Object recognition using a neural network and invariant Zernike features

In this paper, a neural network (NN) based approach for translation, scale, and rotation invariant recognition of objects is presented. The utilized network is a Multi-Layer Perceptron (MLP) classifier with one hidden layer. The back-propagation learning is used for its training. The image is represented by rotation invariant features which are the magnitudes of the Zernike moments of the image. To achieve translation and scale invariancy, the image is first normalized with respect to these two parameters using its geometrical moments. Syed Ahsan Ishtiaque ( 200601096 ) Kumar Srijan ( 200602015 ) Problem Statement: To create a Neural Networks based multiclass object classifier which can do rotation, scale and translation invariant object recognition. Introduction: Pattern recognition is an essential part of any high level image analysis system. A flexible recognition system must be able to recognize an object regardless of its orientation, size, and location in the field of view. This requirement translates into rotation, scale, and translation invariance property for the extracted features. Translation and scale invariance is achieved by the normalizing the images with respect to their first two orders of geometrical moments. Then the image is represented by rotation invariant features which are the magnitudes of Zernike moments of the image. In this report, the performance of a Neural Network based on Multi Layer Perceptron model has been analyzed for the recognition of binary images of letters of English alphabet. Moments The general two-dimensional (2D) moment definition, using a moment weighting kernel φnm(x, y) (also known as the basis function), and an image intensity function f(x, y), is given by With different basis functions, φnm, different types of moments can be obtained. Geometrical Moments The geometric moments are basically projections of the image function onto the monomials, i.e., φnm(x, y) = x y, the (n+m)th order geometric moment, Mnm, is defined as The geometric moments are most widely used in image analysis and pattern recognition tasks. This is due essentially to their simplicity, the invariance and geometric meaning of the low order moment values The zeroth order moment, M00, represents the total mass of the image. The two first order moments, (M10, M01), provide the position of the center of mass. The second order moments, (M20, M11, M02), can be used to determine several useful image features such as the principal axes, the image ellipse and the radii of gyration. Zernike Moments The Zernike moments use the complex Zernike polynomials as the moment basis set. The 2D Zernike moments, Anm of order n with repetition m, are defined as Where, n: Positive integer or zero m: Positive and negative integers subject to constraint n |m| even, |m| n : length of vector from origin to (x,y) pixel : angle between vector p and x axis in counterclockwise direction = Radical polynomial defined as