International Journal of Computer Engineering and Applications, Volume XII, Special Issue, August 18, ISSN

Transcription

1 DETECTION AND CLASSIFICATION OF LEAF DISEASES IN PLANTS Kajal Kumari Verma 1, Annu Kumari 1, Manisha Lakra 1, Manish Singh 1, Sushanta Mahanty 2 [1] Student, [2] HOD of Electronics and Communication Engineering Department [1,2] RVS College of Engineering and Technology, Jamshedpur, India. kajalverma20166@gmail.com, annu24134@gmail.com, manishalakra53@gmail.com, manish5258singh@gmail.com, nitpsushant@gmail.com ABSTRACT- Automatic leaf disease detection is very much essential topic in research. Agricultural growth is directly interacted with nation economy. We know that almost 70% population of India rely upon agriculture. Disease on plants consequentially reduces the quality and quantity of the agricultural products. Detection and Monitoring the disease manually is not acceptable through naked eye observation. This is an old method and requires enormous amount of time and expertise person in that field, moreover it doesn't provides an effective result. This paper presents digital technique for the detection and classification of plant leave disease. To overwhelm the disadvantages of manual method digital image processing technique is being used, which is comparatively faster and provides precise results in the disease detection of plant. The proposed methodology follows some basic steps which involve Image acquisition, Image preprocessing, Image segmentation, Feature extraction and Classification. Index Terms- Image acquisition, Segmentation, Feature extraction, SVM based classification [01] INTRODUCTION NDIA is a cultivated country. Agriculture has played an important role in the development of human civilization than simply a means to feed ever growing population. Plants have become an important source of energy but there are several diseases that affect the plants that will economy, social and ecological losses. The old method for detection and identification of plant this is based on naked eye observation which is very slow and gives less accuracy. For continuous monitoring, a large team of experts is required, to find out plant disease. Due to which consulting experts cost high as well as time consuming too. Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 1

2 DETECTION AND CLASSIFICATION OF LEAF DISEASES IN PLANTS Research work develops the advance computing environment i.e., Automatic detection and classification of plant disease using image processing technique. The paper introduces a MATLAB based system that focuses the leaf diseased area and used image processing technique. Image processing technique is used for measuring the affected area of disease and for accurate detection and identification of plant disease. This starts with capturing the image i.e., healthy and unhealthy images are captured, then the images are applied for pre-processing for image enhancement. Captured images are segmented using k-means clustering method to form clusters and features are extracted before applying to k-mean and lastly SVM algorithm for training and classification. Finally the disease are recognized by the system. Most leaf disease are caused by fungi, bacteria and viruses. So in this paper section 1 gives an introduction and importance of plant disease detection. And then a brief literature review of leaf disease detection technique in section 2 followed by section 3 that describes methodology of the proposed MATLAB image processing. Section 4 provides the experimental results & lastly section 5 that concludes this paper. [02] LITERATURE REVIEW In paper [1] presents the technique to classify & identify the diseases by which plants are affected. a machine learning based recognition system proves to be useful as it saves efforts, money & time. In paper [1] detection of leaf disease has been used, the methods are (a) identifying the infected area or object upon k-means clustering (b) feature extracting the set of the infected area or object (c) detecting and classifying the type of disease using neural networks. In paper [3] enhanced image has high quality and clear image than the original image. color images have primary colors red, green & blue. The RGB color image is difficult to implement in the application because of their range i.e, 0 to 255. Hence they convert RGB to grey images then the histogram equalization distributes the intensities of image is applied on image to enhance the plant disease images. In paper [4] Tushar H. Jaware et al. proposed a novel and improved k-means clustering technique to solve low-level image segmentation. [5] In this paper Support vector machine are a set of related supervised learning method used for classification and regression. The detection accuracy is improved by SVM classifier. In paper [6] authors described techniques for the detection of bacterial leaf scorch infection in plant. In image segmentation, the k-means clustering algorithm is applied for separating background and foreground images. k-means is very simple and effective for the detection of infected area. In paper [7] authors described technique of Sugarcane leaf disease detection and diseases are: Brown Spot, Downy mildew, Sugarcane Mosaic, Downy Fungal, Red stripe and Red rot. Pre-processing involved conversion of RGB image to grayscale and unwanted parts are removed. Healthy area and potentially infected area are located by segmentation. Linear, Non linear and Multiclass SVM are applied for disease detection. In paper [8] authors introduced techniques in which after image acquisition, color transformation structure is created, colors values are converted to space values in image pre-processing also k-means clustering method for segmentation is applied. Unnecessary area in leaf is removed by masking of green pixel. In paper [9] Monica Jhuria et al. uses image processing for detection of disease and fruit grading. In paper [10] author described technique of pomegranate for disease detection & diseases are: alterneria, bacterial blight & anthracnose. Pre-processing involves image resizing, filtering and morphological operations. Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 2

3 [03] OBJECTIVE OF RESEARCH The main objective of this research are :- 1. To collect images of various leaf diseases. 2. To identify region of infection based on clustering algorithm. 3. To extract the features from the segmented image. 4. To develop classification strategies based on features extracted. The objective behind this research is to detect and classify the disease using the image processing technique. The image data set is first collected. The image processing algorithm consists of following techniques ; image acquisition, image pre-processing, image segmentation, feature extraction & classification. The proposed methodology is tested on five types of images: healthy leaf, bacterial blight, alternaria alternate, antharacnose and cerocospora spotted leaf. [04] THEORETICAL BACKGROUND A. The Basic Procedure First step is to capture the images of leaves having disease spots from a digital camera. These acquired images are in RGB format. Then the pre-processing technique is used followed by segmentation and other image processing techniques to extract the features which are essential for the further analysis. After that several techniques are used to classify the images according to the specific problem. Figure shows the basic procedure for the detection and classification algorithm. Fig.1.Framework of the proposed system B. Image acquisition Image Acquisition is the first step of image processing. In this process different type of plant leaf images like -both healthy and affected by various type of diseases, are captured using digital camera having a required resolution for better results. After then captured images are stored in the personal computer for further MATLAB operations. Initially this image is in RGB ( Red, Green,Blue )form. Then to create a Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 3

4 DETECTION AND CLASSIFICATION OF LEAF DISEASES IN PLANTS color transformation structure we have to apply the acquired image to a device- independent color space transformation. C. Image Pre-processing To remove the undesired distortions, noise and to improve the quality of the image data, different Preprocessing steps are used. It includes,image enhancement, color space transformation. Image smoothing, image clipping i.e. cropping of the leaf image to get the interested image region. Image enhancement is carried out for increasing the contrast level. The acquired leaf images in the image acquisition process, are in red, green,blue form are converted into the color space representation. We have used L*a*b color space transformation for better results. D. Image Segmentation Image Segmentation step is used to get the region of interest of the desired leaf. The meaning of image segmentation is to partitioning the object from foreground (binary 1) and background (binary 0 ) into same region and features. Segmentation can be done using different methods such as- K-means clustering, Otsu threshold, Watershed and converting RGB image into HIS. In K-means clustering segmentation process, a technique is used to partition n observation into K clusters. So we can say K is the number of cluster in the segmented image. In our task, we have taken three clusters ( k= 3) and clustering is done depending upon the colors which is present in the segmented image. The algorithm for k-means clustering - Pick center of cluster of k cluster,either randomly or based on some heuristic. Assign each pixel in the image to the cluster that minimizes the distances between the pixel and the cluster. Again compute the cluster centers by averaging all of the pixels in the cluster. Repeat step 2 and 3 until convergence is attained. Fig.2.(a), indicates the import image and Fig.2.( b, c, d ) shows that the corresponding first, second and third clusters respectively obtained by k- means clustering segmentation technique Fig.2.(a) Fig.2.(b) Fig.2.(c) Fig.2.(d) E. Feature extraction Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 4

5 Feature extraction plays an vital role for identification of an segmented leaf. Texture, morphology, edge and color etc. are the features which can be applied in plant leaf disease detection. Monica Jhuria et al, In paper [9], found that morphological result gives better result than other features. In our work we get different feature parameters of clustered image like- mean, standard deviation, energy, entropy, affected region in percentage smoothness etc. These parameters are very much helpful for identification of the disease type. F. Training and classification Support Vector Machine is based on maximizing the minimum distance from the separating hyper plane to the nearest example[11]. There are two basic steps for using classifier : Training & Classification. Training is the process in which contents are taken that belongs to the specified classes and to create a classifier on the basis on these known contents. Classification is the process to built a classifier with a training dataset and running it on the unknown set of contents to determine the class membership for the unknown contents. Training is an iterative process where we have to built the best classifier and classification is a one-time process to run on unknown contents of dataset. Classification algorithms are as follows: a. Linear SVM b. Non Linear SVM c. Multiclass SVM (a)linear SVM Maximum margin hyper plane and margins for an SVM trained with samples from two classes. Fig.3.Hyper planes in Linear SVM (b)non Linear SVM This allows the algorithm to fit the maximum-margin hyper plane in a transformed feature space. Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 5

6 DETECTION AND CLASSIFICATION OF LEAF DISEASES IN PLANTS Fig.4.Kernel machines used in Nonlinear SVM for comparison (c)multiclass SVM Multiclass SVM aims to assign labels to instances by using support vector machines, where the labels are drawn from a finite set of several elements. In machine learning, support vector machines are supervised learning models with associated Learning algorithms that analyze data and recognize patterns, used for classification and regression analysis. Database Database contained large amount of image samples files which are containing the disease image and nondisease images. According to Image feature extraction techniques, the features will be extracted and then that image file will be stored in the database [12]. [05] EXPERIMENTAL RESULT In this experiment, two main dataset were generated, namely: (a)training dataset, and (b)testing dataset. Each column had a unique number (1 to 5) which represented the class (i.e., the disease). 1st, represented Healthy leaf. 2nd, represented Bacterial Blight disease infected leaf. 3rd, represented Anthracnose disease infected leaf. 4th, represented Alternaria Alternata disease infected leaf. 5th, represented Cercospora disease infected leaf. Then, a software program was written in MATLAB that would take in.mat files representing the training and testing data, train the classifier using the train dataset, and then use the test dataset to perform the classification task on the test data. Consequently, a MATLAB routine would load all the data files (training and testing data files) and make modifications to the data. Twenty five images of leaf are used for the learning and leaf disease detection and classification using Multi SVM classifier. Table 1: Feature results for different classes Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 6

7 The above table shows the feature extraction & statistical analysis of the modified infected cluster image obtained in the segmentation process. Then after a Graphical User Interface(GUI) is built to carry out different image processing steps and classification of images into various types. Dataset folder is created consisting of various images of leafs under test. The collected images are classified into two groups: healthy and unhealthy leaf images. SVM classifier is used to carry out the various classification based on the feature extraction & statistical analysis of the images. The GUI represents the various image processing steps like; Image Acquisition, Image pre-processing, segmentation, Feature extraction and Classification. 14 features are calculated from the segmented image these are: Affected area, Mean, Standard_Deviation, entropy, RMS, Variance, Smoothness, Kurtosis, Skewness, IDM, Contrast, Correlation, Energy and Homogeneity. Based on these calculated features the images are classified into various diseases. The GUI for the healthy and unhealthy leaf images are shown below: Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 7

8 DETECTION AND CLASSIFICATION OF LEAF DISEASES IN PLANTS Fig.5.System shows leaf disease detection result (of unhealthy leaf) Fig.6.System shows leaf disease detection result (of healthy leaf) [06] CONCLUSION The accurate detection and classification of plant leaf disease plays an important role in the field of agriculture and this can be done using digital image processing. This paper gives effective and efficient plant disease detection and classification technique using MATLAB image processing. The methodology in this paper depends on k-means and Multi SVM techniques. K- means and SVM algorithm takes very less time and provides high accuracy for complete processing. Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 8

9 REFERENCES [1] Mrunalini R. Badnakhe and Prashant R. Deshmukh An Application of K-Means Clustering and Artificial Intelligence in Pattern Recognition for Crop Diseases, International Conference on Advancements in Information Technology 2011 IPCSIT vol.20 (2011) [2] H. Al-Hiary, S. Bani-Ahmad, M. Reyalat, M. Braik and Z. ALRahamneh, Fast and Accurate Detection and Classification of Plant Diseases, IJCA, 2011, 17(1), 31-38, IEEE [3] Wenjiang Huang, Qingsong Guan, Juhua Luo, Jingcheng Zhang, Jinling Zhao, Dong Liang, Linsheng Huang, and Dongyan Zhang, New Optimized Spectral Indices for Identifying and Monitoring Winter Wheat Diseases, IEEE journal of selected topics in applied earth observation and remote sensing,vol. 7, No. 6, June 2014 [4] Tushar H. Jaware, Ravindra D. Badgujar and Prashant G. Patil, Crop disease detection using image segmentation, National Conference on Advances in Communication and Computing, World Journal of Science and Technology, pp: , Dhule, Maharashtra, India, [5] S. Arivazhagan, R. Newlin Shebiah, S. Ananthi, S. Vishnu Varthini, Detection of unhealthy region of plant leaves and classification of plant leaf diseases using texture features, CIGR, 2013, 15(1), [6] Murali Krishnan and Dr.M.G.Sumithra A Novel Algorithm for Detecting Bacterial Leaf Scorch (BLS) of Shade Trees Using Image Processing IEEE 11th Malaysia International Conference on Communications, Kuala Lumpur, Malaysia pp /13, 26th -28th November [7] Prajakta Mitkal, Priyanka Pawar, Mira Nagane, Priyanka Bhosale, Mira Padwal and Priti Nagane Leaf Disease Detection and Prevention Using Image processing using Matlab International Journal of Recent Trends in Engineering & Research (IJRTER) Volume 02, Issue 02, [ISSN: ], February [8] Khot.S.T, Patil Supriya, Mule Gitanjali, Labade Vidya Pomegranate Disease Detection Using Image Processing Techniques International Journal of Advanced Research in Electrical, Electronics and Instrumentation Engineering Vol. 5, Issue 4, p-issn: , [9] Monica Jhuria, Ashwain Kumar, and Rushikesh Borse, Image Processing For Smart Farming: Detection Of Disease And Fruite Grading, Proceeding of the 2013 IEEE Second International Conference on Image Information Processing (ICIIP-2013) [10] S. P. Bingulac, On the compatibility of adaptive controllers (Published Conference Proceedings style), in Proc. 4th Annu. Allerton Conf. Circuits and Systems Theory, New York, 1994, pp [11] Sandesh Raut, Amit Fulsunge, Plant Disease Detection in Image Processing Using MATLAB,International Jouranal of Innovation Research in Science, Engineering and Technology,(An ISO 3297:2007 Certified Organisation) Vol.6,Issue 6, June 2017 [12] P.Revathi, Classification of Cotton leaf spot Disease Using Image Processing Edge Detection Techniques,Research Scholar, Computer Science Karpagam University Coimbatore-21, Tamil Nadu, India, 2012 IEEE. Kajal Kumari Verma, Annu Kumari, Manisha Lakra, Manish Singh, Sushanta Mahanty 9