Texture Classification Using Multiple Wavelet Analysis

Abstract

The successful use of t he wavelet t ransform in the field o f t exture analysis has been well documented in literature. Simple e nergy features, as well as first and second-order s tatistics of t he wavelet coefficients, have both b een u sed successfully for this purpose. It i s our conjecture that analysing the image with more than one wavelet will provide additional i nformation abou t t he texture, thus improving classification rates. Experimental evidence supports this theory, showing that f or s imple e nergy features error r ates are halved when multiple wavelets are employed.