This paper describes a methodology termed texscale for texture analysis. This hierarchical approach is based on the group method which aims to group different textures into super-classes and determine whether a texture belongs in a particular texture super-class in conjunction with a mask tuning scheme to characterize texture features. Unlike the traditional two-step classification operation involving feature extraction followed by classification rule construction, our aim has been to introduce the texture energy computed using texture 'tuned' masks to directly function as a classifier in a single stage. An evaluation study of texscale classification scheme has been taken via the confusion matrix, which examines the extent to which arbitrary texture samples drawn from the total set of sample textures in two separate studies can be correctly assigned to the classes (15 in the study). One involves 360 samples, the other involves 1440 samples.
|Journal||ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings|
|Publication status||Published - 1 Jan 1994|
ASJC Scopus subject areas
- Signal Processing
- Electrical and Electronic Engineering