KR100903704B1 - The White and Brown Rice Appearance Characteristics Measurement System and the Same Method - Google Patents

Abstract

The present invention is a ratio of the complete rice and powdered rice, wraps, colored rice and cracked rice contained in a certain amount of rice and the ratio of the complete brown rice and powdered brown rice, wraps, colored brown rice, immature brown rice and the like contained in brown rice Apparatus and method for measuring the appearance quality of rice and brown rice which can be automatically determined.

In particular, in order to determine the appearance quality, instead of feeding rice or brown rice samples one by one in order to obtain high judgment accuracy in a short time, dozens of samples are supplied at once, and the samples are contained in a single image and the software is processed by computer image processing. An apparatus for determining the appearance quality of each grain is disclosed.

Apparatus for determining the appearance quality of the present invention, the sample stage 30 is composed of a transparent or translucent material and the sample is placed on the upper surface; A rotating disc 50 positioned below the sample table 30 to rotate the sample table 30; A sample supply member positioned above the sample stand 30 to supply rice or brown rice to the sample stand 30; An upper light 21 and a lower light 22 installed to be spaced apart from the sample table 30 to illuminate the sample; A CCD camera 20 installed at an upper side spaced apart from the sample stage 30 to take a sample on the sample stage 30; A computer 10 for processing an image photographed by the camera 20 using an image processing program to determine appearance quality; A frame grabber 11 installed inside the computer 10 to control the CCD camera 20 and to adjust the lens of the CCD camera 22 according to the amount of light of the upper light 21 and the lower light 22; A speed regulator 60 for receiving a signal relating to the operation state of the CCD camera 20 from the frame grabber 11 and controlling the sample supply speed from the sample supply container 40 and the rotational speed of the rotation disc 50; It is characterized by including.

Rice, brown rice, appearance grade, appearance, frame grabber, CCD camera, binarization, powdered rice, powdered brown rice, wrapped, whole rice, perfect brown rice, colored rice, colored brown rice, immature brown rice, image processing, color difference, pixel, pixel Otsu algorithm

Description

The White and Brown Rice Appearance Characteristics Measurement System and the Same Method

The present invention relates to an apparatus for evaluating the appearance quality of rice and brown rice, in particular, the ratio of complete rice and powdered rice, nagigi, colored rice and cracked rice, etc. contained in a certain amount of rice and complete brown rice and powder contained in brown rice The present invention relates to an apparatus for measuring the appearance quality of rice and brown rice, which is capable of automatically determining the ratio of fine brown rice, crumb, colored brown rice, and immature brown rice.

Korea has a long history of rice cultivation, but it is only 20 years ago that rice can be eaten to the fullest. However, in recent years, the inventories of rice have increased rapidly due to the absolute decrease in rice consumption and the increase in import volume due to the continuous harvest and minimum market access (MMA). As a result, inventories exceeded 10 million islands by the end of 2005. In addition, our rice does not have price competitiveness because the production cost is significantly higher than rice produced in the United States or China.

 On the other hand, due to the recent development of the national economy and changes in the national diet, the preference for quality rice has increased, and the necessity for producing high quality rice is increasing. In particular, in order to protect our rice in the situation of low price competitiveness compared to foreign rice, it is important to produce high quality rice that meets consumer's preference.

Finally, nutrition and taste are important for the quality of rice. However, the appearance of rice is often the subject of quality evaluation, depending on how the rice is used and who handles the rice. Related to the appearance quality of the rice is the ratio of complete rice, wraps, powdered rice, colored rice, etc. contained in the rice, these are known to affect the taste of rice as well as the commerciality of the rice. The Ministry of Agriculture and Forestry divides the quality of rice into three grades (Ministry of Agriculture and Forestry Notice No. 2005-59), and each grade has an appearance quality of rice such as moisture, waste, powdery rice, damaged grains, heat loss and other foreign matters. It is divided based on the criteria.

On the other hand, evaluation of the appearance quality of the rice is generally made by the naked eye of the tester. However, the results produced based on visual judgment can generally be influenced by the proficiency and personal bias of the inspector, the degree of fatigue, the type and brightness of the lighting. As a result, the objectivity of the judgment is low, not only is it difficult to make a quick judgment, but also requires a lot of effort and time. Therefore, there is a demand for the development of a mechanism for quickly and objectively evaluating the appearance quality of rice.

At present, foreign countries obtain a rice image by supplying grains of rice one by one, and a machine that evaluates the appearance quality by analyzing the image with a computer is used, and it is widely distributed in Korea. However, these devices are faster than visual judgment, but are relatively time-consuming, low precision, and expensive.

Machine vision is achieved by combining image processing technology and pattern recognition technology with computer technology. Machine vision is a technology with high potential for use in agri-food processing, and many studies have been conducted to apply machine vision to quality inspection and grading of agri-food. Traditionally, quality inspection of agri-food has relied on visual inspection. In most cases, visual inspection takes a long time, requires a lot of people, and reduces the precision of work. In contrast, inspection of agri-food using machine vision has proved to be much more consistent, convenient and cost-effective (Lu et al., 2000; Tao et al., 1995).

The use of machine vision can be divided into the image acquisition process of capturing images with a camera and converting them into digital images that can be recognized by a computer, and the image analysis process of extracting necessary information by analyzing digital images.

The success of image acquisition depends primarily on proper lighting. The characteristics of the illumination depend on the object and the characteristics of the image acquisition device used. Various types of lighting devices may be used, such as fluorescent lamps, halogen lamps, and LEDs, and methods such as front lighting, rear lighting, and inclined lighting may be used depending on the direction of light shining on the object. The imaging device scans the image formed on the sensor and converts the image into an electrical signal by using optical devices such as a point sensor, a line sensor, and a surface sensor implemented as a photo diode (PD) or a charge coupled device (CCD). . The analog video signal obtained here is spatially sampled in the x and y directions, and then converted into a digital video signal by A / D conversion. The image data after A / D conversion is stored as one image frame in the frame buffer. The frame buffer can acquire and display an image in real time.

The goal of applying machine vision is to extract the necessary information from the acquired images. To this end, the original image obtained by machine vision is improved to a shape related to a desired signal, which is called image processing. That is, various operations are performed to remove the miscellaneous signals included in the image and to extract features of the image. In the process of image processing, the important features of the image are emphasized without converting the information contained in the image, and techniques such as contrast enhancement, edge enhancement, drift removal, sharpening, and zooming are used. Image segmentation is an operation of separating an object from a background in order to extract necessary information from an image. For example, the background is separated from the rice image and only the rice image is extracted. The image segmented image is expressed in a form suitable for post-processing. Expression methods include expression of external characteristics (boundary) and expression of internal characteristics (pixel). The image is displayed using a two-dimensional equation, a function, or a special indicator, and the image in the two-dimensional space may be displayed as an area, a boundary line, or an outline. An area is a part of an image in which parts of similar or identical characteristics such as contour points, textures, and colors are distinguished and displayed. The boundary line is a boundary line that emerges from a process of obtaining an outline of an object or is obtained by various partitioning methods. Image Processor is dedicated hardware that executes specific image processing at high speed. It performs basic functions such as image acquisition, image storage, low-level image processing, and image display. It is necessary to process the image. Most image processing digitizes one screen in unit frame time (about 1/30 second).

Many studies have been carried out to measure and classify morphological, optical and organizational characteristics of grain using machine vision. Zayas et al. (1996) assessed wheat varieties, Ruan et al. (1997) measured the degree of infection of wheat rot, Ni (1997) measured corn grain size, and Xie and Paulsen (1997) measured the whiteness of corn. , Ni et al. (1997) for the identification of corn granules and grains, Steenhoek and Precetti (2000) for corn size grading, Liu et al. (1997) for rice degree measurement, and Yadav and Jindal (2001). Majumdar et al. (1997) studied the application of machine vision for the identification of wheat, barley, oats and rye. However, in these studies, grains were manually fed under the camera to obtain images so that grains could not be in contact with each other and spread over a large area. In order to solve this problem, Casady, Paulsen (1989) and Jayas (1999) developed a device for automatically supplying grains one by one. However, the measuring device is complicated, expensive, and takes a long time to measure.

To solve this problem, the image is processed to extract and analyze necessary information by supplying grains in contact with each other, acquiring images with a camera, and then separating the images of individual grains from the whole image by software. Attempt to develop an algorithm. The development of this algorithm is expected to contribute to the automation and simplification of the grain inspection system using machine vision.

As a study, Shatadal et al. (1995) developed an image processing algorithm using Mathematic Morphology that finds the boundary of contact area by eroding the grain image and then dilation. The rate of separation of grains was 95% in HRS and Durum wheat, 94% in barley, 89% in rye and 79% in oats.The thinner and sharper the grain, the longer the grain was in the longitudinal direction. When several pieces are in contact with each other, the accuracy of the grain recognition is lowered.

Shashidhar et al. (1997) developed a computer algorithm that uses ellipse fitting to create an ellipse that most closely approximates the shape of a grain and to recognize each ellipse as a grain. Then, Visen et al. (2001) recognized the contacted grains as contact ellipses using the elliptic spelling developed by Shashidhar et al. We developed an algorithm that recognizes individual grains by separating them. As a result, the results of experiments on wheat, barley, rye and oats showed an average success rate of 98.6%.

However, according to this algorithm, if the curvature change at the contact point is not clear or when separate grains are separated from the images of the grains contacted with each other, the boundary is irregular and the contact surface is not clear, and the degree of identification is poor. In addition, these methods have high applicability for grains of similar shape, but if heterogeneous grains or broken grains of different shapes are mixed or irregular shapes are complicated and complicated, the accuracy of separating and recognizing individual grains becomes low. Said.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned conventional problem, and in evaluating the appearance quality of rice and brown rice, dozens of rice or brown rice instead of feeding rice or brown rice one by one in order to obtain high determination accuracy in a short time It is an object of the present invention to provide an apparatus and method for analyzing the appearance quality of each grain by software by computer image processing by supplying all of the rice or brown rice in a single image.

That is, a test device for acquiring images of rice or brown rice while continuously supplying rice or brown rice, and recognizes one rice or brown rice by separating one from one image containing dozens of rice or brown rice, A computer that can identify the appearance quality of the grains of rice, ie, complete rice, cracked rice, wrap, colored rice, powdered rice, etc. and the appearance of brown rice, ie, complete brown rice, cracked brown rice, rapeseed, colored brown rice, powdered brown rice, etc. It aims to open image processing algorithm and program and finally automatic measuring device of appearance quality of rice and brown rice so that faster and more precise work can be done in determining appearance quality.

When rice is processed into brown or rice, brown rice contains whole brown rice, baekbaek brown rice, colored brown rice, broken brown rice, cracked brown rice, etc.The rice contains whole rice, baekbaek rice, colored rice, broken rice, cracked rice, etc. The higher the percentage of whole rice, whole rice or whole brown rice, the better the quality. Accordingly, the present invention constitutes a mechanical device with a supply device for supplying a sample of rice or brown rice, a CCD camera for taking an image of the sample, and a computer system for analyzing the image taken by the camera, and the computer system automatically supplies the sample when the sample is supplied. The appearance quality is determined and its number and composition ratio are shown. In determining the appearance quality of rice or brown rice, unlike the conventional system for supplying a sample one by one, the appearance of rice and brown rice can be determined even if dozens of samples are continuously supplied. An object of the present invention is to provide a quality measuring device and a method thereof.

Apparatus for measuring the appearance of the rice and brown rice of the present invention for achieving the above object, the sample stage is composed of a transparent or translucent material and the rice or brown rice sample is placed on the upper surface; A rotating member positioned below the sample stage to rotate the sample stage; A supply member positioned at an upper portion of the sample stage to supply rice or brown rice as a sample; An illumination member installed spaced apart from the sample stage and illuminating rice or brown rice; An image acquisition member installed at an upper side spaced from the sample stage to photograph rice or brown rice on the sample stage; A computer for processing the image photographed by the image acquisition member to determine a road quality by using an image processing program; A frame grabber installed inside the computer to control the image acquisition member and to adjust a lens of the image acquisition member according to the amount of light by the illumination member; And a speed controller configured to receive a signal regarding an operating state of the image acquisition member from the frame grabber, and to control a supply speed of rice or brown rice from the rice or brown rice supply member and a rotational speed of the rotating member. .

In addition, according to the device for measuring the appearance quality of rice and brown rice of the present invention, the image acquisition unit is characterized in that the CCD camera is installed to be able to change the focal length by adjusting the lens and to be movable in the vertical direction.

In addition, according to the milling quality measuring device of the rice of the present invention, the computer is a complete rice, powdered rice, chopped, cracked rice, colored rice or whole brown rice, powdered brown rice, chopped, golden persimmon brown, colored brown rice, immature brown rice It is characterized by automatically determining the appearance quality of the whole rice and brown rice by measuring the ratio or number of.

In addition, the method for measuring the appearance quality of rice and brown rice according to the present invention, the step of photographing the rice or brown rice placed on the sample stage with a camera; Storing an original image of rice or brown rice obtained by the camera in a data format corresponding to an RGB or YUV model by a frame grabber; Binarizing the stored image; Creating a contour two-dimensional image by distance transforming the binarized image to a shortest distance; Three-dimensionalizing the two-dimensional image and extracting a boundary line through a watershed transform; Separating the attached rice or brown rice based on the boundary line, and labeling each rice or brown rice; Each rice or brown rice labeled in the step is analyzed to determine the ratio of one or more of complete rice, powdered rice, cassia, cracked rice, colored rice, or whole brown rice, powdered brown rice, cassia, gold persimmon rice, colored brown rice, immature brown rice, and Characterized in that the step of determining the number.

In addition, according to the method for measuring the appearance quality of the rice and brown rice of the present invention, the step of binarizing the stored image, using the Otsu algorithm to actively calculate the threshold of concentration for distinguishing the rice or brown rice and the background It is done.

According to the apparatus and method for measuring the appearance quality of rice and brown rice of the present invention, when using the appearance quality measuring device of rice and brown rice, it is necessary to reduce the effort and time required for discriminating by the human eye, and to discriminate by mechanical criteria. There is an effect that can ensure the objectivity of.

In addition, according to the apparatus and method for measuring the appearance quality of rice and brown rice of the present invention, unlike in the system for determining the appearance quality of rice and brown rice by supplying one grain of rice or brown rice one by one, dozens of rice or brown rice are continuously It is possible to discriminate the quality of each rice or brown rice by supplying it as a product, and it is possible to reduce the product price drastically by not needing a device for supplying one grain of rice or brown rice one by one. It is effective to shorten the discrimination time since it is determined by.

Hereinafter, with reference to the accompanying drawings illustrating a device and method for measuring the appearance quality of rice and brown rice according to the present invention.

1 is a schematic view showing a device for measuring the appearance of rice and brown rice according to the present invention, Figure 2 is a real picture of the prototype of the device for measuring the appearance of rice and brown rice of the present invention, Figure 3 is 4 is a flowchart illustrating an image processing process, and FIG. 4 is a graph illustrating a density histogram of a digital image according to the present invention. 5 is a flowchart illustrating a separation process of attached rice, and FIG. 6 is a diagram illustrating a process of binarization without using an Otsu algorithm, and FIG. 7 is a diagram illustrating a process of binarization using an Otsu algorithm. to be. FIG. 8 is a diagram illustrating a distance conversion process, FIG. 9 is a diagram illustrating a process of finding a boundary line of rice, and FIG. 10 is a diagram of a map line process according to boundary separation of rice. to be.

Apparatus for measuring the appearance of the rice and brown rice according to the present invention, the sample stage 30 is composed of a transparent or translucent material and the rice or brown rice sample is placed on the upper surface; A rotating disc 50 positioned below the sample table 30 to rotate the sample table 30; Located in the upper portion of the sample table 30, the supply tank 40 for supplying rice or brown rice to the sample table 30 and the rice is installed in the supply cylinder 40 is supplied to the sample table 30 Or rotating roller 41 for adjusting the amount of brown rice; Upper and lower lights 21 and lower lights 22 respectively installed at upper and lower sides of the sample table 30 so as to be spaced apart from the sample table 30 to illuminate rice or brown rice; A CCD camera 20 installed at an upper side spaced from the sample stand 30 to photograph rice or brown rice on the sample stand 30; A computer (10) for determining an appearance quality by processing an image photographed by the camera (20) using an image processing program; A frame grabber 11 installed inside the computer 10 to control the camera 20; Receiving a signal regarding the operating state of the camera 20 from the frame grabber 11, the speed regulator for controlling the supply speed of rice or brown rice from the supply container 40 and the rotational speed of the rotating disc 50 ( 60); A discharge separator 51 installed on an upper surface of the sample stage 30 to separate rice or brown rice photographed by the camera 20; It is provided in the lower portion of the sample table 30 is provided with a collecting container 52 for recovering rice or brown rice separated by the discharge separation plate 51 and a support 70 for supporting these devices.

Here, the upper light 21 uses a higher illuminance than the lower light 22, the camera 20 is installed to be able to change the focal length by adjusting the lens and to move in the vertical direction. The frame grabber 11 adjusts the lens of the camera 20 according to the brightness of the upper light 21 and the lower light 22, and at the same time, the rotary roller 41 inside the supply container 40. Control the rice or brown rice supply. In addition, the computer 10, by measuring the ratio or number of one or more of complete rice, powdered rice, wraps, cracked rice, colored rice or complete brown rice, powdered brown rice, wrapped, gold persimmon brown, colored brown rice, immature brown rice The appearance quality of whole rice and brown rice is automatically judged.

Here, the process of the computer 10 to determine the appearance quality of the rice and brown rice is as follows.

Taking a sample of rice or brown rice placed on the sample table 30 with a camera, and stores the original image of the sample obtained by the camera 20 in the form of data corresponding to the RGB or YUV model by the frame grabber 11 After that, the stored image is binarized. Subsequently, a distance transformation is performed on the binarized image to a shortest distance to create a contour-shaped two-dimensional image, three-dimensionalize the two-dimensional image, and a boundary line is extracted through a watershed transform. Then, the attached samples are separated on the basis of the boundary line, labeling of each sample is performed, and each labeled sample is analyzed for complete rice, powdery rice, crumb, cracked rice, colored rice or whole brown rice, It is to determine the ratio and number of one or more of powdery brown rice, loaves, fried brown rice, colored brown rice, and immature brown rice.

Here, in order to binarize the stored image, it is preferable to use an Otsu algorithm so as to actively calculate a threshold value of the concentration for distinguishing the sample from the background.

Hereinafter, specific examples of the above-described general configuration, the experimental example of the appearance quality of the rice and brown rice of the present invention are shown as follows.

1. Experimental Materials

The rice used for the experiment was taken from the 2002 rice processed at the Migok General Treatment Plant and collected at the processing site to visualize the whole kernel, Broken kernel, Chalky kernel, and colored kernel. It was classified and used by inspection. The criteria for classifying rice were in accordance with the standards announced by the Ministry of Agriculture and Forestry. Complete grains are those that have unbreakable and broken rice, which have more than 3/4 of the total grain average length, excluding powdery granules, heat loss and damage grains.The average length of grains is complete grain 15 Or more were measured and calculated. The crumb was broken down by less than three quarters of the average length of the complete grain among those remaining on the sieve by striking a mesh of nominal dimensions of 1.7 mm in KS A 5101 standard. Powdery granules were granules in which at least 1/2 of the volume was powdery. The colored granules were grains in which at least 1/4 of the grain length was colored, including damaged grains and thermally damaged grains.

2. Appearance grade determination device

In order to measure the appearance quality of rice, an appearance quality judgment device consisting of a lighting device, an image processing device, and a rice supply / discharge device was fabricated. The configuration will be described with reference to FIGS. 1 and 2 as follows.

end. Lighting equipment

An incandescent lamp having a capacity of 12W is installed on the upper portion of the sample rack 30 to which the rice to be analyzed is set to the upper light 21, and two fluorescent lamps having a 4W capacity are installed on the lower portion of the sample rack 30. Illumination 22 illuminated the rice. At this time, the sample stage 30 was made of a white translucent acrylic plate in order to control the amount of light of the lower illumination 22 irradiated from the lower portion of the sample stage 30. Here, the sample stage 30 is preferably transparent or translucent, and if the brightness of the illumination is changed during measurement, an error may occur in the measurement result. When the brightness is different from the reference value by measuring the brightness of the illumination before the measurement, the camera 20 ) Lens was adjusted to match the reference value.

I. Image processing device

The image processing apparatus includes a color CCD camera 20 (LG Honeywell, GC-450NA-G), a frame grabber 11 (WDM method, BT878chip, self-made), and a PC 10 (Pentium IV 2.0G, DDR-RAM). 512MB). The image processing device fabricated for use in this study captures images of rice supplied using the CCD camera 20 and the frame grabber 11 installed in the PC 10, and displays the captured images of the PC 10. The image processing program was analyzed and stored.

The CCD camera 20 was adjusted to be able to move in the vertical direction, and a lens having a focal length of 6 mm and a brightness of F1.2 was attached. If the size of the rice grain image is changed due to the position of the camera 20 or the lens change, an error occurs in the measurement result. Therefore, if the size of the coin is different from the standard size before measuring, adjust the position and the lens of the camera 20. The size was adjusted to the reference value.

The frame grabber 11 has no RAM recognition limit compared to Windows 98 (depending on the specifications of the motherboard, but recognizes up to about 4GB of DDR-RAM recently). The self-manufactured using BT878 chip of) method was used. That is, since the conventional frame grabber recognizes only 128 MB even when using 128 MB or more RAM using the Windows 98 OS, it processes the high-capacity (640 x 480 x 3 = 921 600 Bytes) data transmitted from the camera 20. It takes a lot of time.

In addition, it is possible to capture with VFW (Video For Window) method by using the capture function of general low-cost TV card, but it takes more time to capture the video with 640 × 480 resolution suitable for research. Limited drivers must be provided. If the driver is not a VxD method, it is possible to capture at a resolution of 320 × 240 and below, and if a higher resolution is required, the API of the card maker must be known. In addition, the image processing program of the PC 10 was written using the Visual C ++ language in the MFC (Microsoft Foundation Class Library) method.

All. Rice feed and discharge device

The rice supply device for continuously supplying the sample includes a rice supply container 40 for discharging rice by a predetermined amount by the rotary roller 41 and a rotating disc 50 for transporting the supplied sample to the camera 20 below. The discharge separation plate 51 for separating the finished sample and a collecting container 52 for collecting it. At this time, if the rice is moved while the image is taken by the camera 20, accurate image acquisition is impossible, so the supply of the rice is stopped during the image capture and then supplied again when the image is finished. That is, the frame grabber 11 detects the operation state of the camera 20 to control the opening and closing of the power supply relay with an accuracy of 0.001 seconds unit to transfer the rotating roller 41 and the rice of the rice supply container 40 The operation of the motor driving the rotating disc 30 was controlled. In order to control the amount of rice to be supplied, the speed controller 60 (speed controller) to manually adjust the rotational speed of the motor driving the rotary roller 41 and the transfer rotary disk 30 of the rice supply container 40 It was. At this time, the lower light 22, the speed controller 60, the collecting container 52 was mounted on the support 70.

3. Image Processing

Image processing is the most basic processing for digital images, which removes noise included in the image, extracts features of the image, and performs a filtering operation for performing various basic operations. Image processing techniques can be classified into various types according to the application object. In this study, the order of the rice grains can be separated and identified from the images of rice supplied several times at a time. Image processing was performed by.

end. Image acquisition

The original image of the rice photographed by the color CCD camera 20 was stored in the memory in the form of data corresponding to the YUV2 model of 640 × 480 pixels by the frame grabber 11, and again the data of the YUV2 model was the data of the RGB24 model. Converted and stored in a buffer. In general, the color image is displayed in the RGB component, but the YUV format displays the image in the luminance component and the chrominance component because the memory can be used more efficiently than the RGB and can be processed faster. The reason why the YUV format can effectively use memory is that the color difference component is not given for each pixel but one color difference component per four pixels. That is, in the case of the 4: 1: 1 YUV format, four pixels constituting the image are bundled to sample each luminance Y, and the color difference U and V mean an average value of four pixels. Therefore, four pixels have four luminance and one color difference (U, V), resulting in a total of six samples. Thus, in the case of RGB, the memory is reduced to 1/2 when compared to 12 samples, 3 for 4 pixels each. In this case, 4: 2: 1 instead of 4: 1: 1 would result in a YUV2 model.

I. Binarization

In order to separate the background and the analyte to extract specific information from the acquired image, the digital image is represented only with two values of black (0) and white (255). When dividing the background and the characteristics of the analyte, it is easy to identify the object by setting the threshold of brightness and the value below 0 and the value above 255. The setting of the threshold may be determined by the user's judgment, or a technique of determining the dynamic threshold may be used. In this study, dozens of rice are supplied at one time to acquire one image, so the brightness of the image varies according to the number of rice supplied.

Therefore, when using a fixed threshold, an error occurs in binarization, so the dynamic threshold is determined using the Otsu method. The Otsu method is an algorithm that Otsu proposed in 1979 and is one of the most frequently referred algorithms. It dynamically finds the threshold value from the histogram of the image dynamically. The Otsu method is briefly described as follows. When an image of an image having gray levels from 0 to L-1 exhibits the same distribution as that of the histogram of FIG. 2, if k is an arbitrary gray level between 0 and L-1, 0 based on k It can be divided into the left area from to k points and the right area from k + 1 to L-1. In this case, the value of k is determined as the point where the variance between the two regions (σ2B (k)) becomes maximum.

Where k is an arbitrary threshold, μT is the mean value of the histogram, ω (k) is the zeroth-order cumulative moment, μ (k) is the first-order cumulative moment, and p (i) is the normal probability density function of the histogram, n (i ) Denotes the number of pixels (Pixel) at gray level i, and N denotes the total number of pixels.

In this method, the threshold value can be found actively according to the histogram that is changed when the brightness of the rice is changed due to the amount of rice to be measured, not the change in the histogram caused by external factors such as lighting and noise.

All. Boundary separation of attached rice

Even though the rice and the background are separated from the digital image, the rice is still attached to each other. Therefore, in order to extract information of individual grains of rice, all attached rice must be separated and recognized. In order to find and separate the boundary of the attached rice, a watershed algorithm was applied as shown in FIG. 3. First, the binarized image is distance-transformed to the shortest distance from the outside using a 3 × 3 Euclidean mask to create a contour-shaped image and three-dimensional the two-dimensional image to perform a watershed transform. . Identify the basin ridges of the attached rice from the three-dimensionalized image, make a separate image, synthesize it with the binarized original image, make a boundary line at the rice attached area, and separate the attached rice grains Extract the image.

Next, the rice is cut off by touching the edge of the image, the image is cut off, small particles, foreign matter much larger than ordinary rice (Edge off), and recognizes each rice grain separately from the image of the selected rice grain Labeling.

la. Determination of appearance quality of rice

In order to use as a standard sample for determining the appearance quality of rice, the rice was divided with the naked eye and classified into complete rice, wrap, colored rice, powdered rice, and cracked rice. These standard samples were photographed using the rice appearance quality determination device produced in this study, and image data used for appearance quality determination were extracted. In order to determine the appearance quality of rice in this study, long axis length, short axis length, width, area, circumference, red, green, blue, cyan, magenta, yellow ), Black (Hue Value), saturation (Saturation), the mean and standard deviation of the intensity (Intensity) and extracted the image data were used.

3. Rice appearance quality judgment algorithm

end. Image binarization and correction of rice brightness change

In separating the rice and the background from the digital image, as shown in FIG. 4, when binarization is applied to two images having different brightnesses by applying a threshold value of T = 165, the brightness and the background are not clearly separated from the rice. It can be seen that the threshold must be changed accordingly. This phenomenon occurs because the concentration histogram changes due to the change in brightness of each rice depending on the amount of rice supplied.

In order to prevent this, a computer program was developed to actively calculate the threshold of concentration for distinguishing rice from background using Otsu algorithm. FIG. 7 illustrates threshold values and binary results calculated for two images having different brightness using the Otsu algorithm. By using the Otsu algorithm, the concentration threshold for distinguishing rice from the background can be actively calculated. In comparison with Fig. 6 showing the result of binarization without using the Otsu algorithm, it can be seen that the performance of binarization is significantly improved.

I. Boundary separation of attached rice

Even though the rice is separated from the background in the digital image, several rice is attached to each other. Therefore, in order to extract information of individual grains of rice, all attached rice must be separated and recognized. The distance transform and watershed transform algorithms are applied to find and isolate the bounds of rice. FIG. 8 illustrates a process of distance transformation, and a distance map having a contour shape as shown in FIG. 8 (d) is created from the binarized image as shown in FIG. 8 (b).

When the rice overlaps, the boundary line is extracted as shown in FIG. 9 (a) by branch point transformation from the distance map image as shown in FIG. 8 (d) to divide the boundary. This is called watershed segmentation, which is divided using distance transformation, segmentation using slope, and segmentation using marker-control. In this study, watershed division by distance transformation is used. When the image of the basin ridge obtained in this way is inversed and then joined with the binary image, the boundary of the attached rice is recognized as being separated. Figure 9 shows the process of finding the boundary of the rice attached.

The accuracy of finding basin ridges is also strongly influenced by the distance map. In other words, if the distance of the distance map is too large, the basin ridges are hardly found, and if too tight, the normal rice image is separated into several directions as shown in FIG. 10 (b).

Here, to analyze each rice grain, as shown in Figure 9 (d) to independently distinguish the rice grains to create a name is called labeling. To do this, it is determined whether the pixels are separated from each other or attached to each other. The different pixels are given different names, and the same names are given to the attached pixels. In this study, a recursive component labeling algorithm was used.

All. Determination of appearance of rice

When rice is numbered by labeling, the area of each rice is calculated. The cheapness is determined based on the area of rice. In general, the area of rice is equal to the sum of the number of pixels occupied by the rice image. Therefore, the number of pixels of the rice becomes the area of the rice, and in this study, it is determined that the number of pixels of the rice image is less than 1,000. FIG. 11 shows an image of fire wrapped with a rice appearance quality determining device in this embodiment.

Powdered rice and colored rice were determined by comparing the color information of the rice image. First, powdered rice was divided into powdery rice when Red value> Blue value and Red value> Yellow value in RGB color coordinates. However, some of the complete rice may be distinguished as powdery rice. Therefore, if green value-black value> 6,000 among powdery rice, it was classified as perfect rice. On the other hand, the colored rice was divided into colored rice if Red value> Blue value and Red value <Yellow value, or if the number of colored pixels was 30 or more. However, among some powdery rice, it may be discriminated as colored rice. Therefore, if yellow value-Red value is 5000 or more in colored rice and (color pixel number) X 3 <(white pixel number), it was divided into powdery rice.

FIG. 12 shows images of powdered rice and colored rice photographed by a rice appearance quality determining device manufactured in this study. Any rice that was not broken, colored or cracked rice was determined as complete rice. Among cracked rice, cracked rice was determined as cracked rice.

The edge discrimination method was used for the determination of cracked rice. That is, the brightness of the cracked portion of the rice surface is identified using the difference between the brightness of the other portion. To this end, images of each grain of rice were separated from the images taken by the camera, and one image was made for each grain of rice. After calculating the orientation angle of the long axis of the rice with respect to the horizontal direction, the rice grains were rotated and arranged so that the long axis was horizontal, and the vertical edge components were extracted using the vertical edge discrimination operator. However, since all of the extracted vertical edge components are not crack components, vertical enhancement filtering was performed to make the vertical components thicker, and then the crack components were separated by binarization using the mean-standard deviation value. Then, thinning was performed to thin thick crack components, and small gold was removed to make pure crack images, and the original and crack images were synthesized to show cracks in rice.

Figure 13 shows the image of the complete rice taken by the rice appearance quality determination device produced in this study. In this case, the process according to the algorithm created to find the crack is shown in FIG.

4. Manufacturing of automatic grader

end. Structure of automatic grader

The rice milling grade automatic determiner was constructed as shown in Figures 1 and 2 to produce a starter. The rice milling quality measuring apparatus of this invention is comprised by the drive part which supplies rice, and photographs an image, and the computer which analyzes the photographed image. The driving unit is provided with a rotating roller 41, a rice supply cylinder 40 for supplying a predetermined amount of rice, and a rotating disc (Φ500mm) 50 made of translucent acrylic for transferring the supplied rice to the lower portion of the CCD camera 20. It was. A sample stand 30 is attached to the rotating disc 50, and one incandescent lamp having a capacity of 12 W illuminating rice by the upper light 21 and a CCD camera 20 (LG Honeywell, GC-450NA-G) are mounted. It was. Here, the CCD camera 20 is attached to the front of the lens with a focal length of 6mm, brightness F1.2 to adjust the focus and the amount of light, it was manufactured to adjust the height in the vertical direction. In addition, the discharge separation plate 51 and the collecting container 52 for discharging the finished rice image was installed, and under the rotating disc 50, two lower fluorescent lights 22 for 4W capacity was installed as the lower illumination (22), A speed controller 60 for adjusting the speed of the rotating roller 41 and the rotating disc 50 of the rice feed container 40 was attached.

The computer 10 used in the starter was Pentium IV (2,0G, DDR2-RAM 512M), and the O / S used Windows XP. The operation method of the starter is to put the rice to be measured in the rice feed container 40 and then press the rotary disk 50 operation switch to ensure that the rice supplied in front is supplied to the lower CCD camera 20, then the rotating disk 50 Stops working. Then press the brightness mode on the computer monitor to see if the brightness is within the reference range, then press the start mode, the rice is automatically supplied and analyzed. During the measurement, the rice image and analysis results are displayed on the monitor in real time. After the analysis, the result value is displayed on the monitor and saved as a text file in the computer storage. The frame grabber (11) built in the computer was manufactured by using BT878 chip of WDM (Windows Driver Model) method to be used in OS of Windows 2000 and Windows XP. The image processing program is MFC (Microsoft Foundation). Class Library) is written using Visual C ++ language.

I. Rice appearance quality determination performance

When rice was prepared by randomly preparing various types and colors of rice and analyzed using a starter, the accuracy of separating the attached rice by software and recognizing them individually was 98%. When Visen et al. (2001) used elliptic spelling to recognize the contacted grains as individual grains, the precision for wheat, barley, rye, and oats averaged 98.6%, and the curvature change at the point of contact was not obvious. When the individual grains are separated from the images of grains that are not in contact with each other or when the grains are in contact with each other, if the boundary is irregular and the contact surface is not clear, the degree of identification is reduced. The accuracy is much improved when compared with the results of the study, in which broken grains are mixed or irregular shapes are complicated and complex, the precision of separating and recognizing individual grains is lowered.

The starter was able to determine 100% cheapness, and shows the result of analyzing broken rice in FIG. 15. FIG. 16 shows the results of analyzing 94 powdery rice and 6 loaves. One of the powdery rice was identified as complete rice, and the discrimination accuracy was 98.9%. FIG. 17 shows the results of analyzing 95 colored rice and 5 loaves. One of the colored rice was identified as powdered rice, and the discrimination accuracy was 98.9%. FIG. 18 shows the results of analyzing 93 full rice and 12 broken rice. Three of the full rice were identified as powdery rice, and the discrimination accuracy was 96.8%. In addition, the number of cracked rice by visual inspection was 73, but the discrimination accuracy was 68. The accuracy of discrimination was 93.2%.

In addition, the time required for judging 100 grains of grain was about 15 seconds, which was very superior to that of the conventionally supplied judging device.

Although the preferred embodiment of the present invention has been described above, the present invention can be applied not only to brown rice but also to the appearance quality determination of other grains such as wheat, barley, oats and rye. In addition, the scope of the present invention is not limited only to this specific embodiment, and those skilled in the art will be able to appropriately change within the scope described in the claims of the present invention.

1 is a schematic view showing an appearance quality measuring device of rice and brown rice according to the present invention.

Figure 2 is a real picture of the appearance of the rice and brown rice of the present invention appearance measuring device of the present invention.

3 is a flowchart illustrating an image processing process of the present invention.

4 is a graph showing a density histogram of a digital image according to the present invention.

5 is a flow chart showing a separation process of the attached sample.

6 is a diagram illustrating a process of binarization without using an Oost algorithm.

7 is a diagram illustrating a process of binarization using an Otsu algorithm.

8 is a diagram illustrating a distance conversion process.

9 is a diagram illustrating a process of finding a boundary line of a sample to which the sample is attached.

10 is a diagram illustrating a map line process according to the boundary separation of the attached sample.

11 is a view showing an image of broken rice.

12 is a view showing images of colored rice and powdered rice.

13 is a view showing an image of a complete rice.

14 is a diagram illustrating a process of finding a crack.

15 is a diagram showing an analysis result of broken rice.

16 is a view showing the analysis results of powdered rice.

17 is a diagram showing an analysis result of colored rice.

18 is a view showing the results of analysis of complete rice.

19 is a view showing an image of broken brown rice.

20 is a view showing images of colored brown rice and powdery brown rice.

21 is a view showing images of whole brown rice and immature brown rice.

Fig. 22 is a diagram showing an analysis result of broken brown rice.

Fig. 23 is a diagram showing an analysis result of powdery brown rice.

24 is a diagram showing an analysis result of colored brown rice.

25 is a diagram showing an analysis result of complete brown rice.

<Explanation of symbols for the main parts of the drawings>

10: Computer 11: Frame Grabber

20: CCD camera 21: top lighting

22: lower illumination 30: sample stage

40: supply container 41: rotary roller

50: rotating disc 51: discharge separator

52: Collector 60: Speed Controller

70: support

Claims (9)

A sample table composed of a transparent or semi-transparent material and having a rice or brown rice sample placed on an upper surface thereof; A rotating member positioned below the sample stage to rotate the sample stage; A sample supply member positioned above the sample stage and supplying rice or brown rice as a sample; An illumination member installed spaced apart from the sample stage and illuminating rice or brown rice; An image acquisition member installed at an upper side spaced from the sample stage to photograph rice or brown rice on the sample stage; A computer which determines an appearance quality by processing an image photographed by the image acquisition member using an image processing program; A frame grabber installed inside the computer to control the image acquisition member and to adjust a lens of the image acquisition member according to the amount of light by the illumination member; And a speed controller configured to receive a signal regarding an operating state of the image acquisition member from the frame grabber, and to control a supply speed of rice or brown rice from the sample supply member and a rotation speed of the rotating member. And an appearance grade measuring device of brown rice. The method of claim 1, A discharge separator installed on an upper surface of the sample stage to separate rice or brown rice photographed by the image acquisition member; Apparatus for measuring the appearance of the rice and brown rice, characterized in that it further comprises a collecting tank for recovering the rice or brown rice separated by the discharge separating plate installed in the lower portion of the sample table. The method according to claim 1 or 2, The illumination member is composed of an upper light installed on the upper portion of the sample table, and a lower light installed on the lower portion of the sample table; Appearance of the rice and brown rice characterized in that the light reflected from the upper portion of the rice or brown rice by the upper light and the light transmitted through the rice or brown rice by the lower illumination to collect the image acquisition member to take an image Quality measuring device. The method according to claim 1 or 2, The image acquisition member is a device for measuring the appearance quality of rice and brown rice, characterized in that the CCD camera is installed to be able to change the focal length by adjusting the lens to move in the vertical direction. delete The method according to claim 1 or 2, The sample supply member is provided with a supply container in which rice or brown rice is stored, and a rotating roller installed inside the supply container and operated under the control of the frame grabber to adjust the amount of rice or brown rice supplied to the sample table. Apparatus for measuring the appearance of rice and brown rice. The method according to claim 1 or 2, The computer allows the menu to be selected according to the rice or brown rice, so that the whole rice, powdered rice, chopped rice, cracked rice, colored rice or whole brown rice, powdered brown rice, wrapped rice, cracked brown rice, colored brown rice, immature brown rice Apparatus for measuring the appearance of rice and brown rice, characterized in that automatically determine the appearance quality of the whole rice and brown rice by measuring the above ratio or number. Photographing rice or brown rice placed on a sample table with a camera; Storing the original image of rice or brown rice obtained by the camera in a data format corresponding to an RGB or YUV model by a frame grabber; Binarizing the stored image; Creating a contour 2-dimensional image by distance transforming the binarized image to a shortest distance; Three-dimensionalizing the two-dimensional image and extracting a boundary line through a watershed transform; Separating the attached rice or brown rice based on the boundary line, and labeling each rice or brown rice; Analyzing each of the labeled rice or brown rice in this step to determine the proportion and number of one or more of complete rice, powdered rice, chopped, cracked rice, colored rice or whole brown rice, powdered brown rice, wrapped, cracked brown rice, and colored brown rice. Appearance quality measurement method of rice and brown rice, characterized in that consisting of steps. The method of claim 8, The step of binarizing the stored image, the appearance quality measurement method of rice and brown rice using an Otsu algorithm to actively calculate the threshold of the concentration for distinguishing the rice or brown rice and the background.

Images