Surface Mapping Using Auto-adaptable Similarity Measures

In this expanded abstract, we introduce a new method that finds the global structure of seismic events and allows the automatic mapping of sub-surface data. Seismic reflection data is transformed from the amplitude space into a multi-dimensional amplitude space. In order to minimize the interference of noise and uncertainties naturally present in the data, we apply a clustering procedure. After this step each voxel is represented by its corresponding cluster label. Based on these clusters, we compute a similarity function optimized for each particular dataset. This similarity function is non-local and auto-adaptable. The sub-surface mapping is provided by this function. The experimental results indicate the efficiency of the proposed method and illustrate its advantages.