CN112556601B - Single closed fringe interference pattern phase method and device based on deep learning - Google Patents
Single closed fringe interference pattern phase method and device based on deep learning Download PDFInfo
- Publication number
- CN112556601B CN112556601B CN202011377723.5A CN202011377723A CN112556601B CN 112556601 B CN112556601 B CN 112556601B CN 202011377723 A CN202011377723 A CN 202011377723A CN 112556601 B CN112556601 B CN 112556601B
- Authority
- CN
- China
- Prior art keywords
- phase
- neural network
- interferogram
- interference pattern
- denseblock
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
- G01B11/2441—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures using interferometry
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Instruments For Measurement Of Length By Optical Means (AREA)
Abstract
The method and the device have the advantages of simple structure, high precision and dynamic measurement, remarkably reduce the time required by interferogram phase, reduce the hardware requirement and cost required by interferogram phase resolution, and enable the method to be operated on portable equipment. The method comprises the following steps: (1) establishing a neural network and initializing; (2) preprocessing an interference pattern, namely performing gray scale range normalization on the input interference pattern to obtain a gray scale range normalized interference pattern; (3) processing the interference pattern by the neural network, inputting the interference pattern with normalized gray scale range into the neural network, and outputting a wrapping phase by the neural network; (4) and performing phase unwrapping on the wrapped phase to acquire the absolute phase of the interferogram.
Description
Technical Field
The invention relates to the technical field of photoelectric detection, in particular to a single closed fringe interference pattern phase solving method based on deep learning and a single closed fringe interference pattern phase solving device based on deep learning.
Background
Optical interferometry is an important class of non-contact surface profiling methods. Coherent light waves generate interference patterns with alternate light and shade through interference, and the interference patterns carry surface profile information of a measured object. By analyzing the interferogram, the phase distribution therein is obtained, and then the surface profile of the measured object can be obtained. The phase resolving of the interferogram, i.e. the acquisition of the phase distribution of the interferogram, is a central problem in optical interferometry, and the accuracy of the phase resolving of the interferogram directly determines the accuracy of the final measurement result.
Conventional interferogram phase-resolving methods can be broadly classified into two broad categories, phase-shifting methods and fourier methods. The phase shifting method extracts the phase by acquiring a plurality of fringe images, and has the advantages of good robustness, high precision and insensitivity to background illumination. However, the phase shift method is difficult to realize dynamic measurement due to the need of acquiring a plurality of interferograms. Compared with the phase shifting method, the Fourier method can extract the phase by only one fringe pattern, but the phase-shifting precision is generally lower than that of the phase shifting method. Moreover, the Fourier method cannot process a fringe pattern containing closed fringes, and the uneven background illumination of the fringe pattern can also influence the precision of the Fourier series method.
Since closed fringes are very common in interferograms and interferometers often use laser sources with non-uniform illumination, the phase-shifting method is mainly relied on for interferogram solution. However, the phase shift of the interferometer is generally realized by mechanical moving parts such as piezoelectric ceramics, so that the phase shift of the interferometer requires a long time, usually several seconds. This results in interferometers that are susceptible to vibration, gas turbulence, during phase shifting. Although there are now several methods available to achieve single closed fringe interference pattern dephasing. But these methods have some limitations more or less. In some occasions requiring dynamic measurement, the solution by a single interference pattern containing closed fringes still is a very practical problem.
The deep learning technology is a universal optimization method, and has the main advantage of being independent of the knowledge of people on problems. Traditional rule-based methods require a person to model a problem first, the accuracy of the method often depends on the degree of conformity of the model with realistic physical phenomena, and a high-accuracy model relies on the experience of the person with knowledge of the problem. The deep learning technology can dig out the intrinsic relation from the input and the output of the problem, and can automatically deduce the optimal model of the problem only by preparing a certain number of ideal inputs and outputs. In recent years, the deep learning technique is also gradually applied to a problem in the interferometer information solution.
For this purpose, the applicant filed an invention patent application (application No.: 2020107824652) at 2020.08.04 entitled: a closed fringe compatible single interference diagram phase method and device based on deep learning. The invention only needs one interference pattern in the measuring process and can realize dynamic measurement. However, the neural network used in the invention is a U-Net structure, which extracts features of an interferogram by scaling an input interferogram multiple times, thereby realizing the phase resolution of the interferogram. Therefore, the time required for the phase resolution of the interferogram is relatively long, and the hardware requirement for the phase resolution of the interferogram is strict and the cost is high.
Disclosure of Invention
In order to overcome the defects of the prior art, the technical problem to be solved by the invention is to provide a single closed fringe interference pattern phase method based on deep learning, which has the advantages of simple structure, high precision and dynamic measurement, obviously reduces the time required by interference pattern phase, and reduces the hardware requirement and cost required by interference pattern phase solution, so that the method can be operated on portable equipment.
The technical scheme of the invention is as follows: the single closed fringe interference pattern phase method based on the deep learning comprises the following steps:
(1) establishing a neural network and initializing;
(2) preprocessing an interference pattern, namely performing gray scale range normalization on the input interference pattern to obtain a gray scale range normalized interference pattern;
(3) processing the interference pattern by the neural network, inputting the interference pattern with normalized gray scale range into the neural network, and outputting a wrapping phase by the neural network;
(4) performing phase unwrapping on the wrapped phase to obtain an absolute phase of the interferogram;
the neural network structure in the step (1) is as follows: D1L1 is a gray scale normalized interferogram, which is an input of a neural network, D1L1 obtains D1L2 through a two-dimensional convolutional layer Conv2D, D1L2 obtains D1L3 through a dense block DenseBlock, and D1L3 obtains D1L4 through a dense block DenseBlock; rotating D1L2 by 90 degrees counterclockwise to obtain D2L2, passing D2L2 through a dense block DenseBlock to obtain D2L3, and passing D2L3 through a dense block DenseBlock to obtain D2L 4; rotating D1L2 counterclockwise by 180 degrees to obtain D3L2, passing D3L2 through a dense block DenseBlock to obtain D3L3, and passing D3L3 through a dense block DenseBlock to obtain D3L 4; rotating D1L2 counterclockwise by 270 degrees to obtain D4L2, passing D4L2 through a dense block DenseBlock to obtain D4L3, and passing D4L3 through a dense block DenseBlock to obtain D4L 4; D1L4, D2L4, D3L4 and D4L4 are connected by a connecting layer Concat to obtain D1L 5; D1L5 was passed through a dense block DenseBlock to yield D1L 6; D1L6 is obtained through a two-dimensional convolutional layer Conv2D to obtain D1L7, and D1L7 is the wrapping phase of the neural network output.
The method obtains the wrapping phase of the single interference image containing the closed fringes by establishing the neural network, obtains the absolute phase by phase unwrapping, realizes the unwrapping of the single interference image containing the closed fringes based on the deep learning technology, and has the advantages of simple structure, high precision and dynamic measurement; the neural network structure of the invention does not comprise the scaling process of the input interferogram, and the number of nodes in the network is less, so that the time required by the phase resolution of the interferogram is obviously reduced, and the hardware requirement and the cost required by the phase resolution of the interferogram are reduced, so that the method described by the invention can be operated on portable equipment.
Also provided is a single closed fringe interference pattern dephasing device based on deep learning, which comprises:
a building module configured to establish and initialize a neural network;
the preprocessing module is configured to preprocess the interference pattern, and perform gray scale range normalization on the input interference pattern to obtain a gray scale range normalized interference pattern;
a dephasing module configured to process the interferogram through a neural network, input the interferogram with a normalized gray scale range into the neural network, the neural network outputting a wrapped phase;
an unwrapping module configured to phase unwrapp the wrapped phase to obtain an absolute phase of the interferogram;
the neural network structure in the building module is as follows: D1L1 is a gray scale normalized interferogram, which is an input of a neural network, D1L1 obtains D1L2 through a two-dimensional convolutional layer Conv2D, D1L2 obtains D1L3 through a dense block DenseBlock, and D1L3 obtains D1L4 through a dense block DenseBlock; rotating D1L2 by 90 degrees counterclockwise to obtain D2L2, passing D2L2 through a dense block DenseBlock to obtain D2L3, and passing D2L3 through a dense block DenseBlock to obtain D2L 4; rotating D1L2 counterclockwise by 180 degrees to obtain D3L2, passing D3L2 through a dense block DenseBlock to obtain D3L3, and passing D3L3 through a dense block DenseBlock to obtain D3L 4; rotating D1L2 counterclockwise by 270 degrees to obtain D4L2, passing D4L2 through a dense block DenseBlock to obtain D4L3, and passing D4L3 through a dense block DenseBlock to obtain D4L 4; D1L4, D2L4, D3L4 and D4L4 are connected by a connecting layer Concat to obtain D1L 5; D1L5 was passed through a dense block DenseBlock to yield D1L 6; D1L6 is obtained through a two-dimensional convolutional layer Conv2D to obtain D1L7, and D1L7 is the wrapping phase of the neural network output.
Drawings
Fig. 1 is a flow chart of a single closed fringe interference pattern phase method based on deep learning according to the present invention.
Fig. 2 is a diagram of a neural network architecture.
Fig. 3 is an interference pattern including closed fringes.
FIG. 4 is a gray scale normalized interferogram.
Fig. 5 is a wrapped phase.
Fig. 6 is absolute phase.
Detailed Description
As shown in fig. 1, the single closed fringe interference pattern phase method based on deep learning includes the following steps:
(1) establishing a neural network and initializing;
(2) preprocessing an interference pattern, namely performing gray scale range normalization on the input interference pattern to obtain a gray scale range normalized interference pattern;
(3) processing the interference pattern by the neural network, inputting the interference pattern with normalized gray scale range into the neural network, and outputting a wrapping phase by the neural network;
(4) performing phase unwrapping on the wrapped phase to obtain an absolute phase of the interferogram;
the neural network structure in the step (1) is as follows: D1L1 is a gray scale normalized interferogram, which is an input of a neural network, D1L1 obtains D1L2 through a two-dimensional convolutional layer Conv2D, D1L2 obtains D1L3 through a dense block DenseBlock, and D1L3 obtains D1L4 through a dense block DenseBlock; rotating D1L2 by 90 degrees counterclockwise to obtain D2L2, passing D2L2 through a dense block DenseBlock to obtain D2L3, and passing D2L3 through a dense block DenseBlock to obtain D2L 4; rotating D1L2 counterclockwise by 180 degrees to obtain D3L2, passing D3L2 through a dense block DenseBlock to obtain D3L3, and passing D3L3 through a dense block DenseBlock to obtain D3L 4; rotating D1L2 counterclockwise by 270 degrees to obtain D4L2, passing D4L2 through a dense block DenseBlock to obtain D4L3, and passing D4L3 through a dense block DenseBlock to obtain D4L 4; D1L4, D2L4, D3L4 and D4L4 are connected by a connecting layer Concat to obtain D1L 5; D1L5 was passed through a dense block DenseBlock to yield D1L 6; D1L6 is obtained through a two-dimensional convolutional layer Conv2D to obtain D1L7, and D1L7 is the wrapping phase of the neural network output.
The method obtains the wrapping phase of the single interference image containing the closed fringes by establishing the neural network, obtains the absolute phase by phase unwrapping, realizes the unwrapping of the single interference image containing the closed fringes based on the deep learning technology, and has the advantages of simple structure, high precision and dynamic measurement; the neural network structure of the invention does not comprise the scaling process of the input interferogram, and the number of nodes in the network is less, so that the time required by the phase resolution of the interferogram is obviously reduced, and the hardware requirement and the cost required by the phase resolution of the interferogram are reduced, so that the method described by the invention can be operated on portable equipment.
Preferably, in the step (1), the neural network is trained by using the artificially generated virtual interferogram, so that the neural network is enabled.
Preferably, in the step (2), the gray scale normalization is as shown in formula (1):
where I is the input interferogram, I ' is the result of 5 × 5 median filtering of I, max (I ') is the maximum value of the interferogram pixel gray scale, min (I ') is the minimum value of the interferogram pixel gray scale, InormIs a normalized interferogram. The method for normalizing the gray scale range of the interference pattern has the step of median filtering, and the step of median filtering improves the precision and robustness of normalization of the gray scale range of the interference pattern, thereby improving the precision and robustness of resolving the phase of the interference pattern.
Preferably, in the step (4), the fast two-dimensional phase unwrapping method proposed by Miguel arevallo Herraez is used to perform unwrapping operation on the wrapped phase output by the neural network, so as to obtain the absolute phase of the interferogram.
It will be understood by those skilled in the art that all or part of the steps in the method of the above embodiments may be implemented by hardware instructions related to a program, the program may be stored in a computer-readable storage medium, and when executed, the program includes the steps of the method of the above embodiments, and the storage medium may be: ROM/RAM, magnetic disks, optical disks, memory cards, and the like. Therefore, corresponding to the method of the present invention, the present invention also includes a single closed fringe interference pattern resolving device based on deep learning, which is generally expressed in the form of functional modules corresponding to the steps of the method. The device includes:
a building module configured to establish and initialize a neural network;
the preprocessing module is configured to preprocess the interference pattern, and perform gray scale range normalization on the input interference pattern to obtain a gray scale range normalized interference pattern;
a dephasing module configured to process the interferogram through a neural network, input the interferogram with a normalized gray scale range into the neural network, the neural network outputting a wrapped phase;
an unwrapping module configured to phase unwrapp the wrapped phase to obtain an absolute phase of the interferogram;
the neural network structure in the building module is as follows: D1L1 is a gray scale normalized interferogram, which is an input of a neural network, D1L1 obtains D1L2 through a two-dimensional convolutional layer Conv2D, D1L2 obtains D1L3 through a dense block DenseBlock, and D1L3 obtains D1L4 through a dense block DenseBlock; rotating D1L2 by 90 degrees counterclockwise to obtain D2L2, passing D2L2 through a dense block DenseBlock to obtain D2L3, and passing D2L3 through a dense block DenseBlock to obtain D2L 4; rotating D1L2 counterclockwise by 180 degrees to obtain D3L2, passing D3L2 through a dense block DenseBlock to obtain D3L3, and passing D3L3 through a dense block DenseBlock to obtain D3L 4; rotating D1L2 counterclockwise by 270 degrees to obtain D4L2, passing D4L2 through a dense block DenseBlock to obtain D4L3, and passing D4L3 through a dense block DenseBlock to obtain D4L 4; D1L4, D2L4, D3L4 and D4L4 are connected by a connecting layer Concat to obtain D1L 5; D1L5 was passed through a dense block DenseBlock to yield D1L 6; D1L6 is obtained through a two-dimensional convolutional layer Conv2D to obtain D1L7, and D1L7 is the wrapping phase of the neural network output.
Preferably, in the building module, the neural network is trained by using the artificially generated virtual interferogram, so that the neural network is enabled.
Preferably, in the preprocessing module, the gray scale range normalization is shown in formula (1):
where I is the input interferogram, I ' is the result of 5 × 5 median filtering of I, max (I ') is the maximum value of the interferogram pixel gray scale, min (I ') is the minimum value of the interferogram pixel gray scale, InormIs a normalized interferogram.
Preferably, in the unwrapping module, the unwrapping operation is performed on the wrapped phase output by the neural network by using a fast two-dimensional phase unwrapping method proposed by Miguel arevallo Herraez, so as to obtain the absolute phase of the interferogram.
The invention has the following beneficial effects:
1. the closed fringe compatible single interference graph phase method based on deep learning solves the problem of phase solving of a single closed fringe interference graph by establishing a neural network, only one interference graph is needed in the measuring process, and dynamic measurement can be achieved.
2. According to the method for the closed fringe compatible single interference diagram phase based on the deep learning, the related calculation can effectively improve the calculation speed through the modes of multithreading or GPU calculation and the like, so that the target can be measured in real time.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, and all simple modifications, equivalent variations and modifications made to the above embodiment according to the technical spirit of the present invention still belong to the protection scope of the technical solution of the present invention.
Claims (8)
1. A single closed fringe interference pattern phase method based on deep learning is characterized in that: which comprises the following steps:
(1) establishing a neural network and initializing;
(2) preprocessing an interference pattern, namely performing gray scale range normalization on the input interference pattern to obtain a gray scale range normalized interference pattern;
(3) processing the interference pattern by the neural network, inputting the interference pattern with normalized gray scale range into the neural network, and outputting a wrapping phase by the neural network;
(4) performing phase unwrapping on the wrapped phase to obtain an absolute phase of the interferogram;
the neural network structure in the step (1) is as follows: D1L1 is a gray scale normalized interferogram, which is an input of a neural network, D1L1 obtains D1L2 through a two-dimensional convolutional layer Conv2D, D1L2 obtains D1L3 through a dense block DenseBlock, and D1L3 obtains D1L4 through a dense block DenseBlock; rotating D1L2 by 90 degrees counterclockwise to obtain D2L2, passing D2L2 through a dense block DenseBlock to obtain D2L3, and passing D2L3 through a dense block DenseBlock to obtain D2L 4; rotating D1L2 counterclockwise by 180 degrees to obtain D3L2, passing D3L2 through a dense block DenseBlock to obtain D3L3, and passing D3L3 through a dense block DenseBlock to obtain D3L 4; rotating D1L2 counterclockwise by 270 degrees to obtain D4L2, passing D4L2 through a dense block DenseBlock to obtain D4L3, and passing D4L3 through a dense block DenseBlock to obtain D4L 4; D1L4, D2L4, D3L4 and D4L4 are connected by a connecting layer Concat to obtain D1L 5; D1L5 was passed through a dense block DenseBlock to yield D1L 6; D1L6 is obtained through a two-dimensional convolutional layer Conv2D to obtain D1L7, and D1L7 is the wrapping phase of the neural network output.
2. The single closed fringe interference pattern phase method based on deep learning of claim 1, wherein: in the step (1), the neural network is trained by using the artificially generated virtual interferogram, so that the neural network reaches an available state.
3. The single closed fringe interference pattern phase method based on deep learning as claimed in claim 2, wherein: in the step (2), the normalization of the gray scale range is shown as formula (1):
where I is the input interferogram, I ' is the result of 5 × 5 median filtering of I, max (I ') is the maximum value of the interferogram pixel gray scale, min (I ') is the minimum value of the interferogram pixel gray scale, InormIs a normalized interferogram.
4. The single closed fringe interference pattern phase method based on deep learning as claimed in claim 3, wherein: in the step (4), a fast two-dimensional phase unwrapping method proposed by Miguel Arevallo Herraez is used for unwrapping the wrapped phase output by the neural network, so that the absolute phase of the interferogram is obtained.
5. Single closed fringe interference pattern looks device based on degree of deep learning, its characterized in that: it includes:
a building module configured to establish and initialize a neural network;
the preprocessing module is configured to preprocess the interference pattern, and perform gray scale range normalization on the input interference pattern to obtain a gray scale range normalized interference pattern;
a dephasing module configured to process the interferogram through a neural network, input the interferogram with a normalized gray scale range into the neural network, the neural network outputting a wrapped phase;
an unwrapping module configured to phase unwrapp the wrapped phase to obtain an absolute phase of the interferogram;
the neural network structure in the building module is as follows: D1L1 is a gray scale normalized interferogram, which is an input of a neural network, D1L1 obtains D1L2 through a two-dimensional convolutional layer Conv2D, D1L2 obtains D1L3 through a dense block DenseBlock, and D1L3 obtains D1L4 through a dense block DenseBlock; rotating D1L2 by 90 degrees counterclockwise to obtain D2L2, passing D2L2 through a dense block DenseBlock to obtain D2L3, and passing D2L3 through a dense block DenseBlock to obtain D2L 4; rotating D1L2 counterclockwise by 180 degrees to obtain D3L2, passing D3L2 through a dense block DenseBlock to obtain D3L3, and passing D3L3 through a dense block DenseBlock to obtain D3L 4; rotating D1L2 counterclockwise by 270 degrees to obtain D4L2, passing D4L2 through a dense block DenseBlock to obtain D4L3, and passing D4L3 through a dense block DenseBlock to obtain D4L 4; D1L4, D2L4, D3L4 and D4L4 are connected by a connecting layer Concat to obtain D1L 5; D1L5 was passed through a dense block DenseBlock to yield D1L 6; D1L6 is obtained through a two-dimensional convolutional layer Conv2D to obtain D1L7, and D1L7 is the wrapping phase of the neural network output.
6. The single closed fringe interference pattern phase device based on deep learning of claim 5, wherein: in the construction module, the artificial generated virtual interferogram is used for training the neural network, so that the neural network can reach an available state.
7. The single closed fringe interference pattern phase device based on deep learning of claim 6, wherein: in the preprocessing module, the normalization of the gray scale range is shown as a formula (1):
where I is the input interferogram, I ' is the result of 5 × 5 median filtering of I, max (I ') is the maximum value of the interferogram pixel gray scale, min (I ') is the minimum value of the interferogram pixel gray scale, InormIs a normalized interferogram.
8. The single closed fringe interference pattern phase device based on deep learning of claim 7, wherein: in the unwrapping module, unwrapping operation is carried out on the wrapped phase output by the neural network by using a rapid two-dimensional phase unwrapping method provided by Miguel Arevallo Herraez, so that the absolute phase of the interferogram is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011377723.5A CN112556601B (en) | 2020-11-30 | 2020-11-30 | Single closed fringe interference pattern phase method and device based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011377723.5A CN112556601B (en) | 2020-11-30 | 2020-11-30 | Single closed fringe interference pattern phase method and device based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112556601A CN112556601A (en) | 2021-03-26 |
| CN112556601B true CN112556601B (en) | 2021-08-20 |
Family
ID=75045646
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011377723.5A Active CN112556601B (en) | 2020-11-30 | 2020-11-30 | Single closed fringe interference pattern phase method and device based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112556601B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113340211B (en) * | 2021-08-03 | 2021-10-29 | 中国工程物理研究院激光聚变研究中心 | Interference image phase demodulation method based on deep learning |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110751268A (en) * | 2019-09-27 | 2020-02-04 | 北京理工大学 | Phase aliasing error removing method and device based on end-to-end convolutional neural network |
| CN111351450A (en) * | 2020-03-20 | 2020-06-30 | 南京理工大学 | Single-frame stripe image three-dimensional measurement method based on deep learning |
| CN111461224A (en) * | 2020-04-01 | 2020-07-28 | 西安交通大学 | Phase data unwrapping method based on residual self-coding neural network |
| CN111561877A (en) * | 2020-04-24 | 2020-08-21 | 西安交通大学 | Variable resolution phase unwrapping method based on point diffraction interferometer |
| CN111797678A (en) * | 2020-05-15 | 2020-10-20 | 华南师范大学 | Phase unwrapping method and device based on composite neural network |
| CN111928794A (en) * | 2020-08-04 | 2020-11-13 | 北京理工大学 | Closed fringe compatible single interference diagram phase method and device based on deep learning |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11526753B2 (en) * | 2019-02-12 | 2022-12-13 | Irida Labs S.A. | System and a method to achieve time-aware approximated inference |
-
2020
- 2020-11-30 CN CN202011377723.5A patent/CN112556601B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110751268A (en) * | 2019-09-27 | 2020-02-04 | 北京理工大学 | Phase aliasing error removing method and device based on end-to-end convolutional neural network |
| CN111351450A (en) * | 2020-03-20 | 2020-06-30 | 南京理工大学 | Single-frame stripe image three-dimensional measurement method based on deep learning |
| CN111461224A (en) * | 2020-04-01 | 2020-07-28 | 西安交通大学 | Phase data unwrapping method based on residual self-coding neural network |
| CN111561877A (en) * | 2020-04-24 | 2020-08-21 | 西安交通大学 | Variable resolution phase unwrapping method based on point diffraction interferometer |
| CN111797678A (en) * | 2020-05-15 | 2020-10-20 | 华南师范大学 | Phase unwrapping method and device based on composite neural network |
| CN111928794A (en) * | 2020-08-04 | 2020-11-13 | 北京理工大学 | Closed fringe compatible single interference diagram phase method and device based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 基于神经网络的干涉仪测向方法;崔功杰 等;《无线电工程》;20130205;第43卷(第2期);第16-20页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112556601A (en) | 2021-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106092158B (en) | Physical parameter method of estimation, device and electronic equipment | |
| CN101957182B (en) | Large-caliber high-gradient optical mirror surface online measurement system | |
| CN107917676B (en) | Interference measurement method based on fringe image spectrum analysis | |
| CN111928794B (en) | Closed fringe compatible single interference diagram phase method and device based on deep learning | |
| WO2021184686A1 (en) | Single-frame stripe analysis method based on multi-scale generative adversarial neural network | |
| CN106091978B (en) | The joining method of interference fringe image in inclined in type measurements by laser interferometry | |
| CN108765476A (en) | Polarized image registration method | |
| Yu et al. | PDNet: A lightweight deep convolutional neural network for InSAR phase denoising | |
| CN112556601B (en) | Single closed fringe interference pattern phase method and device based on deep learning | |
| CN102607455A (en) | Three-dimensional measurement method for microstructure based on optical microscope and variable illumination | |
| CN111189414B (en) | Real-time single-frame phase extraction method | |
| Flores et al. | Robust two-step phase estimation using the Simplified Lissajous Ellipse Fitting method with Gabor Filters Bank preprocessing | |
| Sun et al. | High-accuracy simultaneous phase extraction and unwrapping method for single interferogram based on convolutional neural network | |
| Lei et al. | The sensitivity and the measuring range of the typical differential optical flow method for displacement measurement using the fringe pattern | |
| Vithin et al. | Deep learning based single shot multiple phase derivative retrieval method in multi-wave digital holographic interferometry | |
| CN104331857A (en) | Phase position difference iteration compensation method in light intensity transmission equation phase retrieval | |
| Dvoynishnikov et al. | Optoelectronic method of contactless reconstruction of the surface profile of complexly shaped three-dimensional objects | |
| Lehmann et al. | Universal Fourier optics model for virtual coherence scanning interferometers | |
| CN103267485A (en) | Point-diffraction three-dimensional absolute displacement measuring method | |
| CN113011107B (en) | One-dimensional optical fiber sensing signal phase recovery method based on deep convolutional neural network | |
| Liu et al. | A Subpixel Displacement Estimation Algorithm for Digital Image Correlation Based on a Nonlinear Intensity Change Model. | |
| CN104463849A (en) | Color image registration method based on color invariant and phase correlation | |
| CN109781153A (en) | Physical parameter estimation method, device and electronic equipment | |
| Fan et al. | UNet++ for estimating physical parameters from Newton’s rings | |
| CN115128040B (en) | Method for recovering phase from single interference pattern based on inverse cosine pattern and continuity criterion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |