One-shot phase retrieval method for interferometry using a hypercolumns convolutional neural network

One step accurate phase demodulation from a closed fringe pattern with the convolutional neural network HRUnet

Retrieving a phase map from a single closed fringe pattern is a challenging task in optical interferometry. In this paper, a convolutional neural network (CNN), HRUnet, is proposed to demodulate phase from a closed fringe pattern. The HRUnet, derived from the Unet model, adopts a high resolution network (HRnet) module to extract high resolution feature maps of the data and employs residual blocks to erase the gradient vanishing in the network. With the trained network, the unwrapped phase map can be directly obtained by feeding a scaled fringe pattern. The high accuracy of the phase map obtained from HRUnet is demonstrated by demodulation of both simulated data and actual fringe patterns. Compared results between HRUnet and two other CNNS are also provided, and the results proved that the performance of HRUnet in accuracy is superior to the two other counterparts.

Phase retrieval for single-frame interferogram with an irregular-shaped aperture based on deep learning

This paper proposes a high-precision phase retrieval method based on deep learning to extract the Zernike coefficients from a single-frame interferogram with an irregular-shaped aperture. Once the Zernike coefficients are obtained, the phase distribution can be retrieved directly using the Zernike polynomials. For many apertures, the root mean square (RMS) of the residual wavefront between the true and estimated wavefronts reached the order of 10-3 λ. Simulations were conducted under different noise conditions, indicating that the proposed method has high measurement accuracy and robustness. Experiments demonstrated that the accuracy achieved by this method was comparable to that of commercial phase-shifting interferometers. We believe that this method is useful for measuring optical surfaces with irregular apertures.

Phase retrieval of two random phase-shifting interferograms using Zernike coefficient extraction network

This paper proposes a deep learning method for phase retrieval from two interferograms. The proposed method converts phase retrieval into the Zernike coefficient extraction problem, which can achieve Zernike coefficient extraction from two interferograms with random phase shifts. After knowing Zernike coefficients, the phase distribution can be retrieved using Zernike polynomials. The pre-filtering and phase unwrapping process are not required using the proposed method. The simulated data are analyzed, and the root mean square (RMS) of phase error reaches 0.01 λ. The effectiveness of the method is verified by the measured data.

Two-step orthogonalization phase demodulation method based on a single differential interferogram

To reduce the acquisition time of interferogram and provide a dynamic phase retrieval method with arbitrary phase shift using a dual-channel simultaneous polarization phase-shifting system, a two-step orthogonalization phase demodulation method (TOPD) based on a single differential interferogram is proposed in this paper. In this method, the differential interferogram obtained by subtracting two phase-shifting interferograms and one of the Gaussian filtered based-interferograms are used to normalize and orthogonalize, and then the phase related parameters are solved by the Lissajous ellipse fitting method. Finally, the measured phase is obtained with high accuracy. The proposed method further reduces the deviation caused by the filtering operation performed in the two-step phase demodulation method. At the same time, combined it with the Lissajou ellipse fitting method reduces the limitation associated with the approximation conditions of the orthogonalization and normalization method. The experimental and simulation results demonstrate that this method provides a solution with high accuracy, high stability, strong practicability, and few restrictions for phase extraction in quantitative phase imaging.

The Progress of Medical Image Semantic Segmentation Methods for Application in COVID-19 Detection

Image medical semantic segmentation has been employed in various areas, including medical imaging, computer vision, and intelligent transportation. In this study, the method of semantic segmenting images is split into two sections: the method of the deep neural network and previous traditional method. The traditional method and the published dataset for segmentation are reviewed in the first step. The presented aspects, including all-convolution network, sampling methods, FCN connector with CRF methods, extended convolutional neural network methods, improvements in network structure, pyramid methods, multistage and multifeature methods, supervised methods, semiregulatory methods, and nonregulatory methods, are then thoroughly explored in current methods based on the deep neural network. Finally, a general conclusion on the use of developed advances based on deep neural network concepts in semantic segmentation is presented.