Application of Discrete Cosine Transfor in Noise Reduction and Improvement in the Classification of Lithofacies

Methods of Multivariate Analysis (MVA) has always been the object of research in geophysics, because of wide applicability in situations with little information. The abundant amount of research in this area could strengthen the method and applications of MVA as a standard data processing in several areas. This research is based on an attempt to increase the hit rate of MVA methods, improving their applications. The Discrete Fourier of cosines, known by the acronym DCT, literal translation from English Discrete Cosine Transform, a data transformation is widely used for data compression, dimensionality reduction and noise reduction, the latter the reason for its application in geophysical data. The lithofacies classification was performed by combining the method of Independent Component Analysis (ICA) associated with the method of K-nearest neighbor (K-NN) or the junction of the method of Principal Component Analysis (PCA) and the K-NN. The DCT is used as preprocessing of data prior to applying the methods of PCA and ICA. The results obtained in the classification with and without the application of DCT are compared to demonstrate the effectiveness of the method. The result of these methods is analyzed quantitatively by success rate of the process.