Unsupervised clustering of texture features using SOM and Fourier transform

Texture analysis has a wide range of real-world applications. This paper presents a novel technique for texture feature extraction and compares its performance with a number of other existing techniques using a benchmark image database. The proposed feature extraction technique uses 2D-DFT transform and self-organizing map (SOM). A combination of 2D-DFT and SOM with optimal parameter settings produced very promising results. The results from large sets of experiments and detailed analysis are included in this paper.