The Denoising Method of Magnetic Gradient Tensor Tracking System Based on SVD

Due to the noises in signals of magnetic gradient tensor system influence the tracking precision heavily, put forward a denosing method based on SVD. Described the principle of magnetic gradient tensor tracking and gave a plane structure of tensor system. Analysed the noises in magnetic gradient tensor signals. Described the process of SVD, and analysed the characteristic of singular values represent signal and noises. To confirm a good reconstruction order k, adopt the method of fuzzy mean C to do cluster analysing on singular values. At last designed an experiment of magnetic gradient tensor tracking and collected the real signals. Using the denosing method based on SVD to reduce the noises in signals and the signals reached to a good SNR.