Research On The Denoising Of Digital Holographic Microscopic Measurement

Digital holographic micro-measurement technology has the advantages of noncontact,fast measurement speed,and can calculate the full-field three-dimensional shape using a single hologram.It is a research hotspot in the field of optical measurement in recent years,in biological sample measurement,microelectronic component measurement.MEMS device measurement,particle image velocimetry and other fields have broad application prospects.The noise problem is inevitable in the process of digital holography measurement,and it becomes a key factor affecting the measurement results.This thesis conducts research on this problem,and proposes a series of methods to improve the accuracy of reconstructing 3D topography,such as hologram denoising,zero-order image elimination,original image extraction and engraving phase map denoising.The main research contents include:Firstly,the basic principle of image holographic micro-measurement technology is studied.The technology realization process from hologram recording to subsequent three-dimensional shape reproduction in digital holographic micro-measurement technology is introduced in detail,and a single-wavelength reflective off-axis image surface is built.The digital holographic micro-measurement system has designed and developed a digital holographic micro-test software system to match the real-time measurement function of the measurement system.Secondly,the noise situation in the process of hologram recording is studied.The effect of optical path difference on the contrast of hologram interference fringes is verified by experiments.It is proved that the optical path difference adjustment unit improves the quality of hologram.Based on Lee filter and NLM algorithm,the corresponding improved algorithm is studied.It is proved that the improved NLM algorithm and SUSAN algorithm have advantages in image detail retention and noise level reduction.Thirdly,the digital holographic zero-order image elimination method based on Contourlet transform is studied.Compared with the hologram reduction intensity mean method,HRO subtraction method,local mean method and wavelet transform filtering method,it is proved that the zero-order image removal is more thorough and zero.The level of image energy attenuation is the largest.A weighted adaptive angular spectral filtering method is studied,which is compared with manual filtering,Butterworth filtering and binary adaptive filtering.It is proved that this method can preserve the narrow edge transition band while ensuring low noise level.The sinusoidal cosine filter denoising algorithm is studied.Compared with the spin filter denoising algorithm and the sine cosine mean filtering algorithm,the advantages of the proposed method in terms of guard fringe characteristics and retention of transition boundaries are proved.The above research provides support for the accurate measurement of the image digital holographic real-time measurement platform.The measurement platform will be further optimized,the data processing method will be improved,and the measurement platform will be developed in the direction of industrialization and commercialization.