This paper proposes a checkerboard technique of wavelet transformation of i
mages, whose basis is formed by functions with a high degree of isotropy. I
f isotropic elements (spots) predominate on the image, a large number of sm
all coefficients can be expected when it is expanded by a checkerboard tech
nique, with the dot structure being created in the very process of wavelet
analysis when a reduced version of the image is constructed. It is shown th
at, in experiments on data compression, the proposed transformation techniq
ue made it possible at the last stages of the analysis to appreciably reduc
e the error associated with discarding the small expansion coefficients. (C
) 2001 The Optical Society of America.