Abstract: Traditional fracture detection methods are based on difference detection among traces, so the noise will be detected when the fracture was detected. If large-scale fracture detecting methods were adopted or noise suppression methods were used together with fracture detection methods, fracture edges would be dim though signal-to-noise ratio will be raised. In order to solve these problems, the method based on wavelet multi-scale product was introduced for fracture detection. This method made use of the product of adjacent wavelet coefficients, and concealed the causal relationship of fracture in the different scale wavelet coefficients, at the same time of noise was suppressed, this method can enhance the precision of fracture detection. Application in theoretic model and real data had proved the effect of this method.

Key words: fracture detection, wavelet domain, multiscale product, noise suppression