Model Based Ceramic tile inspection using Discrete Wavelet Transform and Euclidean Distance

Samir Elmougy1, Ibrahim El-Henawy2, and Ahmed El-Azab3 1Dept. of Computer Science, College of Computer and Information Sciences, King Saud Univ., Riyadh 11543, Saudi Arabia 1Dept. of Computer Science, Faculty of Computer and Information Sciences, Zagazig University, Zagazig, Egypt 3Dept. of Computer Science, Misr for Engineering and Technology (MET) Academy, Mansoura, Egypt

Abstract— Visual inspection of industrial products is used to determine the control quality for these products. This paper deals with the problem of visual inspection of ceramic tiles industry using Wavelet Transform. The third level the coefficients of two dimensions Haar Discrete Wavelet Transform (HDWT) is used in this paper to process the images and feature extraction. The proposed algorithm consists of two main phases. The first phase is to compute the wavelet transform for an image free of defects which known as reference image, and the image to be inspected which known as test image. The second phase is used to decide whether the tested image is defected or not using the Euclidean distance similarity measure. The experimentation results of the proposed algorithm give 97% for correct detection of ceramic defects. Keywords- Visual inspection; DWT; Euclidean distance. I. INTRODUCTION

Visual inspection of industrial product is one of the main important phases in many industries. It is used to determine the quality for the control for some products such as wood [1], textile [2], leather [3], steel [4], Printed Circuit Board (PCB) [5] and ceramic tiles [6]. In realty, ceramic tiles industry has hazardous, high polluted and unhealthy environment [7]. The inspection in this industry was usually made by humans so it is important to use mechanical technology to instead of standard humans to keep them healthy. A large variety of fast and different algorithms for object detection and recognition has been studied during the last decade by the computer vision community [8]. These algorithms can be divided into main approaches such as statistical, structured, filter based and model based [9]. In this work, a model based approach using DWT and Euclidean distance is introduced.

The earliest based model for visual inspection of ceramic tiles was carried by image difference operation (pixel- by-pixel comparison like XOR logic operator) [5]. Although this model has a good recognition, its operation costs too much processing time, and requires much memory more over the alignment of the tested image which should be identical to the reference image. Figure (1) depicts the image difference operation on ceramic tile.

In this paper, the proposed model is based on discrete wavelet transform because wavelet transform usually lead to a better image modeling, a better image encoding (this is the reason why wavelet is used as one of the best compression methodologies), and a better texture modeling.

Figure 1. Image difference operation of ceramic tile.

XOR Reference Test Image Defects 252 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security,
Vol. 7, No. 2, 2010

The rest of the paper is organized as follows. Overview of Wavelet processing is given in Section 2 followed by the Continuous Wavelet Transform (CWT), 1-D Discrete Wavelet Transform and 2-D Discrete Wavelet Transform in Subsections 2.1, 2.2 and 2.3 respectively. The proposed algorithm and its results are shown in Section 3. LVQ neural network structure and its algorithm explained in Section 4. Finally, conclusion and future work are discussed in Section 5. II. WAVELET PROCESSING Because the frequency contents of signals are very important, transforms are usually used. The earliest well known transform is Fourier transform which is a mathematical technique for transforming our view of the signal from time domain to frequency domain. Fourier transform breaks down the signal constituents into sinusoids of different frequencies. However, Fourier transform comes with serious shortage that is the lost of time information which mean it is impossible to tell when a particular event take place [10]. This shortage vanishes with using wavelet transform. A shifted version of the original signal is called mother wavelet which it is a wave form effectively a limited duration and its average value is zero. The most well known wavelets are Haar. Figure (2) depicts some types of these wavelets [11].
A. Continuous Wavelet Transform The Continuous Wavelet Transform (CWT) given in Equation (1), where x(t) is the signal to be analyzed, and ψ(t) is the mother wavelet or the basis function which it must be integrated to zero as give

…(Full text truncated)…