Image Segmentation of Color Image using Threshold Based Edge Detection Algorithm in MatLab Neha Yadav, M.Tech [1] Vikas Sindhu [2] UIET, MDU Rohtak Abstract: The basic feature of an image is Edge. Edges can be defined as boundary between two different intensity regions in an image. The process of identifying and locating sharp discontinuities in an image is called Edge detection. Edge detection process significantly reduces the amount of data and filters out useless information, while preserving the essential structural properties in an image.[1] Since computer vision involves the recognition and classification of objects in an image, edge detections is a vital tool. In this paper, the main aim is to study edge detection process based on different techniques, edge detection methods such as Sobel, Prewitt, Robert, Canny, Laplacian of Gaussian (LOG) are also used for segmenting. A new edge detection technique is proposed which detects the acute and precise edges that are not possible with the pre-existing techniques. This implemented edge detection technique can be improved by combining it with other types of filters namely STD, Weiner, Harmonic, Geometric filters to eliminate the noise from the image [2]. Keywords: Edge-detection, Image segmentation, Canny operator, Sobel, Prewitt, Robert, Canny, Laplacian of Gaussian (LOG) Introduction: For pc vision and image processing systems to convert an image, the detachment of the image into object and background is a basic stride. Segmentation separation the image into a set of disjoint regions that are visually not similar, uniform and meaningful with respect to few qualities or processed properties, for example grey level, intensity, texture or color to enable easy image analysis.[3] A large number of methods are present in the literature to segment images. This is a critical work because the result of an image segmentation algorithm can be feed as input to upper-level processing tasks. The most commonly method based on edge technique is used to perform image segmentation. The Edge detection algorithms technique commonly detects sharp transitions of intensity and/or color within an image. These sharp transitions are characteristic of object edges. When edges of an object are detected then other processing such as region detection, text finding, and object detection can take place. In using color images there are many benefits. Such as, the increment in the quantity of information can be used for more accuracy of object location, processing, and the probability of processing images which is more complicated. Image detection is generally used to locate objects and boundaries (lines, curves, etc.) in images.[4] In brief, image detection is the process in which a label is assigned to every pixel in an image such that pixels having same label share certain properties. Edge detection produces as a line drawing of an image, which shows the intensity changes.[5] Usually, the boundaries of objects cause to produce immediate changes in the image intensity. Image segmentation is done by using the various edge detection techniques for example Sobel, Prewitt, Robert, Canny, LOG. Some of the practical applications of image segmentation are: Medical Imaging 1. Locate tumors and other pathologies 2. Measure tissue volumes 3. Computer-guided surgery 4. Diagnosis 5. Treatment planning 6. Study of anatomical structure Locate objects in satellite images (roads, forests, etc.) Face recognition Fingerprint recognition 49 Neha Yadav, Vikas Sindhu
Traffic control systems Brake light detection Proposed Method The different types of edge detection methods used are Sobel, Prewitt, Roberts, Canny, LoG. Sobel Operator: Sobel Operator carry out 2-D spatial gradient evaluation on an image and so highlight the areas of high spatial frequency that relative to edges. The convolution mask of Sobel operator are indicated in Figure 1, which are used to obtain the gradient magnitude of the image from the original 1 2 1 0 0 0-1 -2-1 -2 0 2 Figure 1. Sobel Mask Prewitt Operator: The prewitt operator is a way to measure the magnitude and orientation of the edge. The convolution mask of prewitt operator is indicated in figure 2. 1 1 1 0 0 0-1 -1-1 Figure 2. Prewitt Mask Roberts Operator: It performs 2-D spatial gradient evaluation on an image. It emphasizes areas of high spatial frequency which often related to edges. The cross convolution mask is indicated in figure 3. -1 0 0-1 Figure 3. Roberts Mask Laplacian of Guassian (LoG) Operator: LoG is a second order derivative. The digital execution of the Laplacian function is done using the mask indicated in figure 4. 0 1 0-1 4-1 0-1 0 0 1-1 0 Figure 4. Laplacian of Guassian (LoG) Operator Canny Operator: To examine edges by without considering noise from the image before find edges of image the Canny is a very important method. Without disturbing the characteristics of the edges in the image Canny method is a better method afterwards it applying the tendency to examine the edges and the serious value for threshold. The algorithmic steps are following: 1) Intervolve the image f(r, c) with a Gaussian function to get even and regular surface image f^(r, c). f^(r,c)=f(r,c)*g(r,c,6) 2) By applying the first difference gradient operator to determine the edge strength then edge magnitude and direction are gotten as earlier. 50 Neha Yadav, Vikas Sindhu
3) Apply non-maximal to the gradient magnitude. 4) Then apply threshold to the non-maximal suppression image. International Journal of Innovations & Advancement in Computer Science Implementation and Results: The techniques were applied to image. Colored image (figure 5) was transformed into gray scale image (figure 6) and then segmentation and recognition methods were applied. Grayscale image (Figure 6) is acceded for segmentation and object recognition using Prewitt, Sobel, Canny, Roberts, LoG,. By using different edge operators techniques the segmented image and recognized image are as shown in table Figure 5. Original Image Figure 6: Gray scale image Table 1. Segmented and Recognized images using different operators/algorithms Operator/ Algorithm Segmented Image Recognized Image Sobel Prewitt 51 Neha Yadav, Vikas Sindhu
Roberts Laplacian of Gaussian (LoG) Canny Canny operator performed better than Sobel, Prewitt, Roberts and LoG Conclusions: The aim of this research is to present a review of various approaches for image segmentation based on edge detection techniques. By studying the different Edge detection techniques and the result shows that canny gives the best results. In this research an attempt is made to review the edge detection techniques which are based on irregularity intensity levels. By using MATLAB software the relative consequences of various edge detection techniques is carried out with an image. From the observation, the result of LoG and Canny edge detectors gives almost similar edge map. When compared to the results of other techniques Canny is superior one, since under different conditions different edge detections show different results. The new edge detection technique is proposed which detects the acute and accurate edges that are not possible with the 52 Neha Yadav, Vikas Sindhu
already available techniques. The proposed method with different Thresholds value for given input image is in range of 0 to 1 and it is observed that when the threshold value is 0.68 the acute edges can be recognized properly. Inspite so many edges detection techniques are present in the literature, since it is a difficult task to the researchers to detect the exact image without noise from the original image. REFERENCES 1) Y.Ramadevi, T.Sridevi, B.Poornima, B.Kalyani; Segmentation And Object Recognition using Edge Detection Techniques ; International Journal of Computer Science & Information Technology (IJCSIT), Vol 2, No 6, December 2010. 2) Sonam Saluja1, Aradhana Kumari Singh2, Sonu Agrawal3; A Study of Edge-Detection Methods ; International Journal of Advanced Research in Computer and Communication EngineeringVol. 2, Issue 1, January 2013. 3) Muthukrishnan.R1 and M.Radha2; Edge Detection Techniques For Image Segmentation ; International Journal of Computer Science & Information Technology (IJCSIT) Vol 3, No 6, Dec 2011. 4) Poonam Dhankhar1, Neha Sahu2; A Review and Research of Edge Detection Techniques for Image Segmentation ; IJCSMC, Vol. 2, Issue. 7, July 2013, pg.86 92. 5) Shubhashree S. Savant1, Ramesh Manza2; Color Image Edge Detection using Gradient Operator International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)Web Site: www.ijettcs.org Email: editor@ijettcs.org;volume 4, Issue 1, January-February 2015 ISSN 2278-6856. 6) Iasonas Kokkinos, and Petros Maragos (2009), Synergy between Object Recognition and image segmentation using Expectation and Maximization Algorithm., IEEE Trans. on Pattern Analysis and Machine Intelligence (PAMI), Vol. 31(8), pp. 1486-1501, 2009. 7) Wen-Xiong Kang, Qing-Qiang Yang, Run-Peng Liang (2009), The Comparative Research on Image Segmentation Algorithms, First International Workshop on Education Technology and Computer Science. 53 Neha Yadav, Vikas Sindhu