Space-time linear predictive filtering for the improvement of signal-to-noise ratio in ultrasonic imaging

Poor quality of ultrasonic images is the result of scattering noise that is caused by the target´s surroundings, when the material´s grain size is smaller than the wavelength of the transmitted signal. A C-scan can be obtained by moving the transducer over a sample and acquiring M×N number of A-scans, each having T time samples, from M×N different equally spaced locations. The nature of the C-scan implies a three-dimensional signal, since it consists of M×N time series (A-scans) which were acquired in space. It is shown how a three-dimensional least squares linear predictor can be used to increase the signal-to-noise ratio in ultrasonic imaging. The optimum linear predictor´s coefficients are obtained by minimizing the total squared error in a localized space analysis area, which consists of K×L number of A-scans