Plant Disease Image Segmentation applying the K-Means Algorithm, Firefly, and Particle Swarm Optimisation

Abstract

The process of dividing an image to several pieces with comparable features is known as segmentation of images. A difficult phase in the image segmentation process is to extract the information from the image. Clustering is used to segment photos of a similar type or to assess the relevance of data. In essence, clustering is an unsupervised learning process. The data elements that make up one cluster share the same kinds of characteristics. In other words, each cluster has a minimum difference between its points and a maximum difference from the data points of other clusters. The suggested approach clusters image pixels by combining Particle Swarm Optimization with K-Means. Since combined techniques produce the best results, K-means should be thanked. K-means combined with Particle Swarm Optimization performs better than firefly combined with K-means. The firefly algorithm is used to tackle optimization issues and has a variety of uses. The Firefly method has been utilized in numerous research and optimization fields. Firefly algorithm and Firefly have been effectively utilized to solve a variety of issues. The firefly algorithm needs to be altered or combined with other algorithms in order to be used to a wide range of problems. Due to their metaheuristic nature, contemporary algorithms are crucial for solving NP-Hard problems.

Introduction

I. INTRODUCTION

A. Introduction of Image Segmentation

Image segmentation is one of the essential phases in the study and interpretation of images. The process of breaking down a digital image into several subsets is known as image segmentation. The image is divided into several parts with constant properties due to varied textures, shapes, boundaries, and colours. Four key picture segmentation techniques include threshold segmentation, edge detection, region extraction, and clustering. The process of breaking down a digital image into several subsets is known as image segmentation. https://www.sic.ici.ro168 Romana Capor Hrosik, Eva Tuba, Edin Dolicanin, Raka Jovanovic, and Milan Tuba are divided into numerous discontinuous segments, each of which contains elements that are related to yet distinct from those found in the other segments depending on factors such as intensity level, textural features, frequency components, etc. When brain tumour identification is the focus of segmentation

is to clearly distinguish the anomaly, in this case the tumors, from the rest of the image. Different methods were put forth for image segmentation. The most popular techniques include thresholding, edge detection, histogram-based techniques, clustering-based techniques, and others. In this article, we suggest a method for medical image segmentation based on the clustering algorithm. Romana Captor Hrosik, Eva Tuba, Edin Dolicanin, Raka Jovanovic, and Milan Tuba into numerous discontinuous parts, with elements in one segment being similar to, but different from, elements in other segments based on various criteria, such as intensity level, textural aspects, frequency components, etc. When brain tumor detection is taken into consideration, segmentation's purpose is to distinguish the anomaly—in this case, the tumor—clearly from the surrounding area of the image.

 Thresholding, histogram-based approaches, clustering-based techniques, region-growing methods, edge detection, etc. are some of the most used methods. In this study, we offer a clustering-based approach for segmenting medical images. Different methods for image segmentation have been suggested. Thresholding, histogram-based approaches, clustering-based methods, region-growing methods, edge detection, etc. are some of the most used techniques.

Using segmentation based on the clustering technique, we offer a method in this study for identifying illnesses in plant images. A method of unsupervised classification is clustering. Data handling, thickness, centroids, and other factors can all be used to group data into clusters. In this research the K-means algorithm among other clustering techniques.

However, determining the initial centroids for the k-mean is a significant issue K-means can be used to create centroids for the original cluster, but it becomes trapped in local optimum [9]. It is challenging to match the time restrictions and complexity of the standard approach when optimizing centroids' effectiveness through picture analysis. Many academics plan to apply swarm intelligence to improve the K-means algorithm. The cuckoo algorithm, artificial bee colony optimization, and differential algorithm are currently used successfully optimization algorithms for image processing. Swarm intelligence-based algorithms are developed from applications of the firefly method and are used to resolve optimizations issues. When used in many optimized and research fields, the firefly algorithm produces very precise and effective results. Using an algorithm, the flashing pattern and behavior of fireflies was created. [3]. The goal of this study is to offer a clustering k-means picture segmentation algorithm that is merged with firefly and particle swarm optimization [7].

B. Firefly Algorithm

Swarm intelligence optimisation techniques are now primarily used in information technology applications. In many fields, optimisation techniques are employed to enhance specific functions and their results. After being used in numerous applications, the Firefly algorithm produces findings that are backed up and precise. The Firefly algorithm has to be changed because it cannot handle complex issues. Author Yang.X.S. was the one who first created the Firefly algorithm. On the basis of Firefly's flashing pattern behaviors, the Firefly algorithm was created.

The behaviors and flashing pattern of fireflies served as the foundation for the swarm intelligence programme known as the "firefly algorithm." [3].

Following are the assumptions of Firefly algorithm:

1. All fireflies are unisexual, which means they are attracted to other fireflies and are neither male nor female.
2. The primary key element is brightness. One firefly will be drawn to a brighter firefly if it is less luminous. Increases in distance and decreases in light intensity cause brightness to automatically dim.
3. A firefly will move at random if there are no other fireflies around.

Two crucial processes are included in the firefly algorithm: varying the light intensity and determining attractiveness. For the suggested method, we stipulate that the firefly's brightness, which is determined by an objective function, is utilised to calculate attractiveness.

The light intensity I(r) varies with distance ‘r’ monotonically and exponentially, is given by:

I = IOe–y                                                  (1)

Where I0 is defined as the initial light intensity and y is defined as the light absorption coefficient.

The firefly attractiveness þ is:

Q =Qe–yr2                                               (2)

Where the distance between the two fireflies is denoted by ‘r’ and þ define them attractiveness at r=0

C. Image Segmentation Using firefly and k-Means

Data clustering is a method of unsupervised learning. The K mean clustering algorithm separates the data into k clusters. Each data point with the shortest distance from the original centroids is assigned to the cluster. The data point is disassembled to a useful centroid, and the fresh centroid is calculated repeatedly to inform Confluence. Clusters formed from pixels of the same type. Initial centroids are a major factor in the K-means algorithm's performance and efficiency. Centroids are immediately generated now the method is running [5]. Due to this, the k mean approach is prone to local optimum trapping and is unable to deliver accurate and efficient clustering results. We employ a heuristic technique to look for global solutions to this issue.

A. Integration of K-means with Firefly Obtaining relevant findings and dynamic clustering are influenced by the following two key elements: It assists in determining the competent centroid position during initiation. It discovers universal solutions in addition to local ones. A universal solution can be attained by having a maximum intra-connection and a minimum inter-connection. Between K-means and K-means integrated with Firefly, the addition of the exploration function is the only distinction. The existing solution is continuously enhanced and optimized by this exploring feature. When a certain standard value is reached, the firefly searching agent finds a new answer and replaces the old one. K-means and universal approaches are absent from the search [10].

K-means has three steps:

1. Initialization
2. Cluster Assignment
3. Exploration and Evaluation

Figure 1 has shown working of my research following steps are to be considered in research

a. Step1: Start

b. Step2.upload data of plant images which is to be taken from plant village set

c. Step3: Extract GT as {d1n1, d2n2, d3n3 d4n4, d5n5} where where, d represents disease and n represents total no. of element Step4: For each cat in GT S= Perform Segmentation using K-means

d. Step5: Perform K-means hybrid with Particle swarm optimization

e. Step6: Perform K-means hybrid with firefly

Conclusion

This work introduces a novel hybrid picture segmentation technique called KFA (k-mean firefly algorithm). The suggested approach is a meta-heuristic that effectively segments images and achieves clustering convergence. The correlation coefficient for the numerous images in the various clusters was calculated, and it revealed that the image segmentation had an accuracy of 88%. The experimental finding demonstrates that this strategy yields superior outcomes and correlation coefficient. Numerous uses for this proposed technique exist, such as the detection of plant diseases, medical imaging, content-based picture retrieval, etc. This method can be improved upon to produce outcomes and applications that are more effective. Future modifications to this algorithm could include automatic cluster count calculation and clustering itself.

References

[1] Amit Yerpude Bhavana Vishwakarma, \"A Meta-Heuristic Approach for Image Segmentation Using Firefly Algorithm,\" International Journal of Computer Trends and Technology (IJCTT), Vol. 11, No. 2, Pp. 69-73, May 2014. [2] Anand, R., S. Veni, and J. Aravinth. ” An application of image processing techniques for detection of diseases on brinjal leaves using kmeans clustering method.” Recent Trends in Information Technology (ICRTIT), 2016 International Conference on. IEEE, 2016. [3] Annie George,”Anomaly Detection based on Machine Learning:Dimensionality Reduction using PCA and Classification using SVM.”International Journal of Compute Applications (0975-8887),Vol 47-No.21,June-2012. [4] H. Suhl, Eds. New York: Academic, 1963, pp. 271-350. [5] Hong Yao, Qingling Duan, Daoliang Li, etal. An improved K-means clustering algorithm for fish image segmentation. Mathematical and Computer Modelling. Vol.58, No. 3-4, 2013, P:784-792 [6] Jagadesh D.Pujari, Rajesh, Abdul munaf, “Image processing based detection of fungal diseases in plants”, International journal of Information and communication Technologies, (Elsevier), vol. 46, pp 1802-1808, March 2015.R. Nicole, “Title of paper with only first word capitalized,” J. Name Stand. Abbrev., in press. [7] Jayamala K. Patil, Raj Kumar, “Advances in Image Processing for Detection in Plant Diseases”, Journal of Advanced Bioinformatics Applications and Research ISSN 0976-2604, Vol 2, Issue 2, June2011, pp 135-141. I.S. Jacobs and C.P. Bean, “Fine particles, thin films and exchange anisotropy,” in Magnetism, vol. III, G.T. Rado and [8] Kennedy J., Eberhart R., Shi Y. (2001), Swarm intelligence, Academic Press. [9] M. Masroor Ahmed and Dzulkifli Bin Mohamad \"Segmentation of Brain MR Images for Tumor Extraction by Combining Kmeans Clustering and Perona-Malik Anisotropic Diffusion Model\" International Journal of Image Processing, Volume (2) Issue (l),20 10. [10] Mandar Sohani, Avinash Shrivas Vijay Jumb, \"Color Image Segmentation Using K- Means Clustering and Otsu’s Adaptive Thresholding,\" International Journal of Innovative Technology and Exploring Engineering (IJITEE), Vol. 3, No. 9, Pp. 72- 76, February 2014. [11] Mingchao Chiang,Chunwei Tsai, Chusing Yang, A time-efficient pattern reduction algorithm for k-means clustering. Information Sciences. Vol.181, No.4, 15 2011, P: 716-731 [12] Ming-Ni Wu, Chia-Chen Lin and Chin-ChenChangI3 \"Brain Tumor Detection Using Color-Based K-Means Clustering Segmentation“International Conferenceon Computer Graphics ,2009 [13] Noble, and I.N. Sneddon, “On certain integrals of Lipschitz-Hankel type involving products of Bessel functions,” Phil. Trans. Roy. Soc. London, vol. A247, pp. 529-551, April 1955. (references) [14] P. B. Padol, A. A. Yadav, \"SVM classifier based grape leaf disease detection\", Proc. Conf. Adv. Signal Process. (CASP), pp. 175-179, June2016, IEEE,2016. [15] Priyadarshini Patil,Nagaratma Yaligar,Meena S M, “Comparison of Performance of classifiers - SVM, RF and ANN in Potato Blight Disease Detection using Leaf Images .”2017 IEEE Conference on Computational Intelligence and Computing Research. [16] Processing\",International Conference on Innovations in Information,Embedded and Communication Systems (ICIIECS), IEEE, 2017. [17] Q Yao, Z Guan, Y Zhou, J Tang, Y Hu, B Yang, \"Application of support vector machine for detecting rice diseases using shape and color texture features\", 2009 International Conference on Engineering Computation IEEE, pp. 79-83, 2009.. Eason, [18] R. M. Prakash, \"Detection of Leaf Diseases and Classification using Digital Image [19] R.Jothi Chitra, K.Sakthidasan @ Sankaran & V.Nagarajan 2016, „Iterative Non Local Image Restoration using Interpolation of Up and Down Sampling?, in International Journal of Advanced Engineering Technology, Vol.07/Issue 02/April- June,2016/633-638 [20] Rajneet Kaur Gurjeet Kaur Seerha, \"Review on Recent Image Segmentation Techniques,\" International Journal on Computer Science and Engineering (IJCSE), Vol. 5, No. 2, Pp. 109-112, February 2013. [21] Usama Mokthar, Nashwa El Bendary , Aboul Ella Hassenian, E.Emary,Mahmoud A Mahmoud,Hesham Hefny and Mohamed F. Tolba “SVM-Based Detection of Tomato Leaves Diseases.”D.Filev et al. (eds) systems 2014,Advances in Intelligent Systems and Computing,Springer International Publishing Switzerland 2015. [22] Vijay Singh,A.K.Misra”Detection of plant leaf diseases using image segmentation and soft computing techniques”,2214-3173,2017 China Agricultral university [23] Viraj A Gulhane and Maheshkumar H Kolekar. Diagnosis of diseases on cotton leaves using principal component analysis classifier. In India Conference (INDICON), 2014 Annual IEEE, pages 1–5. IEEE, 2014 [24] Yang, X. S. (2009) ‘Firefly algorithms for multimodal optimization’, in: Stochastic Algorithms:Foundations and Appplications (Eds O. Watanabe and T. Zeugmann), SAGA 2009, Lecture Notes in Computer Science, 5792, Springer-Verlag, Berlin, pp. 169-178. [25] Yang, X. S., (2010) ‘Firefly Algorithm, Stochastic Test Functions and Design Optimisation’, Int. J. Bio-Inspired Computation, Vol. 2, No. 2, pp.78–84.

Copyright

Copyright © 2023 Mandeep Kaur. 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.