Using discrete wavelet transform for texture classification by using energy

Abstract EN

In this paper suggests a new image classification scheme of texture images, using the discrete wavelet transformation (DWT).

A good classification relies on feature extraction stage.

Several samples (texture images) of several classes were analyzed using wavelet transform which is new and promising set of tools.

In this work, it was shown that wavelet transform was appropriate for texture feature extraction.

The performances of statistical properties of wavelet detail coefficients were compared in several classification experiments.

And it was shown that, when used together (Mean, and energy ) the energy and mean absolute of approximation coefficients were appropriate wavelet signatures for achieving a successful image classification with acceptable low error rates.