Tree classification and linking process in the seismogram
Two purposes of this study are to use decision- theoretic pattern recognition for detecting the physical anomalies of seismic patterns and to use linking technique for linking seismic reflection patterns. Tree classification is used in decision-theoretic pattern recognition to the two-dimensional seismogram for the detection of physical anomalies in bright spot pattern, pinch-out pattern, and flat spot pattern. Using tree classifier, we get the 1-D classification result. In 2-D processing technique, linking process can extract the seismic patterns from the seismogram. Two methods are used in the linking process. One is the corresponding technique of the Levenshtein distance, the other is the proposed pattern growing technique. In the Levenshtein distance technique, a string-to-string minimum distance calculation is used for waveform correlation with the neighboring traces. In the pattern growing technique, the branch and bound search algorithm is used to find the best way to correlate the seismic reflectors. After obtaining the 1-D tree classification result, some singularities may still exist in the seismogram. These singularities can be removed with 2-D linking processing technique.