International Journal of Electronics, Communication & Instrumentation Engineering Research and Development (IJECIERD) ISSN 2249-684X Vol.

3, Issue 1, Mar 2013, 71-76 TJPRC Pvt. Ltd.

A NOVEL METHOD FOR BIAS CORRECTION AND SEGMENTATION OF IMAGES

ABRAHAM MATHEW1 & R. RADHAKRISHNAN2
1 2

Research Scholar, Sathyabama University, Chennai, India

Principal, Mahindra Institute of Engineering & Technology, Salem, India

ABSTRACT
We propose a variation level set framework, for segmentation and bias correction of images with intensity in homogeneities. Based on a generally accepted model of images with intensity in homogeneities and a derived local intensity clustering property, we define an energy of the level set functions that represent a partition of the image domain and a bias field that accounts for the intensity in homogeneity. Segmentation and bias field estimation are therefore jointly performed by minimizing the proposed energy function. The slowly varying property of the bias field derived from the proposed energy is naturally ensured by the data term in our variation framework, without the need to impose an explicit smoothing term on the bias field. Our method is much more robust to initialization than the piecewise smooth model. Experimental results have demonstrated superior performance of our method in terms of accuracy, efficiency, and robustness.

KEYWORDS: Level Set, Homogeneities, Clustering, Estimation, Robust INTRODUCTION

Several general-purpose algorithms and techniques have been developed for image segmentation. Since there is no general solution to the image segmentation problem, these techniques often have to be combined with domain knowledge in order to effectively solve an image segmentation problem for a problem domain. Intensity in-homogeneity often occurs in real-world images, which presents a considerable challenge in image segmentation. The most widely used image segmentation algorithms are region-based and typically rely on the homogeneity of the image intensities in the regions of interest, which often fail to provide accurate segmentation results due to the intensity in-homogeneity. Intensity in-homogeneity often occurs in real-world images, which presents a considerable challenge in image segmentation. The most widely used image segmentation algorithms are region-based and typically rely on the homogeneity of the image intensities in the regions of interest, which often fail to provide accurate segmentation results due to the intensity inhomogeneity. We, proposes a novel region-based method for image segmentation, which is able to deal with intensity inhomogeneities in the segmentation. First, based on the model of images with intensity in-homogeneities, we derive a local intensity clustering property of the image intensities, and define a local clustering criterion function for the image intensities in a neighborhood of each point. This local clustering criterion function is then integrated with respect to the neighborhood center to give a global criterion of image segmentation. In a level set formulation, this criterion defines an energy in terms of the level set functions that represent a partition of the image domain and a bias field that accounts for the intensity in-homogeneity of the image. Therefore, by minimizing this energy, our method is able to simultaneously segment the image. Level Set Method has become a thriving technique in this field[1].We begin by reviewing some typical PDE based applications in image processing. In typical PDE methods, images are assumed to be continuous functions sampled on a grid[1,3]. Model-based vision[7] is firmly established as a robust approach to recognizing and locating known rigid objects in the presence of noise[5,7], clutter, and occlusion.

72

Abraham Mathew & R. Radhakrishnan

PROPOSED METHODOLOGY
Intensity in homogeneity often occurs in real-world images due to various factors, such as spatial variations in illumination and imperfections of imaging devices, which complicates many problems in image processing and computer vision. In particular, image segmentation may be considerably difficult for images with intensity in homogeneities due to the overlaps between the ranges of the intensities in the regions to segmented. This makes it impossible to identify these regions based on the pixel intensity. Those widely used image segmentation algorithms usually rely on intensity homogeneity, and therefore are not applicable to images with intensity inhomogeneities. In general, intensity inhomogeneity has been a challenging difficulty in image segmentation. The level set method, originally used as numerical technique for tracking interfaces and shapes, has been increasingly applied to image segmentation in the past decade. In the level set method, contours or surfaces are represented as the zero level set of a higher dimensional function, usually called a level set function. With the level set representation, the image segmentation problem can be formulated and solved in a principled way based on well-established mathematical theories, including calculus of variations and partial differential equations (PDE). An advantage of the level set method is that numerical computations involving curves and surfaces can be performed on a fixed Cartesian grid without having to parameterize these objects. Moreover, the level set method is able to represent contours/surfaces with complex topology and change their topology in a natural way. Previous image processing studies have used grey level co-occurrence matrices (GLCMs) within texture-based feature classi. cation algorithms. Researchers in medical imaging have used GLCMs to identify tumor. The Gray Level Co-occurrence Matrix1 (GLCM) and associated texture feature calculations are image analysis techniques. Given an image composed of pixels each with an intensity (a specific gray level), the GLCM is a tabulation of how often different combinations of gray levels co-occur in an image or image section. Texture feature calculations use the contents of the GLCM to give a measure of the variation in intensity (a.k.a. image texture) at the pixel of interest. From a generally accepted model of images with intensity in-homogeneities, we derive a local intensity clustering property, and therefore define a local clustering criterion function for the intensities in a neighborhood of each point. This local clustering criterion is integrated over the neighborhood center to define an energy functional, which is converted to a level set formulation. Minimization of this energy is achieved by an interleaved process of level set evolution and estimation of the bias field. As an important application, our method can be used for segmentation and bias correction . Let be the image domain, and I: R be a gray level image. In a segmentation of the image I is achieved by into disjoint regions 1.. N, and a piecewise smooth function

finding a contour , which separates the image domain

U that approximates the image and is smooth inside each region following Mumford-Shah functional

i. This can be formulated as a problem of minimizing the

where

is the length of the contour . In the right hand side of (1), the first term is the data term, which forces u to

be close to the image I, and the second term is the smoothing term, which forces u to be smooth within each of the regions separated by the contour . The third term is introduced to regularize the contour \C= . Let be the regions in

separated by the contour C, i.e.

Then, the contour C can be expressed as the union of the boundaries of the . Region-based image segmentation methods typically relies on a specific

regions, denoted by C1,C2CN , i.e.

A Novel Method for Bias Correction and Segmentation of Images

73

region descriptor (e.g. intensity mean or a Gaussian distribution) of the intensities in each region to be segmented. However, it is difficult to give such a region descriptor for images with intensity inhomogeneities. Moreover, intensity inhomogeneities often lead to overlap between the distributions of the intensities in the regions . Therefore, it is

impossible to segment these regions directly based on the pixel intensities. Nevertheless, the property of local intensities is simple, which can be effectively exploited in the formulation of our method for image segmentation with simultaneous estimation of the bias field.based on the image model in (3) and the assumptions A1and A2, we are able to derive a useful property of local intensities, which is referred to as a local intensity clustering property as described and justified below. To be specific, we consider a circular neighborhood with a radius The partition of the entire domain centered at each point y , defined by . . For a

induces a partition of the neighborhood

, i.e., forms a partition of are close to b(y), i.e..

slowly varying bias field b, the values b(x) for all x in the circular neighborhood b(x)=b(y) for x Thus, the intensities b(x)J(x) in each subregion b(x)J(x) = b(y) for x

are close to the constant b(y)

, i.e.

Then, in view of the image model in we have I(x) = b(y) + n(x) for

Where n(x) is additive zero-mean Gaussian noise. Therefore, the intensities in the set = {I(x) : x }

form a cluster with cluster center , which can be considered as samples drawn from a Gaussian distribution with mean . Obviously, the N clusters , are well-separated, with distinct cluster centers. This local

intensity clustering property is used to formulate the proposed method for image segmentation and bias field estimation.

SEED POINT
The seed points for growing the regions are identified by computing a score for every pixel: n(x) = varR*varG*varB where varR, varG and varB are variance of the R, G and B component of pixel respectively taken over a window of 20 pixels taking x as center (boundary pixels are taken care accordingly). Each pixel i is stored as a set, V = ( xi, yi, ni ) where xi and yi denotes the respective co-ordinates. Now, a set of all the pixels is stored in S, S = (V1, V2,..., Vn). The Region Segmentation Iteration The minimum element in S signifies that the region around this pixel is more homogeneous and can serve as a good seed point for the region growing algorithm. Following steps are taken then: Pick the minimum element in S. Create a region to hold the pixel. The seed area window Rj centered at the this current minimum. is initialized using the

74

Abraham Mathew & R. Radhakrishnan

The state information of the region growing algorithm, and L, are then initialized as given in equations mentioned later.

Compute Mean j and Covariance j of the region Rj.

Compute the likelihood which is defined as: j(x) = MD (f(x), N(j , j)) T where value of T can be defined manually in order to get best segmentation results and MD(f(x), N(j , j)) represents the Mahalanobis Distance of the image feature vector f(x) = [fr(x), fg(x), fb(x)] to the appearance of the region modeled as a single gaussian using the mean j and the covariance j of the features in the region. The function j(x) is the level set formulation with level defined by the value of T.

PROPOSED VARIATIONAL EM ALGORITHM

The expectation-maximization (EM) algorithm has been widely applied to the estimation of Gaussian mixture model(GMM) in brain MR image segmentation. However, the EM algorithm is deterministic and intrinsically prone to overfitting the training data and being trapped in local optima. In this paper, we propose a variation EM(GA-VEM) algorithm for MAMOGRAM image segmentation. In this approach, the VEMalgorithm is performed to estimate the GMM, and the GA is employed to initialize the hyperparameters of the conjugate prior distributions of GMM parameters involved in the VEM algorithm. SinceGA has the potential to achieve global optimization and VEM can steadily avoid overfitting, the hybrid GA-VEM algorithm is capable of overcoming the drawbacks of traditional EM-based methods. The VEM algorithm provides an automated preference for simple models that sufficiently explain the data without unnecessary complexity, and therefore has distinct advances over the EM algorithm in avoiding overfitting. However, both the EM and VEM algorithms are deterministic optimization techniques, which can achieve merely a local optimum with heavy dependence on initialization.

RESULTS

Figure 1: MRI Image Segmentation Using Level Set Method

A Novel Method for Bias Correction and Segmentation of Images

75

REFERENCES
1. G. Aubert and P. Kornprobst, Mathematical Problems in Image and the Calculus of Varia- tions. New York: Springer-Verlag, 2002. 2. V. Caselles, F. Catte, T. Coll, and F. Dibos, A geometric model for active contours in image processing, Numer. Math., vol. 66, no. 1, pp. 131, Dec. 1993. 3. V. Caselles, R. Kimmel, and G. Sapiro, Geodesic active contours,I Image. Process., vol. 10, no. 2, pp. 266277, Feb. 2001. 4. T. Chan and L. Vese, Active contours without edges, IEEE Trans.Image. Process., vol. 10, no. 2, pp. 266 277, Feb. 2001. 5. D. Cremers, A multiphase levelset framework for variational motion segmentation, in Proc. Scale Space Meth. Comput. Vis., Isle of Skye, U.K., Jun. 2003, pp. 599614.L. 6. 7. Evans, Partial Differential Equations. Providence, RI: Amer.Math. Soc., 1998. S. Kichenassamy, A. Kumar, P. Olver, A. Tannenbaum, and A. Yezzi, Gradient flows and geometric active contour models, in Proc. 5th Int. Conf. Comput. Vis., 1995, pp. 810815. 8. R. Kimmel, A. Amir, and A. Bruckstein, Finding shortest paths on surfaces using level set propagation, IEEE Trans. Pattern Anal. Mach. Intell., vol. 17, no. 6, pp. 635640, Jun. 1995. 9. C. Li, R. Huang, Z. Ding, C. Gatenby, D. Metaxas, and J. Gore, A variational level set approach to segmentation and bias correction of medical images with intensity inhomogeneity, in Proc. Med. Image Comput. Comput. Aided Intervention, 2008, vol. LNCS 5242, pp.10831091, Part II. 10. R. Malladi, J. A. Sethian, and B. C. Vemuri, Shape modeling with front propagation: A level set approach, IEEE Trans. Pattern Anal. Mach. Intell., vol. 17, no. 2, pp. 158175, Feb. 1995. 11. D. Mumford and J. Shah, Optimal approximations by piecewise smooth functions and associated variational problems, Commun.Pure Appl. Math., vol. 42, no. 5, pp. 577685, 1989. 12. S. Osher and J. Sethian, Fronts propagating with curvature-dependent speed: Algorithms based on HamiltonJacobi formulations, J. Comp.Phys., vol. 79, no. 1, pp. 1249, Nov. 1988. 13. N. Paragios and R. Deriche, Geodesic active contours and level sets for detection and tracking of moving objects, IEEE Trans. Pattern Anal.Mach. Intell., vol. 22, no. 3, pp. 266280, Mar. 2000. Processing: Partial Differential Equations

Abraham Mathew, received his B.E in Electronics engineering from Ban galore university, ME in Applied Electronics from Sathyabama University,Chennai.He is currently pursuing PhD from Sathyabama University. His research interest include Video & Image Processing, Signal Processing. He is having more than ten years of experience in teaching. He is a life member of IETE.

76

Abraham Mathew & R. Radhakrishnan

R.Radhakrishnan, received his B.E in Electronics and communication engineering from Bharathidasan university, ME in Applied Electronics from Bharathiar University,and PhD from Anna University ,Chennai.He is currently working as the Principal of Sri Nandhanam Engineering College, Vellore. He has published more than ten research papers in various international and national journals.