Scale Invariant Texture Representation Using Galois Field for Image Classification.

This paper presents a Galois Field method to scale invariant representation of texture image. The method is based on addition of neighbours in Galois Field. Scale invariance is achieved by considering the neighbours at different levels. The texture is represented using the features extracted by transforming it into a Galois Field based addition. Then the normalized cumulative histogram (NCH) bin values are considered as textures. For scale invariance, features are extracted at different levels. Thus obtained features are used for scale invariant classification. The average classification accuracy of 80.77%, 91.74%, 98.52% and 74.08% is achieved for Mondial Marmi, Brodatz, Vectorial and Outex datasets at level 3. The features can be used for suitable applications.