Technique for Signal Noise Reduction based on Sparse Representation
Carlos Morales-Perez, Jose Rangel‐Magdaleno, Hayde Peregrina Barreto, Jorge Martínez‐Carballido · 2018
Noise is present in everyday life, that it is a fact to find it in everything. Depending on the applications, the noise can present or not issues. In sensing process, the noise would be a huge issue when the accuracy of the system is high, when the resolution of the sensor is better than, e.g., 40 mV/A. In this case, the outputs less than 3A have a high SNR. If these signals are used in signals processing, depending on its focus, they can provoke malfunctions and false positives. This paper presents, a methodology for noise reduction based on the sparse representation of a signal using a dictionary based on Discrete Cosine Transform (DCT) and Discrete Sine Transform (DST). The proposed method is tested with synthetic signals to research its effectiveness, and then it is tested with real signals taken from the current that supply Induction Motors. After the methodology is applied, signals are analyzed to prove that the information contained is preserved. The noise reduction of the current signals was ~30dB with the Motor under normal operating conditions.