 900 may allow the defect determination unit 500 to determine defects only in the inspection area where the egg detection unit 400 determines that eggs are present.

In other words, unnecessary inspection processes in areas without eggs can be omitted.

The egg detection unit 400, defect determination unit 500, and control unit 900 may be operated in the form of one program (indicated by a dotted line in FIG. 1).

That is, the egg detection unit 400 may function as an egg detection module, and the defect determination unit 500 may function as a defect determination module.

The egg detection unit 400 of the vision egg sorting system according to an embodiment of the present invention determines the total number of pixels whose brightness is higher than the minimum brightness value of any second setting value selected from 50 to 70% of the number of pixels occupying the entire inspection area. If the number is large, it can be characterized as determining that eggs exist.

Figure 3 shows the process of determining whether there is an egg in area 7 of Figure 2.

In the case where there are no eggs, an image as shown in Figure 4 can be obtained, and can be expressed as a dark area with a brightness value of approximately 20 or less, as in the center of Figure 4.

In other words, if the brightness value is used, the presence or absence of an egg can be determined based on the size of the bright part or the size of the dark part.

For example, if the first setting value is the minimum brightness value and the number of pixels with a brightness value higher than the first setting value is greater than the number specified by the second setting value, it may be determined that there is an egg.

Brightness values can be classified into a total of 256 values, from 0 (black) to 255 (white).

3 as an example, set the minimum brightness value as the first setting value to 41, find pixels with a brightness value of 41 or more, and if the total number of pixels with a brightness value of 41 or more as the second setting value is 18,000 or more, the egg It can be judged that there is.

In the above, an example of determining the presence or absence of an egg based on the minimum brightness value and the total number of pixels above the minimum brightness value was given, but the present invention is not limited to this, and the total number of pixels below the minimum brightness value is used as the basis, etc. Of course, if the presence or absence of an egg can be determined based on the brightness value, various implementations are possible.

Looking at inspection area No. 7 in FIG. 5, when there is an egg, it can be seen that the number of pixels in the dark area is less than the number of pixels in the dark area in inspection area No. 1 in FIG. 5.

As in inspection area 1 in 5, there are bright areas caused by rollers, etc. even in the inspection area without eggs, and the bright area in the inspection area without eggs occupies about 30%-40% of the total area, and in the inspection area with eggs, the bright area exists. The bright part of occupies about 80-90% of the entire area.

Therefore, if the bright area is more than 70% of the total inspection area, it can be judged that there is an egg, but considering unexpected situations such as small eggs mixed in or eggs turning in different directions, it is advisable to lower the standard a little. do.

The defect determination unit 500 of the vision egg sorting system according to an embodiment of the present invention selects a number of adhering foreign substances, chicken manure, stains, lumps, and abnormal eggshell cracks according to the size of the pixel group within the third set value range. It may be characterized as determining defectiveness.

The third set value range may be set in multiple numbers according to the defect determination criteria.

For example, as shown in Figures 6 and 8, the defect determination criteria can be classified into various categories such as faint large foreign matter, dark small foreign matter, lump (RGB), egg shell 14, egg white 17, crack 9, egg shell 74, etc.

A faint large foreign body refers to a foreign body formed in a large color that is little different from the color of the egg shell. Types of defects corresponding to the faint large foreign body include light bruises, dents, attached foreign bodies, chicken manure, stains, lumps, and egg shell abnormalities. This may include cracks, etc.

For example, the defective judgment criterion for a faint large foreign object may be that the size of the group with a brightness value of less than 103 when the sensitivity is set to 0.03 is 150 pixels or more.

Here, an OCR tool can be used to use sensitivity, and sensitivity will be explained later.

Dark small foreign matter refers to small foreign matter formed in a color that is significantly different from the color of the egg shell. Types of defects corresponding to dark small foreign matter include dark spots, adhering foreign matter, chicken manure, stains, lumps, and cracks in eggshell abnormalities. It can be.

For example, the defective judgment criterion for dark small foreign substances may be that the size of the group with a brightness value of less than 103 when the sensitivity is set to 0.2 is 10 pixels or more.

A lump (RGB) may correspond to a lump determined using an RGB image.

For example, the defective judgment criterion for lumps (RGB) may be that the size of the set in which each RGB value is 38 or less is 5 pixels or more.

The numbers for egg shell 14, egg white 17, and egg shell 74 are numbers assigned arbitrarily.

Eggshell 14 means that the types of defects that can be confirmed by eggshell abnormalities are defects 1 and 4.

Defect No. 1 is chicken manure, defect No. 2 is spray, defect No. 3 is lumps, defect No. 4 is egg shell abnormality (including defective eggshell pigment), defect No. 5 is foreign matter attached, defect No. 6 is egg yolk, defect No. 7 is egg shell abnormality (including defective eggshell pigment) Egg white, defect number 8 can mean alum (bruise), defect number 9 can mean cracks.

In other words, egg shell 14 refers to the defect judgment standard for detecting defect No. 1 (chicken manure) and defect No. 4 (eggshell abnormality).

For example, the defective judgment criterion for Eggshell 14 may be that the size of the set corresponding to the H value of 131 to 167, the S value of 5 to 22, and the I value of 92 to 108 is 10 pixels or more.

Egg white 17 refers to the defect judgment standard for detecting defect No. 1 (chicken manure) and defect No. 7 (egg white).

For example, the defective judgment criterion for egg white 17 may be that the size of the set corresponding to the H value of 150 to 240, the S value of 0 to 24, and the I value of 74 to 167 is 15 pixels or more.

Crack 9 refers to the defect judgment standard for detecting defect number 9 (crack).

For example, the defective judgment criterion for crack 9 may be that the size of the set corresponding to the H value of 127 to 167, the S value of 0 to 129, and the I value of 102 to 222 is 1 pixel or more.

Eggshell 74 refers to the defect judgment standard for detecting defects No. 4 (egg shell abnormality) and defect No. 7 (egg white).

For example, the defective judgment criterion for eggshell 74 may be that the size of the set corresponding to the H value of 25 to 230, the S value of 0 to 10, and the I value of 70 to 72 is 10 pixels or more.

When applying the defective judgment criteria for Eggshell 14 and Eggshell 74, when checking the HSI values for each, there are overlapping parts where some are the same and some are different, but if this range is specified at once, different parts are also included. , it is desirable to specify and use each.

Egg shell 14 is useful for distinguishing between white and 74 egg whites (the white part is broken and smeared, causing a reflection: the shiny part).

Another example of the HSI value used for each numbered defect determination criterion can be implemented as follows.

Eggshell 1 4: H(131~167), S(5~22), I(92~108)

Egg white 1 7: H(150~240), S(0~24), I(74~167)

Eggshell 7 4: H(202~242), S(0~46), I(46~82)

The standard range for determining defectiveness of the RGB value or HSI value described above can be calculated (minimum and maximum values) through training.

Training involves collecting the RGB or HSI values of defective eggs by checking the RGB or HSI values of defective eggs through vision images. says

If the color vision image acquired by the vision camera unit 200 is an RGB vision image, it can be changed to an HSI vision image depending on the situation and used, and obtained by the vision camera unit 200. If one color vision image is an HSI vision image, it can be changed to an RGB vision image depending on the situation.

For example, if the color vision image acquired by the vision camera unit 200 is an RGB vision image and the defect determination criteria is a faint large foreign object and a dark small foreign material, a gray scale image is used and a defective judgment is made. If the standard is RGB, the RGB scale image is used, and if the defect judgment standard is egg shell 14, egg white 17, crack and egg shell 74, the HSI scale image is used, the vision camera unit ( 200) can be converted and used to suit the situation.

The defect determination unit 500 of the vision egg sorting system according to an embodiment of the present invention may be characterized in that it undergoes a pre-processing process for the color vision image before determination.

For example, as shown in FIG. 8, when the defect determination standard is a faint large contaminant and a dark small contaminant, binarization in gray scale can be performed as preprocessing, and the defect determination criterion is a lump (RGB). , binarization can be performed on the RGB scale as preprocessing, and when the defective judgment criteria are eggshell 14, egg white 17, cracks, and eggshell 74, binarization can be performed on the HSI scale as preprocessing. can do.

If you perform binarization and extremely change the value range to the object color (black, etc.) or background color (white, etc.), the computational amount of object detection can be reduced.

When performing binarization processing, the image can be transformed into two colors and all values below a certain threshold can be transformed into object color (black, etc.) or background color (white, etc.).

When binarization is processed, the number of data is drastically reduced. Through this, the object to be detected can be transformed into a state that is easy to detect.

The defect determination unit 500 of the vision egg sorting system according to an embodiment of the present invention may be characterized in that it performs the above pre-processing process by equalization using an OCR tool.

Pre-processing in an OCR tool serves to correct the image so that the computer can more easily recognize the text area in the image.

There are several techniques used in the preprocessing process. First, the color image is converted to grayscale, the pixel values are analyzed, the brightness and contrast are increased, and the pixel values are divided into two ranges and classified as 0 and 1. This is to perform binarization.

In addition, several techniques such as spot removal, line removal, and layout analysis can be used.

There is a lot of data in images.

For example, an image of size 1,000 × 1,000 has 1,000,000 pixels.

And if an RGB value is assigned to each pixel, there are 256 × 256 × 256 distributions, so there can be 16,777,216 types of pixels.

That is, (0, 0, 0) becomes a black pixel, and (255, 0, 0) becomes a red pixel.

Each image is composed of various combinations of pixels. Analyzing the image itself without additional processing is very difficult and takes a long time.

Therefore, a process of reducing unnecessary data in the image and refining meaningful data is necessary.

At this time, a preprocessing algorithm can be used.

The preprocessing process is the process of processing images into meaningful information so that they can be efficiently used in image processing algorithms.

In other words, before a full-fledged algorithm is applied, it is desirable to simplify the data and leave only the data necessary for the algorithm.

The preprocessing algorithm serves to minimize parts that have a negative impact on judgment.

There are various preprocessing algorithms.

In image processing, preprocessing algorithms are an essential algorithm, and depending on how these algorithms are used, benefits such as accuracy, precision, and reduction of computation time can be obtained.

The control unit 900 of the vision egg sorting system according to an embodiment of the present invention adjusts the sensitivity used in the pre-processing process of the defect determination unit 500 and the group size, which is the standard for judging whether the defect determination unit 500 is defective. It may be characterized in that egg inspection quality can be adjusted by adjusting .

For example, if the test quality needs to be increased, the sensitivity can be increased, and if the test quality needs to be lowered, the egg test quality can be adjusted by lowering the sensitivity.

In addition, when inspection quality needs to be increased, the set group size can be reduced, and when inspection quality needs to be lowered, the set group size can be increased to adjust the primary inspection quality.

In other words, it is possible to adjust egg test quality by adjusting sensitivity, which is the primary adjustment variable of test quality, and group size, which is the secondary control variable of test quality.

Examples of OCR tools that use sensitivity (Confidence Threshold) are as follows:

Tesseract OCR: Tesseract is an OCR library that allows you to control recognition results by setting sensitivity. After performing OCR, the confidence score of each character or word returned can be checked and filtered if it is lower than a certain threshold.

Tesseract provides 'tessedit_char_whitelist', a parameter that can set sensitivity. This parameter can be used to control the reliability of OCR results. Confidence values are generally expressed in a range from 0 to 100, and sensitivity can be set by converting to an appropriate ratio.

AWS Rekognition: Rekognition from Amazon Web Services is an image and video analysis service that also provides OCR. Rekognition returns a confidence value for OCR results, and you can use this to adjust sensitivity.

AWS Rekognition's OCR feature can set its sensitivity using the 'MinConfidence' parameter. You can set the confidence threshold for OCR results by adjusting this parameter to a value between 0.0001 and 1.0.

Google Cloud Vision API: Google Cloud Vision is an image and video recognition service that also includes OCR. You can check the reliability of OCR results through the API and adjust sensitivity by setting a reliability threshold.

The sensitivity of Google Cloud Vision's OCR feature can be adjusted through the 'confidenceThreshold' parameter. You can adjust the confidence threshold for OCR results by setting this parameter to a value between 0.0001 and 1.0.

In addition, sensitivity may also be available in other commercial OCR tools and services. You can use the reliability of the OCR results to adjust recognition accuracy or filter recognition results as needed. Depending on the tool you use, setup methods and provided functions may vary.

Figure 9 shows a window where you can adjust the sensitivity, which is the first adjustment variable of test quality, and the group size, which is the second control variable of test quality. When the egg test quality is set to "sensitive," a faint foreign body (bruise) is displayed. , Splash) sensitivity is set to 0.03, and the faint foreign matter size is set to 150 pixels.

At this time, if the egg inspection quality is set to "Medium", the sensitivity of faint foreign bodies (bruises, splashes) can be changed to 0.05 to be applied, and if the egg inspection quality is set to "Insensitive", the sensitivity of faint foreign bodies (bruises, splashes) can be changed to 0.05. , Splash) can be applied by changing the sensitivity to 0.07.

The control unit 900 of the vision egg sorting system according to an embodiment of the present invention controls parts other than the egg image, which are background parts that have little influence on the egg presence or absence of the egg detection unit 400 and the defect determination of the defect determination unit 500. It may be characterized by masking.

Figure 3 shows an example of determining whether an egg is present in a masked state.

That is, in Figure 3, an egg is shown in the unmasked part of the test area, and the number of pixels with a minimum brightness value of 41 or more and a maximum brightness value of 255 or less in the unmasked part of the test area is This shows an example in which it is judged that there is egg opening if there are more than 18,000.

In Figure 3, the number of pixels with a minimum brightness value of 41 or more and a maximum brightness value of 255 or less was 19,989, and there were more than 18,000, so it was determined that there was an egg.

The present invention is not limited to the above-described embodiments, and the scope of application is diverse. Of course, various modifications and implementations are possible without departing from the gist of the present invention as claimed in the claims.

100: Egg transfer department
200: Vision camera unit
300: lighting unit
400: Egg detection unit
500: Defect determination unit
900: Control unit

Claims (7)

An egg transfer unit 100 including an egg transfer line including a conveyor;
A vision camera unit 200 that acquires a color vision image of a specific area of the egg transfer unit 100;
A lighting unit 300 that irradiates light to a specific area of the egg transfer unit 100, which acquires a color vision image with the vision camera unit 200;
An egg detection unit 400 that determines whether an egg is present based on a first set value in the inspection area of the color vision image obtained from the vision camera unit 200;
For the inspection area of the color vision image obtained from the vision camera unit 200, the pixels belonging to the RGB value or HSI value in the third set value range are within a predetermined distance between pixels. A defect determination unit 500 that determines defects based on the size of adjacent sets to which they belong; and
Controls the vision camera unit 200, allocates multiple inspection areas among the color vision images obtained from the vision camera unit 200, and determines the criteria for the egg detection unit 400 and the defect determination unit 500. A control unit 900 that sets and sequentially inspects each inspection area, inspects a plurality of vision images for one egg, and determines that an egg that has been judged defective at least once is finally defective;
Includes,
The egg detection unit 400 is
A vision egg sorting system characterized in that it is determined that eggs exist when the total number of pixels with a minimum brightness value is greater than any first setting value selected from 50 to 70% of the number of pixels occupying the entire inspection area.
delete According to paragraph 1,
The defect determination unit 500 is
A vision egg sorting system characterized in that it determines a number of defects selected from adhered foreign matter, chicken manure, stains, lumps, eggshell abnormalities, and cracks, depending on the size of the pixel set group within the third set value range.
According to paragraph 3,
The defect determination unit 500 is
A vision egg sorting system characterized by going through a pre-processing process for color vision images before judging.
According to paragraph 4,
The defect determination unit 500 is
A vision egg sorting system characterized in that the above preprocessing process is performed by leveling using an OCR tool.
According to clause 4,
The control unit 900 is
Vision egg selection, characterized in that egg test quality can be adjusted by adjusting the sensitivity used in the pre-processing process of the defect determination unit 500 and the group size, which is the standard for judging defects by the defect determination unit 500. system.
According to paragraph 4,
The control unit 900 is
A vision egg sorting system characterized by masking parts other than the egg image, which are background parts that have little influence on the egg presence of the egg detection unit 400 and the defect determination of the defect determination unit 500.