View of WHEAT DISEASE IDENTIFICATION SYSTEM USING IMAGE SEGMENTATION AND THRESHOLD FILTER

(1)

Vol.03, Issue 09, Conference (IC-RASEM) Special Issue 01, September 2018 Available Online: www.ajeee.co.in/index.php/AJEEE

1

WHEAT DISEASE IDENTIFICATION SYSTEM USING IMAGE SEGMENTATION AND THRESHOLD FILTER

Rashmi Ranjan¹, Dr. Mehajabeen Fatima²

1M. Tech. Scholar, Department of Electronics and Communication, SIRT, RGPV, Bhopal, India

2Head of Department, Department of Electronics and Communication, SIRT, RGPV, Bhopal, India

1[email protected], ²[email protected]

Abstract: - The rapid growth of digital imaging made image processing techniques widespread over various applications such as medical, industrial, engineering and real life problem. Image enhancement, segmentation, feature extraction, classification and identification play an essential role in image processing applications. Keeping this as motivation, this research work proposes wheat image segmentation and feature extraction techniques using computational intelligence and image processing techniques. The implemented scheme starts with image segmentation using morphological set to extract the zones of interest and then to enhance the edges surrounding it. Further, feature extraction using threshold filter to gray level co-occurrence matrix is presented. Further, rough set is used to engender all minimal reducts and rules. These rules then fed into a classifier to identify different zones of interest and to check whether these points contain decision class value as either cancer or not. In this work, we develop user-friendly wheat disease reference architecture to provide on-field disease detection and prediction using cloud analytics.

Keywords: - Image Processing, Wheat Disease, Image Segmentation, Morphological Operation

I. INTRODUCTION

Image processing is to improve its quality and to modify it through editing for further use in different applications through its enhancement, segmentation, feature extraction, classification etc. Noise is another factor of concern in images and is reduced through image enhancement which is a process of adjusting the brightness, changing the tone of the colour, and sharpening the image etc. [1].

Image segmentation partitions an input image into nonoverlapping, homogeneous and connected regions [2] such that

“union of any two spatially adjacent regions is not homogenous” [3]. A region is homogeneous if all pixels in that region

“satisfy homogeneity conditions defined per one or more pixel attributes, such as intensity, colour, texture, etc. and if a connected path between any two pixels exists within the region” [4].

Enhancement of image quality is an essential step for image perception and is achieved by noise removal and amplification of image contrast. Genetic Algorithms in particular and other evolutionary algorithms in general have been adopted to achieve faster processing times and better results for image enhancement [5]. Various image segmentation techniques are available which are based on either of these i.e.

object matching, colour basis, prior knowledge, wrapper approach, and many more. But, development of automatic segmentation technique is still an active area of research. Image Segmentation is to partition a digital image into multiple segments (sets of pixels, also known as super pixels) for locating objects and boundaries (lines, curves, etc.) in images and its goal is to simplify and/or change the representation of an image into meaningful representation, which is easier to analyze. Amiya Halder et al [6]

proposed an image segmentation technique, which works successfully for both “single and multiple feature data with spatial information”. The method used Fuzzy C-means clustering and GA to automatically segment an image into its constituent parts. “Segmentation assigns a label to every pixel in an image such that pixels with the same label share certain visual characteristics”, such as colour, intensity, or texture. Adjacent regions are significantly different with respect to the same characteristic.

According to the report by United Nations of Food and Agriculture Organization the population will get doubles in 2050. The increased production of the agriculture will support huge economic boost to the nation. In

(2)

2 agriculture rice is the major food crop consumed by majority of the people in the world, particularly in India 70% of the people taking the rice as their major food.

Our focus is going too based on the rice production agriculture system. Across the Globe, India ranks the second largest producer of rice after china.

The farmers face lot of problems in the crops due to the diseases. Major problems in the agriculture include water problem, climatic change, pests and diseases in the plants [2]. Due to pests and diseases along the crop losses up to 37% every year.

II. LITERATURE REVIEW

Priyadarsan Parida et al, [1], these techniques are based on Transition Region Segmentation and Morphological Processing. This technique is applied to various types of image. This method decomposes the colour image into its respective colour components red, green and blue. For each component, a global threshold and local variance is used to extract the transition regions. The transition region, thus obtained undergoes morphological operation to get object contours. The morphological filling operation is employed on object contours to extract object regions. The intersection of results for individual object regions provides the exact object regions. Finally, the objects are extracted using the object regions. The technique provides better parameters but it can be increased further.

Soma Dey et al. [2], text Extraction from different background is always a challenging problem in these emerging field of research area of image processing. In our project we have implemented upon Noise Removal segmentation, Curve-let Transformation features and Support Vector Machine classifier. In that we calculated on successful result purpose on still images and some real time images for different background location image on features extraction and classified techniques.

Previously, Different methodologies have been used for texture feature extraction.

The method which we proposed gives a better accuracy and efficiency towards extract texture feature analysis as with wavelet transform. Curve-let, Gradient Edge Operator on Support Vector Machine Classifier analysis and

performance measure is completely new in this paper.

Ramya et al. [3], image Denoising and Image Segmentation are the two major areas of the medical image processing. The main objective of this paper is to develop a robust segmentation algorithm in order to detect tumor in 2D MRI brain images. Here we use image denoising as the preprocessing step as noise plays an important role in case of accuracy of affected area of the image, especially in medical diagnostics. To denoise the image, fourth order partial differential equation is employed. A seeded region growing segmentation is used to detect the tumor in MRI brain image. Also skull removal procedure is employed using morphological operators to increase the accuracy of brain tumor detection. This method detects the tumor in the brain image efficiently and also tested for several brain tumor images.

D. Chudasama et al. [4], the trial comes about exhibit that the proposed technique is not just better than the previous BTC-based picture ordering plans, additionally the previous existing strategies in the writing identified with the substance based picture recovery. To accomplish higher recovery precision, another component can be included into the EDBTC ordering plan with the other shading spaces, for example, Y Cb Cr, Hue-SaturationIntensity, lab, and so forth. An expansion of the EDBTC picture recovery framework can be conveyed to list video by considering the video as a grouping of pictures. This system should consider the transient data of the video arrangement to meet the client necessity in the CBIR setting. Image segmentation is a process of decomposition of the images into a set of homogeneous nonoverlapping regions which union is the entire image. The technique provides better smoothness but the quality of image segmentation decrease.

Jiann-Shu Lee et al. [5], a relative examination between the proposed half and half technique and the BTC and WHT demonstrates that the proposed strategy performs superior to the BTC and WHT.

Since this strategy includes less number of basic calculations, the time taken by this calculation is likewise less when contrasted and BTC. To optimize the system performance, we treat the image segmentation as a multi-objective

(3)

3 optimization. The technique provides smoothness but the results are not objectively handled. I

III. METHODOLOGY

Binary images may contain countless defects. In some circumstances binary regions constructed by simple thresholding are buckled by noise and textures. Morphology is a vast extent of image processing operations that modifies the images based on shapes. It is considered to be one of the data processing methods useful in image processing. It has many applications like texture analysis, noise elimination, boundary extraction etc. Morphological image processing follows the goal of eliminating all these defects and maintaining structure of image.

Morphological operations are confident only on the associated ordering of pixel values, rather than their numerical values, so they are focused more on binary images, but it can also be applied to grayscale images such that their light transfer functions are unknown and thus their absolute pixel values are not taken into consideration. Morphological techniques verify the image with a small template called structuring element. This structuring element is applied to all possible locations of the input image and generates the same size output. In this technique the output image pixel values are based on similar pixels of input image with is neighbors. This operation produces a new binary image in which if test is successful it will have non-zero pixel value at that location in the input image. There are various structuring element like diamond shaped, square shaped, cross shaped etc. The base of the morphological operation is dilation, erosion, opening, closing expressed in logical AND, OR notation and described by set analysis. Among them in this paper only two operations are used dilation and erosion. Dilation adds pixels while erosion removes the pixels at boundaries of the objects. This removal or adding of pixels depends on the structuring element used for processing the image.

3.1 Dilation

Dilation is one of the basic operators in mathematical morphology. It is applied to binary image but can also be applied to grayscale image. Dilation causes the

objects to grow in size. The effect of this operation will gradually increase the boundaries of foreground pixels, thus areas grow in size and holes in that region become smaller [6]. Dilation takes two parts as data. First one is the input image to be dilated and second is the structuring element also known as kernel.

With the help of this structuring element only it determines how much the image is to be dilated. The mathematical definition of dilation can be as follows [1]: Suppose A be a set of input image coordinates and B be a set of structuring element coordinates and Bx is a translation of B so that its origin is at x. Thus dilation of A by B is set of all points of x such that intersection of Bx with A is not null. In terms of set operations dilation of A by B is defined as [7]:

Figure 1: Dilation image 3.2 Filling the region

Dilation operation makes the boundaries of the object thick so for segmenting the object the next step is to fill the holes. The flood fill operation is most commonly known to fill the holes in the given input image. For binary images, it basically changes the background pixels to foreground pixels until it reaches the object boundaries and for grayscale images it makes the intensity level same i.e. it makes the dark areas surrounded by lighter areas to same intensity levels [2]. In binary images and gray-scale images the boundaries of the objects need to be specified by connectivity. In binary images the starting point for filling can also be specified. If we specify holes as an argument then it is of no need to specify any starting points [2].

(4)

4 Figure 2: Flood fill image

3.3 Erosion

Erosion is also one of the basic operators in mathematical morphology. Erosion causes the objects to shrink or become thin in size. Erosion basically erodes away the boundaries of the foreground which results in areas of those pixels shrink in size and holes of those areas become larger [8]. So, after dilution and filling the holes of object in some images the boundaries get mixed up so to somewhat separate the boundaries erosion is applied so as to make the boundaries of the objects thinner for better output. Erosion like same dilation takes two parts as data.

First one is the input image to be eroded and second is the structuring element.

With the help of this structuring element only it determines how much the image is to be eroded. The mathematical definition of erosion can be as follows [1]: Suppose A be a set of input image coordinates and B be a set of structuring element coordinates and Bx is a translation of B so that its origin is at x. Thus dilation of A by B is set of all points of x such that Bx is a subset of A. In terms of set operations erosion of A by B is defined as [9]:

Figure 3: Erosion Image IV. PROPOSED METHODOLOGY

In Image processing section, initially the image is captured from the camera and

further the image is processed using k means clustering for segmenting the image. The processed image is then edge detected using three different edge detection techniques. The edge detection techniques used are sobel, prewitt and canny algorithm. The diseased sample banana leaf has been taken for the edge detection analysis. Amongst the three edges detection methods used, canny edge detection algorithm gives the better and reliable detection. Owing to its optimality to meet with the three criteria for edge detection and the simplicity of process for implementation, it became one of the most popular algorithms for edge detection method.

As discussed earlier, IoT and Image processing are combined together in agricultural field in order to increase product yield and to reduce the crop failure. We focused on plant failure due to environmental factors through IoT technology. IoT system includes sensors, Arduino and a camera that regularly captures the plant. The color, texture, shape and area of the leaf are the parameters also considered in this work.

After examine the conditions of the plants we go for image processing. The initial test is done by using MATLAB software. In addition to the environmental factors, the plant with a diseased leaf can also be identified using Image processing. Based on the output and constraints the pesticides will be sprayed for the crop/plant where the disease is identified.

If there is any change that corresponds to the deterioration in the plants growth, the farmer is immediately informed. Early diagnosis will thus help in taking the necessary actions to increase the produce and reduce failure of crops.

(5)

5 Figure 4: Block diagram

4.1 Algorithm

Step 1: Select a 3 x 3 matrix size according to the 2-D window size. Assume that the processing pixel is Pij, which lies at the center of window.

Step 2: If 0 < Pij< 255, then the processing pixel or Pij is uncorrupted and left unchanged.

Step 3: On the off chance that Pij = 0 or Pij = 255, then it is considered as tainted pixel and four cases are conceivable as given underneath.

Case 1: In the event that the chose window has all the pixel esteem as 0, then Pij is supplanted by the Salt clamor (i.e.

255).

Case 2: On the off chance that the chose window contains all the pixel esteem as255, then Pij is supplanted by the pepper commotion (i.e. 0).

Case 3: In the event that the chose window contains all the esteem as 0 and 255 both. At that point the handling pixel is supplanted by mean estimation of the window.

Case 4: On the off chance that the chose window contains not all the component 0 and 255. At that point dispose of 0 and 255 and locate the middle estimation of the rest of the component. Supplant Pij with middle esteem. Step 4: Rehash step 1 to 3 for the whole picture until the procedure is finished.

Figure 5: Flow Chart of Proposed Method The window of size 3x3 chooses for noise detection and noise removal. The window contains total 9 elements which are as follows: Z1, Z2, Z3, Z4, Z5, Z6, Z7, Z8, Z9.

Third step use threshold values for noise detection and final median value for noise removal. We can divide the complete process into no. of steps as follows:

Table 1: Filtering window of size 3x3

Step-1:-

First we select all columns of filtering window one by one and then we find three values i.e. Maximum, Minimum and Median in each column. The mathematical expression can be shown as follow: The minimum values of rows and columns are represented as

Min (cln1) = min {Z1, Z4, Z7}

Min (row1) = min {Z1, Z2, Z3}

The maximum value of rows and columns are represented as

Max (cln1) = max {Z1, Z4, Z7}

(6)

6 Max (row1) = max {Z1, Z2, Z3}

Max (row2) = max {Z4, Z5, Z6}

Max (row3) = max {Z7, Z8, Z9}

The median value of the rows and columns are represented as

Med (cln1) = med {Z1, Z4, Z7}

Med (row1) = med {Z1, Z2, Z3}

max_min = Max (Min (cln1), Min (cln2), Min (cln3) Min (row1), Min (row2), Min (row3))

min_max = Min (Max (cln1), Max (cln2), Max (cln3) Max (row1), Max (row2), Max (row3)} median_med = Med {Med (cln1), Med (cln2), Med (cln3) Med (row1), Med (row2), Med (row3))

Now these three values (max_min, min_max, median_med) will be furhter sorted and finally we get minimum threshold, maximum threshold and final median value as follows:

Thmax = max {min_max, median_med, max_min}

Thmin = min {min_max, median_med, max_min}

Final_med= med { min_max, median_med, max_min }

These two threshold values will be used for noise detection and final median will be used for noise removal.

Step-3:-

Now we will perform noise detection and noise removal operation using these three values i.e. Thmax, Thmin, and Med_new.

Here we are parallel calculating the threshold values and median value. So there is no need to perform noise detection and noise removal separately.

V. SIMULATION RESULT

The values of FPR and FNR also vary between 0 and 1. FPR and FNR indicate over-segmentation and under segmentation, respectively. High values of FPR and FNR correspond to serious over- segmentation and under segmentation, respectively. The judgment about over segmentation and under- segmentation

can be validated by FPR and FNR. In the case of over segmentation, a portion of background appears as foreground where as in case of under segmentation a portion of foreground is missed from the actual foreground.

Figure 6: Fungus Wheat Disease Image Table 1: Results of Wheat Fungus

Image

(7)

7 VI. APPLICATION

Visual inspection of image processing allows the user to see how the structure image affects the original image. Variable playback speeds allows the user to control the speed at which the structure image is processed through the image so a user can see how it affects the final image.

User defined structure image lets the user control what the 3x3 structure image looks like and allows users the ability to see how different structure images affect different images. User defined images lets the user define an image up to 16x16. By clicking on the different cells, a user can setup up an image to their specifications before processing.

VII.CONCLUSION

Wheat diseases are discussed in this paper. The wheat diseases are identification with the help of morphological operation and threshold filter. Three types of error are calculated i.e. misclassification (ME), false negative rate (FNR) and false positive rate (FPR).

Basic of these error identified wheat diseased is calculate.

The research is very important in terms to increase the production of agriculture system in India. It is noted that development of the system will helps the farmers to save their crop loss from the diseases. Finally conclude that system will detect the diseases in the earlier time and classify above diseases and give information to the farmers to save their crops.

REFERENCE

1. T. Baranwal, Nitika and P. K. Pateriya,

"Development of IoT based smart security and monitoring devices for agriculture,"

2016 6th International Conference - Cloud System and Big Data Engineering (Confluence), Noida, 2016, pp. 597-602.

2. Kapoor, S. I. Bhat, S. Shidnal and A.

Mehra, "Implementation of IoT (Internet of Things) and Image processing in smart agriculture," 2016 International Conference on Computation System and Information Technology for Sustainable Solutions (CSITSS), Bangalore, 2016, pp. 21-26.

3. Cambra, S. Sendra, J. Lloret and L. Garcia,

"An IoT serviceoriented system for agriculture monitoring," 2017 IEEE International Conference on Communications (ICC), Paris, 2017, pp. 1- 6.

4. S. R. Prathibha, A. Hongal and M. P.

Jyothi, "IOT Based Monitoring System in Smart Agriculture," 2017 International Conference on Recent Advances in Electronics and Communication Technology (ICRAECT), Bangalore, 2017, pp. 81-84.

5. S. Roy et al., "IoT, big data science &

analytics, cloud computing and mobile app based hybrid system for smart agriculture,"

2017 8th Annual Industrial Automation and Electromechanical Engineering Conference (IEMECON), Bangkok, 2017, pp. 303- 304.

6. M. S. Mekala and P. Viswanathan, "A novel technology for smart agriculture based on IoT with cloud computing," 2017 International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I- SMAC), Palladam, 2017, pp. 75-82.

7. Sjaak Wolfert, Lan Ge, Cor Verdouw, Marc- Jeroen Bogaardt, Big Data in Smart Farming – A review, In Agricultural Systems, Volume 153, 2017, Pages 69-80, ISSN 0308-521X.

8. D. Yan-e, "Design of Intelligent Agriculture Management Information System Based on IoT," 2011 Fourth International Conference on Intelligent Computation Technology and Automation, Shenzhen, Guangdong, 2011, pp. 1045-1049.

9. Nikesh Gondchawar1, Prof. Dr. R. S.

Kawitkar2, “IoT based Smart Agriculture”

International Journal of Advanced Research in Computer and Communication Engineering Vol. 5, Issue 6, June 2016.

10. Sachin D. Khirade, A. B.patil,” Plant disease detection Using image processing,”2015, International conference on computing communication control and automation, IEEE. J. Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed., vol. 2. Oxford: Clarendon, 1892, pp.68- 73.

11. Vijai singh, Varsha, A.K.Mishra”Detection of unhealthy region of plant leaves using image processing and genetic algorithm”, 205, ICACEA, India. K. Elissa, “Title of paper if known,” unpublished.

12. Monica Jhuria, Ashwani kumar and Rushikesh Borse, ”Image processing for Smart farming, detection of Disease and Fruit Grading,” proceeding of the 2013, IEEE, second international conference on image Information processing.