SYNTACTIC PATTERN RECOGNITION FOR WAVELET CLUSTERING IN SEISMOGRAM

Abstract

In a seismogram there exist many kinds of wavelets. The reflected wavelet from the gas sand zone has different shape with other layers. We use shape structure of the wavelet in the analysis. Syntactic pattern recognition is applied to the clustering of wavelets in the seismogram. The extracted wavelets can be represented as the strings of symbols. We use Levenshtein distance to calculate the distance between two strings. Then we can construct the hierarchical clustering of the wavelets. Top down hierarchical clustering by recursive method is proposed. A new pseudo F-statistics (NPFS) is proposed to decide the optimal number of clusters. From the experimental results in seismogram the wavelets on the gas sand zone can be detected successfully. It can improve the seismic interpretation.