You are on page 1of 4

Extraction of Infected area from leaf image using image segmentation

School of Technology, SRTM University, Sub-Centre, Latur (MS) India vikashumbe@gmail.com

International Journal of Computer Information Systems, Vol. 3, No. 6, 2011

Vikas Humbe

Computer Science Department, Deogiri College Aurangabad (MS) India Yoginibpatil@gmail.com

Yogini Patil and Yogita S. Patil

Abstract Plants remain one of most difficult kinds of object to model due to their complex geometry and wide variation in appearance. In this paper the extraction of infected area from leaf image using image segmentation has been examined. Here some special leaves are taken for experimental purpose e.g Spinach, groundnut (Arachis Hypogea), Dhatura Alba, Tictona grandis, Musdard (Brassica compestri), Agal Maramus. Some of the plants are food plants, some are wild plants and some are medicinal plants. Some leaf images are .bmp some are .jpg segmentation techniques used in research paper that are i. Canny edge detector method of segmentation, ii. Localized Region Based Active Contour Segmentation. iii. Marker-Controlled Watershed Segmentation. Canny edge detection is edge based segmentation method work on grayscale image, it marks the strong edges of infected area from leaf. Active contour segmentation is useful for image with non-homogeneous foregrounds and background, it works on color image. Finally on the basis of experimental result the marker controlled watershed gives promising result and avoid problem of over segmentation. Keywords- Canny edge detector, Localized Region Based Active Contour Segmentation, Watershed transform, Marker-Controlled Watershed Segmentation

II.

RELATED WORK

I.

INTRODUCTION

Image segmentation has been extensively investigated in the past decades, with the development of a large number of image segmentation methods, but it is not applied on specially one part of tree i.e. leaf. As leaf is one of the sensitive parts of tree as it is infected by many diseases like Cedar apple rust, rust infection through mosaic virus, Mottle, Spotted wilt etc. The goal of this paper is to implement the image segmentation methods on various leaf images and judge the performance. Infected part from that leaf images can be separated and it will easy to identify the disease and required pesticide. Among the above methods watershed is proved as one of the best method in previous works .Section-2 briefly reviews the related work on image segmentation evaluation and summarizes the contribution of this paper. Section-3 introduces the methodology involved in this paper. Section-4 briefly introduces the performance measurement details. Section-5 describes the Experiment results with analysis charts. conclusion is given in Section-6.

There have been a large number of literatures on the image segmentation evaluation developed in the past decades. Most of previous works are focused on developing better ways to measure the accuracy/error of the segmentation and do not require the ground truth image segmentation as the reference [5][6][7]. Segmentation methods are applied based on leaf properties like geometry, color gradients and edges of leaf images. Semiautomatic technique for modeling plants directly from images. This image-based approach has the distinct advantage that the resulting model inherits the realistic shape and complexity of a real plant. We designed our modeling system to be interactive, automating the process of shape recovery while relying on the user to provide simple hints on segmentation. Segmentation is performed in both image and 3D spaces, allowing the user to easily visualize its effect immediately. Using the segmented image and 3D data, the geometry of each leaf is then automatically recovered from the multiple views by fitting a deformable leaf model. Our system also allows the user to easily reconstruct branches in a similar manner. We show realistic reconstructions of a variety of plants, and demonstrate examples of plant editing [10]. Most of the prior image-segmentation evaluation methods, however, need a ground-truth segmentation of the considered image and the performance is measured by calculating the discrepancy between the considered segmentation and the ground-truth segmentation. Since the construction of the ground-truth segmentation for many real images is labor intensive and sometimes not well or uniquely defined, most of prior image segmentation methods are only tested on: (a) some special classes of images used in special applications where the ground-truth segmentations are uniquely defined, (b) synthetic images where ground-truth segmentation is also well defined. (c) a small set of real images. Different from these methods, the paper presents i) Canny edge detector method of segmentation, ii) Localized Region Based Active Contour Segmentation. iii) Watershed segmentation iv) Marker-Controlled Watershed Segmentation In this paper we present the results of an objective evaluation of above four popular segmentation techniques we examine their characteristics

December Issue

Page 27 of 72

ISSN 2229 5208

1. Correctness: the ability to produce segmentations which agree with human intuition. That is, neither segmentations which correctly identify structures in the image at neither too fine nor too coarse a level of detail. 2. Stability with respect to parameter choice: the ability to produce segmentations of consistent correctness for a range of parameter choices. 3. Stability with respect to image choice: the ability to produce segmentations of consistent correctness using the same parameter choice on a wide range of different images .If a segmentation scheme satisfies these three characteristics, then it will give useful and predictable results which can be reliably apply on thousand of variety plants. III. METHODOLOGY

We evaluate the following four image segmentation methods i) Canny edge detector method of segmentation ii) Localized Region Based Active Contour Segmentation iii) Watershed transforms iv) Marker-Controlled Watershed Segmentation 3.1 Canny edge detector method of segmentation Smoothing - It is inevitable that all leaf images taken from a camera will contain some amount of noise. To prevent that noise is mistaken for edges, noise must be reduced. Therefore the image is first smoothed by applying a Gaussian filter. Finding gradients - The Canny algorithm basically finds edges where the grayscale intensity of the image changes the most. These areas are found by determining gradients of the image. Gradients at each pixel in the smoothed image are determined by applying what is known as the Sobel-operator Non-maximum suppression - The purpose of this step is to convert the blurred edges in the image of the gradient magnitudes to sharp edges. Basically this is done by preserving all local maxima in the gradient image, and deleting everything else. Double thresholding - The edge-pixels remaining after the nonmaximum suppression step are (iii) marked with their strength pixel-by-pixel. Many of these will probably be true edges in the image, but some may be caused by noise or color variations for instance due to rough surfaces. The simplest way to discern between these would be to use a threshold, so that only edges stronger that a certain value would be preserved. The Canny edge detection algorithm uses double thresholding. Edge pixels stronger than the high threshold are marked as strong; edge pixels weaker than the low threshold are suppressed and edge pixels between the two thresholds are marked as weak. Edge tracking by hysteresis - Strong edges are interpreted as certain edges, and can immediately be included in the final edge image. Weak edges are included if and only if they are connected to strong edges 3.2 Localized Region Based Active Contour Segmentation This method works on color images. In this Method typical region-based active contour energies are localized in order to handle images with non-homogeneous foregrounds and backgrounds. Here following steps are perform in MATLAB Step1:-load the image Step2:-make image smaller for fast computation Step3:-Segmentation Initialization

International Journal of Computer Information Systems, Vol. 3, No. 6, 2011 Step4:-Final Segmented image after applying fun Localized_ Seg 3.3 Water segmentation - Here contrast of image is enhanced. First by applying top hat filtering uneven illumination when background is dark and then bottom hat filtering is applied .Together used to enhance contrast in image. then structuring elements is used using strel() function in MATLAB to interact with image with purpose that how the shape here disk shape is used is fits or misses in image wherever it is fixed that are traced and then watershed function is applied to highlight the areas. 3.4 Marker-Controlled Watershed Segmentation Separating touching objects in an image is one of the more difficult image processing operations. The watershed transform is often applied to this problem. The watershed transform finds "catchment basins" and "watershed ridge lines" in an image by treating it as a surface where light pixels are high and dark pixels are low. Segmentation using the watershed transforms works well if you can identify, or "mark," foreground objects and background locations. Marker-controlled watershed segmentation follows this basic procedure 1. Compute a segmentation function. This is an image whose dark regions are the objects you are trying to segment. 2. Compute foreground markers. These are connected blobs of pixels within each of the objects. 3. Compute background markers. These are pixels that are not part of any object. 4. Modify the segmentation function so that it only has minima at the foreground and background marker locations. 5. Compute the watershed transform of the modified segmentation function. This example highlights many different Image Processing Toolbox functions, Including fspecial, imfilter, watershed, label2rgb, imopen, imclose, imreconstruct, imcomplement, imregionalmax , bwareaopen , graythresh, and imimposemin. IV. PERFORMANCE MEASURE

There are many quality measures that are used like Mean, variance, directional contrast, mean square error, peak signal to noise ratio and Fourier spectrum etc. In this work we have used Mean, Standard deviation, which are more useful to measure the quality of the digital images. Mean (M): It measures the average intensity of given image
M

x
i 0

l 1

p ( xi )

Standard deviation () : It measures the average contrast of the image..


Where the

2 ( x)

xi is the random variable indicating intensity value,

is the histogram of the intensity levels in a region, l is the number of possible intensity levels.

p(x)

December Issue

Page 28 of 72

ISSN 2229 5208

In this research work Mean and Standard deviation of various leaf images are calculated for segmentation techniques like Canny edge detector method, Localized region based contour, marker control watershed segmentation and results are store in the tables with graphical representation. The comparison with the three region-based image segmentation methods to analyze their average performances. The images are taken from web sources store there by camera. After applying image segmentation method for 11 leaf images, the average performance for each method is observed.
Figure1: Canny edge detector method of segmentation

International Journal of Computer Information Systems, Vol. 3, No. 6, 2011

Here highest peak of mean shows average intensity and standard deviation shows contrast for particular image .this method gives average results for all images.
Figure2: Localized Region Based Active Contour Segmentation

VI. CONCLUSION

Figure3:Results of some intermediate steps for Marker-Controlled Watershed Segmentation

In this research work we have applied various segmentation techniques like segmentation using Canny edge detector method, localized region based active contour segmentation, Watershed Segmentation, Morphological Watershed segmentation. On the basis of experimental result the marker control watershed segmentation gives unwavering and promising results and as it consist flooding the topographic surface of leaf image, it avoid problem of over segmentation. In future infected areas segmented using above methods can be store in database. By finding the similarities between areas stored and extracted, one can suggest insecticides or pesticides so this research will helpful for Biologist. REFERENCES
[1] [2] [3] Gauch, J. and Pizer, S.: Multiresolution analysis of ridges and valleys in grey-scale images, IEEE Transactions on Pattern Analysis and Machine Intelligence, 15:6 (June 1993), pages: 635 - 646, 1993. Olsen, O. and Nielsen, M.: Multi-scale gradient magnitude watershed segmentation, Proc. of ICIAP 97, Florence, Italy, Lecture Notes in Computer Science, pages 613. Springer Verlag, September 1997. Dam, E., Johansen, P., Olsen, O. Thomsen,, A. Darvann, T. , Dobrzenieck, A., Hermann, N., Kitai, N., Kreiborg, S., Larsen, P., Nielsen, M.: "Interactive multi-scale segmentation in clinical use" in European Congress of Radiology 2000. Vincken, K., Koster, A. and Viergever, M.: Probabilistic multiscale image segmentation, IEEE Transactions on Pattern Analysis and Machine Intelligence, 19:2, pp. 109-120, 1997. Y. Zhang, A survey of evaluation methods for image segmentation. Pattern Recogn.29(8),1335- 1346 in the year 1996. N. R. Pal and S. K. Pal, A review on image segmentation techniques, Pattern Recogn. 26,1277-1294 in the year 1993. F. Ge. S. Wang, and T. Liu, Image segmentation evaluation from the perspective of salient object extraction, Proc. IEEE Conf.Comp. Vis. Patt. Recogn., I,1146-1153 (2006). Lidi Wang, Tao Yang and Youwen Tian,, Crop Disease Leaf Image Segmentation Method Based on Color Features, Computer And Computing Technologies In Agriculture, Volume I IFIP International Federation for Information Processing, 2008, Volume , 713-717, DOI: 10.1007/978-0-387-77251-6_78

TABLE I.

SEGMENTED IMAGES USING FOUR SEGMENTATION METHODS

MEAN AND STANDARD DEVIATION OF ORIGINAL AND

[4] [5] [6] [7] [8]

December Issue

Page 29 of 72

ISSN 2229 5208

[9] [10] [11] [12] [13]

Pedro F Felzenszwalb, Daniel P Huttenlocher ,Efficient Graph-Based Image Segmentation , International Journal of Computer Vision, Volume: 59 , Issue: 2 in the year 2004. Image-based plant modeling Long Quan, Ping Tan, Gang Zeng, Lu Yuan, Jingdong Wang, Sing Bing Kang in ACM Transactions on Graphics in the year 2006 A. Shaik , S. V. Kasmirraja and SS Ahamed MOVING TOWARD REGION-BASED IMAGE SEGMENTATION TECHNIQUES Shawn Lankton, Localizing Region-Based Active Contours IEEE Trans Image Process. 2008 November; 17(11): 20292039. doi: 10.1109/TIP.2008.2004611. Canny, John, "A Computational Approach to EdgeDetection," IEEE Transactions on Pattern Analysis and Machine Intelligence,Vol. PAMI8, No. 6, 1986, pp. 679-698.

[14] BESL, P., AND MCKAY, N. 1992. A method for registration of 3D shapes. IEEETransactions on Pattern Analysis and Machine Intelligence 14, 2, 239256. [15] BOYKOV, Y., VEKSLER, O., AND ZABIH, R. 2001. Fast approximate energy minimization via graph cuts. IEEE Transactions on Pattern Analysis and Machine Intel-ligence 23, 11, 12221239. [16] DE REFFYE, P., EDELIN, C., FRANCON, J., JAEGER, M., AND PUECH, C. 1988 Plant models faithful to botanical structure and development. SIGGRAPH, 151158. [17] M.C. Cooper. The tractability of segmentation and scene analysis. International Journal of Computer Vision, vol 30, no 1, pages 27-42, October 1998

International Journal of Computer Information Systems, Vol. 3, No. 6, 2011

December Issue

Page 30 of 72

ISSN 2229 5208

You might also like