Photoacoustic Microscopy Signal Analysis And Denoising Based On Time Frequency Methods

In photoacoustic microscopy imaging, noise will decrease the signal's signal to noise ratio and degrade the imaging quality. Therefore the signal's analysis and denoising is an important research topic. Time frequency methods can simultaneously analyze and denoise the signal from the aspects of time and frequency. This makes them suitable for the research of the photoacoustic microscopy imaging signal. In this paper, we use different kinds of time frequency methods to analyze and denoise the photoacoustic microscopy imaging signal. The main research contents include:Firstly, we simulated the photoacoustic microscopy imaging signal based on physical principles and the noisy signal with different kinds of noise. We used different kinds of time frequency methods to analyze simulated signals, including the short time Fourier transform, Wigner-Ville distribution, Choi-Williams distribution, Zhao-Atlas-Marks distribution, wavelet transform based on haar, db4, sym11 wavelets and Hilbert-Huang transform. The results showed that these methods have different characteristics and can get different aspects of the nature of the signal and noise. Joint application of these methods can have better results.Secondly, we used these time frequency methods to analyze photoacoustic microscopy imaging experiments' signals and study their nature. The results showed that Cohen's class distribution, wavelet transform and Hilbert-Huang transform can distinguish the meaningful signal from the noise by distributing the meaningful signal in a small area of the time frequency plane or a few part of the decomposition components.Finally, we used frequency denoising based on Cohen's class distribution, wavelet denoising and empirical mode decomposition denoising to denoise photoacoustic microscopy imaging experiments' signals. And we quantitatively evaluated the results. We then selected methods with better and more stable denoising results to denoise the experiment's whole data and reconstructed the imaging picture. The results showed that these time frequency methods can improve the signal's signal to noise ratio and signal to noise amplitude ratio, reduce noise' energy, make it easier to identify the meaningful signal and enhance the imaging result.