Transform Features for Texture Classification and Discrimination in Large Image Databases

This paper proposes a method for classification and discrimination of textures based on the energies of image subbands. We show that even with this relatively simple feature set, effective texture discrimination can be achieved. In this paper, subbandenergy feature sets extracted from the following typical image decompositions are compared: wavelet subband, uniform subband, discrete cosine transform (DCT), and spatial partitioning. We report that over 90% correct classification was attained using the feature set in classifying the full Brodatz [3 J collection of 1 12 textures. Furthermore, the subband energy-based feature set can be readily applied to a system for indexing images by texture content in image databases, since the features can be extracted directly from spatial-frequency decomposed image data. In this paper, we also show that to construct a suitable space for discrimination, Fisher Discrimination Analysis [SI can be used to compact the original features into a set of uncorrelated linear discriminant functions. This procedure makes i t easier to perform texture-based searches in a database by reducing the dimensionality of the discriminant space. We also examine the effects of varying training class size, the number of training classes, the dimension of the discriminant space and number of energy measures used for classification. We hope that the excellent performance for texture discrimination of these simple energy-based features will allow images in a database to be efficiently and effectively indexed by contents of their textured regions.