Quality Analysis and Classification of Rice using Image processing

Abstract

Grain quality analysis is a huge challenge in agricultural industries. Internal control is critical in the food industry because food products are characterized and rated into various categories after quality data has been collected. Grain quality assessment is performed by hand, but the results are subjective, lengthy, and pricey. To overcome the limitations and drawbacks of image processing techniques, different resolutions are used for grain quality analysis. Using image processing techniques, this paper proposes a method for grading and analyzing rice based on grain size and form. An edge detection algorithmic software is used in particular to determine the area of each grain\'s borders. we discover the endpoints of each grain using this technique, and we can then live the grain\'s endpoints using caliper.

Introduction

I. INTRODUCTION

Agricultural enterprise is the oldest and maximum huge enterprise withinside the world. Traditionally, the first-class of meals are described via way of means of a human sensory panel primarily based totally on its bodily and chemical properties. Physical parameters encompass grain length and form, moisture content material, chalk, whiteness, and freeness. For the top-of-line storage, the moisture content material needed to be among 12-14%. Various techniques are used for moisture evaluation.

The primary goal of the proposed method is to provide an alternative approach to quality analysis that requires less time and money. Image processing is an important and advanced technological subject that has seen significant progress. Attempts are being made to replace human manual detection. The document suggests a solution to agribusiness problems.

II. PROBLEM DEFINITION

Product quality analysis is critical in the agriculture sector. An experienced technician visually evaluates the quality of the grain seed. However, the outcome of such an assessment is comparative, varies in results, and takes a long time. The technician's attitude also has an impact on quality; as a result, a new and improved methodology, namely an image processing technique, is presented to address the flaws that have evolved as a result of old ways.

A. Quality and Classification

Grain quality assessment is a significant concern in agriculture. In the food sector, quality control is critical since food is categorized and divided into several classes based on quality factors after harvesting. Grain quality testing is manual, but it is subjective, time-consuming, and expensive. Using image processing techniques, the research provides a method for classifying and grading grains based on grain size and form. Specifically, edge detection to determine each grain's border. We may determine the endpoints of each grain using this technique, and then measure the length and width of the rice using vernier calipers. This process takes very little time and is very affordable.

The image processing technology is used to count the number of rice and classify them based on length, breadth, and length-breadth ratio. The length-width ratio is calculated as follows: length equals the average length of the rice grain, and breadth equals the average breadth of the rice grain.

                                          L/B = [(average rice length) / (average rice breadth)] *100

B. Image Acquisition and Processing

A camera is used to capture the image. This is depicted in Figure 1. On the computer, the captured image is saved. Image processing methods are applied to the image after it has been saved.

III. METHODS

Fig. 2 depicts the flow of the image processing method, which consists of a few basic phases. For image acquisition, rice grains are scattered at random on a black background. The image is saved to be analyzed later. The first phase is pre-processing, which involves image registration and noise removal via a filter. The Shrinkage algorithm is used to segment the touching kernels in the second stage. We have a tendency to use edge detection in the third stage to find the boundary region. The rice grain measurement, as well as length, breadth, and length-breadth measurements, are completed in the fourth stage. Rice is categorized in the fifth stage of the algorithm based on its size and form.      

A. Image Pre-processing

The image acquired with a camera is saved in the computer's 3-D RGB color space, as seen in fig 1. The filter is used to remove noise that happens during the image acquisition process. The image is also sharpened by the filter. The rice grains are segmented from the black background using a threshold technique, and the image is converted to a grey image as seen in fig 3.

B. Shrinkage Morphological Operation

Rice grains are scattered randomly across a black background. The grains in Figure 1 are not oriented in any way. When contacting grains occur, morphological operations can be used to categorize them. Grain touching can be separated into two types: point and line touching. The combination of dilatation and erosion is a morphological surgery. Erosion is a technique for separating adjacent parts of a grain of rice without compromising its integrity. The erosion process is followed by the dilation process. The purpose of dilatation is to restore the original shape of degraded features without re-joining the divided elements.

In the vision and motion toolbox, there are different types of morphological operations are available such as;

1. Auto M - Auto median,
2. Close - Dilation followed by an erosion,
3. Dilate - Dilation (opposite of erosion),
4. Erode - Erosion which removes isolated background pixels,
5. Open - Erosion followed by dilation,
6. P close - A succession of seven closings and openings,
7. P open - A succession of seven openings and closings.

C. Edge Detection

As illustrated in Fig.6, edge detection aids in locating the region of rice grain boundaries. Gaussian, Gradient, Prewitt, Canny, Fuzzy, and Sobel are six edge detection algorithms offered in Vision and Motion Toolbox. In the proposed methodology, we use the Sobel method to find edges.

D. Object Measurement

The number of grains of rice in the image represents the number of individual grains in the evaluation. After counting the number of grains of rice, edge detection techniques are implemented to the image, and endpoint values for each grain are obtained as a result of the algorithm's application. We use a caliper to connect the endpoints and evaluate the length and breadth of every grain. Once we have the length and breadth values, we can calculate the length to breadth ratio.

E. Object Classification

All outcomes must be standard, measured, and calculated to be classified. The laboratory manual on rice grain quality, Board of Rice analysis, Rajendranagar, Hyderabad, provides the standard information for measuring the size and form of rice.

The table below shows how rice grains are classified based on the length and length-to-breadth ratio:

IV. RESULT AND DISCUSSION

Table 4 shows the results obtained from putting image processing algorithms into action. The length-breadth ratio of each grain in the input image is shown in the result.

                                                                               Table2. Results for L/B Ratio

Images in which rice grains are randomly arranged and dispersed in a layer are subjected to image analysis techniques. When a fault occurs, such as touching kernels, the shrinking process effectively separates the connecting section from the points where the kernels are touching. Edge detection is used to determine the range of each grain's boundaries and endpoints, after which the length and width can be measured with a caliper. The length-breadth ratio is calculated when the length and breadth values have been determined. A dash app is created to see the results of the Average aspect Ratio Vs Classification chart and a pie chart for Quality analysis of the input image.

Grouped Bar chart – Used for Classification purposes

• Blue Bar indicates the Number of Rice grains.
• Red Bar indicates Average Aspect Ratio.

       Pie chart – Used for Quality Analysis purposes

• Blue Section indicates the percentage of Rice grains in the given sample.
• Red Section indicates the percentage of Dust in the given sample.

V. FUTURE WORK

The majority of quality analysis factors must be measured using image processing techniques. This research could be expanded to develop a method for identifying granules based on any attribute that can be used to improve rice quality. The cost of such a system should be low, as should the time spent on quality analysis.

Conclusion

In this project, we classify the taken rice grain sample into different categories and also analyze its quality based on the aspect ratio, so comparison with other works is not possible. Existing work only detects the rice grains or calculates the number of rice grains in the given sample, but our work helps analyze the quality of the rice sample and place it into a specific category. The quality of the grains in the samples is nearly 100% accurate and capable of efficiently classifying high-value grains, which is otherwise very time-consuming in manual analysis. This function can save a lot of time and manpower.

References

[1] “Laboratory Manual on Rice Grain Quality”, Directorate of Rice Research, Rajendranagar, Hyderabad, September 2013. [2] Mahale, Bhagyashree, and Sapana Korde. \"Rice quality analysis using image processing techniques.\" International Conference for Convergence for Technology-2014. IEEE, 2014. [3] Ali, S.F., Jamil, H., Jamil, R., Torij, I. and Naz, S., 2017, November. “Lowcost solution for rice quality analysis using morphological parameters and its comparison with standard measurements”. In 2017 International Multi-topic Conference (INMIC) (pp. 1-6). IEEE. [4] “Documentation Python”. Online: https://www.python.org/doc/ [5] “Documentation Open CV”. Online: https://opencv.org/about/ [6] IBPGR-IRRI Rice Advisory Committee, “Descriptors for Rice Oryza Sativa L”. International Rice Research Institute and International Board for Plant Genetic Resources, 2020. [7] Divya Mohan, M.Raj., Semanticscholar 2016, ”Quality Analysis of rice grain using ANN and SVM”.

Copyright

Copyright © 2022 Chityala Udhay, Rishabh Semwal, Krishna Goel, G Ravi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.