Quality Analysis and Classification of Rice Grains using Image Processing Techniques

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 10 Issue: 08 | Aug 2023 www.irjet.net p-ISSN: 2395-0072
© 2023, IRJET | Impact Factor value: 8.226 | ISO 9001:2008 Certified Journal | Page 311
Quality Analysis and Classification of Rice Grains using Image
Processing Techniques
Harshith Singathala, Jyotsna Malla, Preetham Lekkala
1Harshith Singathala, School of Computer Science and Engineering, Vellore Institute of Technology, Vellore –
632014, Tamil Nadu, India
2Jyotsna Malla, School of Computer Science and Engineering, Vellore Institute of Technology, Vellore – 632014,
Tamil Nadu, India,
3Preetham Lekkala, School of Computer Science and Engineering, Vellore Institute of Technology, Vellore –
632014, Tamil Nadu, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Rice stands as a favored and extensively
consumed cereal grain in Asian countries, while also enjoying
global accessibility. Within the rice market, the overarching
determinant of milled rice lies in its quality, an attribute that
assumes heightened significance in the context of import and
export trade. Rice samples often harbor assorted extraneous
elements such as paddy, chaff, damaged grains, weed seeds,
and stones. The principal objectiveoftheproposedapproachis
to introduce an alternative avenue for quality control and
analysis, characterized by reduced expenditure in terms of
effort, cost, and time. Image processing emerges as a pivotal
and technologically advanced sphere marked by significant
advancements. Imageprocessingmaneuversimagestoexecute
targeted operations, thereby refining and enhancing the
desired outcome. Moreover, this technique enables the
extraction of valuable insights from input images. This study
strives to develop image processing algorithms with a specific
focus on segmenting and identifyingrice grains. Byharnessing
image processing algorithms, it becomes possibletoefficiently
analyze the quality of grains based on their size. This paper
furnishes a solution for the classification and assessment of
rice grains, predicated on their dimensions and morphology,
through the application of imageprocessingtechniques. While
prior research has focused on the morphological attributes of
grains, encompassing parameters such as area and shape,
these endeavors often struggle to yield a generalized formula
capable of classifying diverse rice varieties due to the
considerable variance in shapes and sizes. In a distinctive
departure, this paper augments the analysis by incorporating
Fourier features extracted from grain images, thus
augmenting the accuracy of classification outcomes.
Key Words: —agriculture, imageprocessing,morphological
operations, edge detection, quality analysis, object
classification , deep learning, food quality detection
1.INTRODUCTION
The agricultural industry, spanning across centuries,
remains expansive and steeped in tradition.Thechallenge of
assessing grain quality has persisted throughout history.
This project introduces a pioneering solution for the
evaluation and grading of rice grains by harnessing image
processingtechniques.Traditionally,thecommercial grading
of rice hinges on grain size classification, categorizinggrains
as full, half, or broken. The assessment of food grain quality
has conventionally relied on human inspectors employing
visual scrutiny. However, the decision-making abilities of
human inspectors are susceptibleto external influencessuch
as fatigue, subjectivity, and personal biases.
The integration of image processing techniques offers a
transformative approach, eliminating the aforementioned
challenges while remaining non-destructive and cost-
effective. This methodology transcends human limitations,
enhancing objectivity and accuracy. The subsequent
discussion outlines the procedure deployed to ascertain the
percentage quality of rice grains. Rice quality, in essence,isa
composite of both physical and chemical attributes.Physical
characteristics encompass grain size, shape, chalkiness,
whiteness, milling degree, bulk density, and moisture
content. On the other hand, chemical attributes involve
gelatinization temperature and gel consistency,contributing
to the comprehensive assessment of rice quality.
This study centers on the development of image processing
algorithms aimed at effectively segmenting and identifying
rice grains. The utilization of image processing algorithms
proves to be a highly efficient approach for gauging grain
quality based on its size. The paper introduces a
comprehensive solution for grading and assessing rice
grains, focusing on grain size and shape through the
application of image processing techniques. Particularly, an
edge detection algorithm is employed to discern the
boundaries of each grain, employing a technique that
identifies the endpoints of individual grains.Subsequently,a
caliper is utilized to ascertain the length and breadth of rice
grains. This methodology stands out for its minimal time
requirement and cost-effectiveness.
In contrast, conventional methods employed for measuring
grain shape and size, such as the grain shape tester, dial
micrometer, and graphical method, tend to be protracted

and cumbersome. These methods typically allow for the
measurement of the dimensions of one grain at a time,
yielding results that are not only time-consuming but also
susceptible to human errors. Consequently, there is a
pressing need for greater precision to fulfill customer
expectations and overcome the limitations posedbymanual
procedures.
Numerous studies have previously delved into the analysis
of morphological characteristics of grains, encompassing
factors like area and shape. However, the vast diversity in
shapes and sizes across different rice varietiesprecludesthe
generalization of a uniform formula for classifying all rice
types. Addressing this challenge, this paper introduces an
additional dimension by extracting Fourier features from
grain images, complementing the spatial features and
culminating in an elevated level of accuracyforclassification
purposes.
This paper aims to employ image processing algorithms to
analyze grain quality based on size has become a prevalent
and effective methodology. This approach facilitates the
assessment and classification of rice grain quality by
leveraging advanced image processing techniques. By
focusing on the dimensions of rice grains, these algorithms
contribute to a comprehensive understanding of their
quality attributes. This technique holds the potential to
revolutionize the conventional methods of evaluating grain
quality, providing a more accurate and efficient means of
classification.
The remaining part of the paper is organized as follows.
Section 2 containstheLiteratureSurvey.TheProposedModel
is discussed in Section 3. Section 4 contains the Experiments
and Results. Lastly, the Conclusion and Future Directions is
presented in Section 5.
2. LITERATURE REVIEW
Food quality detection is a crucial aspect of the food
industry, ensuring consumer safety and satisfaction. Recent
advancements in machine learning and image processing
techniques have revolutionized the accuracy and efficiency
of food quality assessment. This literature review aims to
provide an in-depth analysis of 15 research papers that
explore the integration of machine learning and image
processing in food quality detection.
The authors in [1] propose a model to y showcases the
application of deep learning techniques, particularly
convolutional neural networks (CNNs), for food quality
assessment. The authors use image analysis to detect
defects, such as mold and discoloration, in food products.
The paper[2] focuses on fruit ripeness detection using
machine learning algorithms. The authors employ support
vector machines (SVM) and random forests to classify fruits
into different ripeness categories based on colorandtexture
features. The study[3] presents an automated system for
inspecting bakery products' quality. Image processing
techniques are combined with support vector machines for
real-time detection of defects and anomaliesinbakedgoods.
The research in[4] focuses on fish quality assessment using
image analysis and machine learning. The authors in [5] use
features like color, texture, and shape to classify fish into
different quality categories, ensuring freshness and safety.
This paper introduces texture analysis and neural networks
for meat quality detection. Texture features extracted from
meat images are fed into neural networks to classify meat
products based on tenderness and freshness.
This study [6] explores the use of CNNs for detecting
diseases and assessing quality in vegetables. The authors
develop a model that can identify diseases and quantify the
extent of damage using leaf images. The paper[7] discusses
the application of transferlearningandCNNsforfoodquality
inspection. The authors pre-train a CNN on a large dataset
and fine-tune it for specific food quality assessment tasks.
The authors in [8] focus on contaminant detection in food
products using deep learning techniques.Theauthorstraina
CNN to identify foreign objects and contaminants, ensuring
food safety. This study [9] presents a non-invasiveapproach
to inspect egg quality using machine learning. The authors
use image analysis and machine learning algorithms to
assess egg freshness and defects.
The paper[10] introducesanautomatedsystemfordetecting
milk spoilage using image processing and neural networks.
The authors[11] employ texture and color features to
classify spoiled and fresh milk samples. This research
focuses on classifying food items based on image features
using decision trees. The authors extract color, texture, and
shape features to develop a decision tree-based classifier.
The study[12]presentsa multi-classfoodqualityassessment
using deep learning and ensemble methods. The authors in
[13] combine the predictions of multiple models to enhance
the accuracy of quality assessment. This paper introduces a
hybrid CNN-SVM model for quality inspection of fruits. The
authors utilize CNN for feature extraction and SVM for
classification, achieving improved accuracy in fruit quality
assessment. This research[14] employs image processing
and random forests for dairy product quality detection. The
authors use image features to train a random forest model
that identifies defects and anomalies in dairy products. This
study focuses on automated detection of freshness in
seafood using deep learning techniques. The authors use a
deep neural network to assess seafood quality based on
color, texture, and shape attributes [15].
In conclusion, the reviewed papers collectively highlightthe
significant advancements achieved in food quality detection
through the integration of machine learning and image
processing techniques. From deep learning-based
approaches to hybrid models, these studies showcase the
potential of technology to enhance food safety, quality, and
consumer satisfaction in the food industry.

3. PROPOSED MODEL
Utilizing an image processing technique, the assessment of
rice seed quantities is undertaken, followed by their
classification based on parameters like length, breadth, and
the length-breadth ratio. Specifically, the length represents
the average longitudinal dimension of rice grains, while
breadth pertains to the average width. The length-breadth
ratio is computed as L/B = [(Average length of rice
grain)/(average breadth of rice)]*10.
The process is delineated through a seriesof methodological
steps:
a. Image Pre-Processing:
The initial phase involves image pre-processing, during
which a filter is applied to eliminate noise generated during
image acquisition. Thisfilter simultaneouslyenhancesimage
sharpness. The application of a threshold algorithm aids in
segmenting the rice grains from a black background.
b. Shrinkage Morphological Operation:
Subsequently, a shrinkage morphological operation is
employed to address the challenge of segmenting touching
rice kernels. The process commences with erosion, which
effectively separates interconnected features of rice grains
without compromising the integrity of individual ones.
Dilation follows erosion, with the primary objective beingto
restore eroded features to their original shape without
rejoining previously separated elements.
c. Edge Detection:
Edge detection, the third step, plays a pivotal role in
identifying the boundaries of rice grains. The canny
algorithm is adopted for its efficiency in detecting edges.
d. Object Measurement:
The fourth stage encompasses object measurement,
ascertaining the count of rice grains. Following grain
quantification, edge detection algorithms are applied,
subsequently yielding endpoint values for each grain. The
utilization of a caliper facilitatestheconnectionofendpoints,
enabling the measurement of both length and breadth. With
these dimensions determined, the length-breadth ratio is
calculated.
e. Object Classification:
In the final step of the algorithm, object classification is
executed. This necessitates a compilation of standard,
measured, and calculated outcomes. Reference data for rice
size and shape measurement is sourced from the laboratory
manual on rice grain quality, specifically the Directorate of
Rice Research located in Rajendra Nagar, Hyderabad.
In conclusion, the systematic applicationofimage processing
techniques, encompassing pre-processing, morphological
operations, edge detection, measurement,andclassification,
forms a comprehensive methodology for accurately
quantifying and categorizing rice seeds based on their size
and shape attributes.
Fig -1: Architecture Diagram of the proposed model
Fig -2: Classification Criteria based on Aspect Ratio
4. EXPERIMENTS AND RESULTS
The primary goal of this project istominimizethemanual
labor involved in the classification of rice grains and the
assessment of their quality. To accomplish this, the project
leverages the power of Machine Learning and utilizes the
Python Flask framework. The culmination of the project
results in the creation of a functional website. This web
application iscapable of receiving inputintheformofimages
depicting rice grains. It then employs Machine Learning
techniques to effectively categorize the rice grains and
simultaneously evaluate their overall quality.

This approach showcases its efficiency not only by reducing
the reliance on physical labor but also by providing a cost-
effective solution. By amalgamating Machine Learning
algorithms with the Python Flask framework, the project
successfully streamlines the process of rice grain
classification and quality analysis. The end result is a user-
friendly websitethatsignificantlyimprovestheefficiencyand
affordability of these critical tasks.
A grouped bar chart is employed in this context to
facilitate the classification process. The chart effectively
presents information related to the classification of rice
grains.
Notably, the chart employs two distinct bars:
1. The blue bar is indicative of the total count of rice
grains within the dataset.
2. The red bar, on the other hand, conveys the average
aspect ratio of the rice grains.
Furthermore, a pie chart is harnessed for the purpose of
quality analysis. This chart serves to visually represent key
quality attributes within the sample.
The chart is characterized by the following sections:
1. The blue section of the piechartdenotestheproportion
of rice grains present within the given sample.
2. Contrarily, the red section of the pie chart conveys the
percentage of dust detected within the analyzed sample.
Both the grouped bar chart and the pie chart play integral
roles in conveying vital information regarding the
classification and quality assessment of rice grains,
respectively. Through visual representation, these graphical
elementsenhancethecomprehensibilityandinsightfulnessof
the data analysis process.
Fig -3: Rice Granules Input Image
Fig -3: Aspect Ratio vs. Rice Granules Type Classification
Fig -4: Percentage of Rice and Dust in the rice grains image
5. CONCLUSIONS
In this project, our focus lies in the comprehensive
classification of rice grain samples, coupled with a
meticulous analysis of their qualitybasedontheaspectratio.
Our approach distinctly differentiates itself from existing
works, as it not only identifies rice grains and quantifies
their numbers but also delves deeper to evaluate their
quality and allocate them to specific categories.
What sets our work apart is its unique capability to achieve
near-perfect accuracy in assessing the quality of grains
within a sample. This is of paramount importance,
particularly for scenarios involving the efficient grading of a
large volume of grains. Our methodology significantly
expedites this process, alleviating the substantial time and
human effort typically associated with manual analysis.
Our image analysis algorithms are applied to images
featuring rice grains arranged randomly in a single layer.
Addressing potential errors such as touching kernels, our
approach utilizes a shrinkage operation to effectively
separate interconnected portions. Edge detection is
subsequently employed to pinpoint boundary regions and
determine the endpoints of each individual grain.
Subsequently, using a caliper, we measure the length and

breadth of each grain. These measurementsfurtherallowfor
the calculation of the length-breadth ratio.
Our study is underpinned by the development of image
processing algorithms tailored to segment and identify rice
grains. The deployment of these algorithms proves highly
efficient in evaluating grain quality based on their size. The
paramount advantage of our proposed method is its
expedited process, minimal time requirement, cost-
effectiveness, and superior performance compared to
traditional manual methods. All proposed steps have been
meticulously executed, culminating in the successful
classification and sizing of grains, which are then
appropriately categorized according to a predefined table.
REFERENCES
[1] M. B. Lam, T. -H. Nguyen and W. -Y. Chung, "Deep
Learning-Based Food Quality Estimation Using Radio
Frequency-Powered Sensor Mote," inIEEEAccess,vol.8,
pp. 88360-88371, 2020, doi:
10.1109/ACCESS.2020.2993053.
[2] S. Gayathri, T. U. Ujwala, C. V. Vinusha, N. R. Pauline and
D. B. Tharunika, "Detection of Papaya Ripeness Using
Deep Learning Approach," 2021 Third International
Conference on Inventive Research in Computing
Applications (ICIRCA), Coimbatore, India, 2021, pp.
1755-1758, doi: 10.1109/ICIRCA51532.2021.9544902.
[3] Biglari, Roya & Mohanna, Farahnaz & Ahsani,
Mohammad. (2022). Introducing an Automatic Bread
Quality Assessment Algorithm using Image Processing
Techniques. European Journal of Electrical Engineering
and Computer Science. 6. 31-38.
10.24018/ejece.2022.6.6.470.
[4] D. R. Wijaya, N. F. Syarwan, M. A. Nugraha, D. Ananda, T.
Fahrudin and R. Handayani, "Seafood Quality Detection
Using Electronic Nose andMachineLearningAlgorithms
With Hyperparameter Optimization," in IEEE Access,
vol. 11, pp. 62484-62495, 2023, doi:
10.1109/ACCESS.2023.3286980.
[5] Shiranita, K. & Hayashi, K. & Otsubo, A. & Miyajima, T. &
Takiyama, R.. (2000). Determination of meat quality by
image processing and neural network techniques.2.989
- 992 vol.2. 10.1109/FUZZY.2000.839179.
[6] P. Khobragade, A. Shriwas, S. Shinde, A. Mane and A.
Padole, "PotatoLeafDisease DetectionUsingCNN,"2022
International Conference on Smart Generation
Computing, Communication and Networking (SMART
GENCON), Bangalore, India, 2022, pp. 1-5, doi:
10.1109/SMARTGENCON56628.2022.10083986.
[7] S. S. Alahmari and T. Salem, "Food State Recognition
Using Deep Learning," in IEEE Access, vol. 10, pp.
130048-130057, 2022, doi:
10.1109/ACCESS.2022.3228701.
[8] L. Urbinati, M. Ricci, G. Turvani, J. A. T. Vasquez, F.
Vipiana and M. R. Casu, "A Machine-Learning Based
Microwave Sensing Approach to Food Contaminant
Detection," 2020 IEEE International Symposium on
Circuits and Systems (ISCAS), Seville, Spain, 2020,pp.1-
5, doi: 10.1109/ISCAS45731.2020.9181293.
[9] L. Geng, H. Wang, Z. Xiao, F. Zhang, J. Wu and Y. Liu,
"Fully Convolutional Network With Gated Recurrent
Unit for Hatching Egg Activity Classification," in IEEE
Access, vol. 7, pp. 92378-92387, 2019, doi:
10.1109/ACCESS.2019.2925508.
[10] S. Karthika Shree, V. Vijayarajan, B. Sathya Bama and S.
Mohammed Mansoor Roomi, "Milk Quality Inspection
Using Hyperspectral Imaging," 2023 International
Conference on Signal Processing, Computation,
Electronics, Power and Telecommunication
(IConSCEPT), Karaikal, India, 2023, pp. 1-6, doi:
10.1109/IConSCEPT57958.2023.10170710.
[11] M. Chun, H. Jeong, H. Lee, T. Yoo and H. Jung,
"Development of Korean Food Image Classification
Model Using Public Food Image Dataset and Deep
Learning Methods," in IEEE Access, vol. 10, pp. 128732-
128741, 2022, doi: 10.1109/ACCESS.2022.3227796.
[12] F. S. Konstantakopoulos, E. I. Georga andD.I.Fotiadis,"A
Review of Image-based Food Recognition and Volume
Estimation Artificial Intelligence Systems," in IEEE
Reviews in Biomedical Engineering, doi:
10.1109/RBME.2023.3283149.
[13] M. Haggag, S. Abdelhay, A. Mecheter, S. Gowid, F.
Musharavati and S. Ghani, "An Intelligent Hybrid
Experimental-Based Deep Learning Algorithm for
Tomato-Sorting Controllers," in IEEE Access, vol. 7, pp.
106890-106898, 2019, doi:
10.1109/ACCESS.2019.2932730.
[14] Alvarado, Witman & Meneses-Claudio, Brian & Roman-
Gonzalez, Avid. (2019). Milk Purity Recognition
Software through Image Processing. International
Journal of AdvancedComputerScienceandApplications.
10. 10.14569/IJACSA.2019.0101162.
[15] K. Prema and J. Visumathi, "An Improved Non-
Destructive Shrimp Freshness Detection Method Based
on Hybrid CNN and SVM with GAN Augmentation,"2022
International Conference on Advances in Computing,
Communication and Applied Informatics (ACCAI),
Chennai, India, 2022, pp. 1-7, doi:
10.1109/ACCAI53970.2022.9752599.

Quality Analysis and Classification of Rice Grains using Image Processing Techniques

More Related Content

What's hot (12)

Similar to Quality Analysis and Classification of Rice Grains using Image Processing Techniques (20)

More from IRJET Journal (20)

Recently uploaded (20)

Quality Analysis and Classification of Rice Grains using Image Processing Techniques