Rollover to highlight the lizard; click for an enlargement

Research: Body patterns & camouflage

Many animals exhibit brilliant colour patterns - jacky lizards are not one of these species. With the exception of the dorso-lateral stripe, objective measurements of reflectance properties are unimpressive. However, they do exhibit interesting body patterns which may serve an important functional role in camouflage. I have recently started to examine these body patterns, by comparing the texture patterns of the lizards, with the substrate against which they rely use for camouflage. WAVELET analysis is proving useful for image analysis and compression in other areas of science and industry, and we are currently exploring its application to meet our goal.

The analysis resembles Fourier analysis, whereby a signal is expressed as the sum of a series of sines and cosines. However, a Fourier expansion has only frequency resolution and no time resolution. Recently, several solutions have been developed which are more or less able to represent a signal in the time and frequency domain at the same time. The idea behind these time-frequency joint representations is to cut the signal of interest into several parts and then analyze the parts separately. In wavelet analysis, a fully scalable window is shifted along the signal and for every position the spectrum is calculated. Then this process is repeated many times with a slightly longer window for every new cycle. In the end the result will be a collection of time-scale representations of the signal, all with different resolutions.   However, calculating coefficients at every possible scale (the continuous wavelet transform) generates a lot of data. However, it is possible to sample a reduced number of scales - based on a power of 2 - that is more efficient and just as accurate. This is the discrete wavelet transform (DWT). An efficient way to implement this is using filters. The DWT approach passes the signal through a filter bank of low- and high-pass filters that result in approximation and detail coefficients respectively. Downsampling (keeping every second value) ensures that the amount of data is restricted to the original size. The decomposition of approximation coefficients can be continued resulting in lower resolution components - multiple level decomposition.

Example DWT of lizard pattern and additional methodological considerations here

Home | Research

Last Update: December 8, 2006