Classification of Islamic Geometric Pattern Images Using Zernike Moments

In this paper, we consider the use of orthogonal moments for classify Islamic geometric pattern images. To implement this technique, we use shape based classification. Zernike moments have been utilized as shape image descriptor. In classification stage, two different classifiers namely K-nearest neighbor rule, feed forward neural network have been used, which they are traditional nonparametric statistical classifier. Set of different experiments on binary images of regular, translated, rotated and scaled Islamic geometric shape has been done and variety of results has been presented. The results showed that orthogonal moment invariants are qualified as features to classify Islamic geometric pattern. The best result was 96.03% correct recognition demonstrating Zernike moments and Nearest Neighbor are adequate for Islamic star pattern recognition.