Using Co-occurrence Features Extracted From Ripplet I Transform in Texture Classification

Using Co-occurrence Features Extracted From Ripplet I Transform in Texture Classification

Texture analysis plays an important role in image processing. Nowadays transform based methods such as wavelet or curvelet transform based methods are widely being used. Inthis paper textured images are classified using ripplet type-I transform. Ripplet I is a higher dimension expansion fromcurvelet transform which generalizes its parabolic scaling law. Using this transform two dimensional signals can be represented in different directions and scales. After applying ripplet transform on the textures, we try to classify them in three different ways. First, images are classified directly based onripplet coefficients. Then classification based on statistical features extracted from ripplet coefficients is done. In the thirdcase classification is done based on co-occurrence features extracted from ripplet coefficients. This is the first time cooccurrence features extracted from ripplet coefficients are being used in classification. Classification based on curvelet transform is also done for the purpose of comparison. Experimental results show better performance in the Co-occurrence method