1�� College of Geo- Exploration Science and Technology��Jilin University��Changchun 130026��China; 2�� The Second Institute of Oceanography��State Oceanic Administration��Hangzhou 310012��China; 3�� Key Laboratory of Submarine Geoscience��State Oceanic Administration��Hangzhou 310012��China

The ground penetrating radar (GP��) detection data is a wide band signal�� always disturbed by some noise��such as ambient random noise and multiple reflection waves�� The noise affects the target identification of underground medium seriously�� A method based on principal component analysis (PCA) was proposed to ex- tract the target signal and remove the uncorrelated noise�� According to the correlation of signal��the authors get the eigenvalues and corresponding eigenvectors by decomposing the covariance matrix of GP�� data and make linear transformation for the GP�� data to get the principal components (PCs)�� The lower- order PCs stand for the strong correlated target signals of the raw data��and the higher- order ones present the uncorrelated noise�� Thus the authors can extract the target signal and filter uncorrelated noise effectively by the PCA�� This method was demonstrated on real ultra- wideband through- wall radar data and simulated GP�� data�� Both of the results show that the PCA method can effectively extract the GP�� target signal and remove the uncorrelated noise��