CN109612686B - Calibration method of dispersion confocal measuring device based on neural network - Google Patents

Calibration method of dispersion confocal measuring device based on neural network Download PDF

Info

Publication number
CN109612686B
CN109612686B CN201811490138.9A CN201811490138A CN109612686B CN 109612686 B CN109612686 B CN 109612686B CN 201811490138 A CN201811490138 A CN 201811490138A CN 109612686 B CN109612686 B CN 109612686B
Authority
CN
China
Prior art keywords
light intensity
weight coefficient
spectral response
intensity sequence
neural network
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
Application number
CN201811490138.9A
Other languages
Chinese (zh)
Other versions
CN109612686A (en
Inventor
卢文龙
张篪
陈成
王健
刘晓军
周莉萍
汪洁
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN201811490138.9A priority Critical patent/CN109612686B/en
Publication of CN109612686A publication Critical patent/CN109612686A/en
Application granted granted Critical
Publication of CN109612686B publication Critical patent/CN109612686B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01MTESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M11/00Testing of optical apparatus; Testing structures by optical methods not otherwise provided for

Landscapes

  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Microscoopes, Condenser (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

The invention discloses a calibration method of a dispersion confocal measuring device based on a neural network, which utilizes the dispersion confocal measuring device to measure the measuring position of piezoelectric ceramics and process to obtain a new light intensity sequence of a normalized spectral response signal; forming a light intensity set by the light intensity sequences of all the new normalized spectral response signals and a measurement position set by the corresponding measurement positions, and randomly dividing the light intensity set into a training set and a test set; inputting the light intensity sequences of all the training sets into a neural network, and obtaining a predicted position value corresponding to the light intensity signal after the weight coefficient matrix operation of the neural network; and calculating errors E between the predicted light intensity sequence position values of all the training sets and the actual measured values, adjusting each weight coefficient in the weight coefficient matrix according to the errors E, and verifying that each weight coefficient of the weight coefficient matrix is the calibration value of the dispersion confocal measuring device when the predicted light intensity sequence position values of all the test sets meet preset conditions, thereby improving the accuracy of the calibration value of the dispersion confocal measuring device.

Description

Calibration method of dispersion confocal measuring device based on neural network
Technical Field
The invention belongs to the field of confocal measurement, and particularly relates to a calibration method of a dispersion confocal measurement device based on a neural network.
Background
The confocal microscopic measurement technology is mainly used for eliminating multiple scattered light when a common microscope detects a sample, is similar to confocal microscopic measurement, the dispersion confocal microscopic technology focuses and images a point light source on a rear image surface after passing through a half-mirror and a dispersion lens, and the dispersion lens can focus different incident light waves at different axial positions. The dispersive confocal microscopy technology is commonly used for micro-nano surface structure measurement, rapid coordinate measurement, transparent medium thickness measurement, online process detection and the like.
The dispersive confocal measuring device is an optical testing device manufactured based on the principle. The calibration (calibration) is needed in practical application. When the testing device is calibrated, a large amount of position data in a measuring range and a spectral response signal corresponding to the position need to be collected, and a mapping relation between the position data and the spectral response signal can be established through algorithm processing, wherein the related algorithm is called a calibration algorithm or a calibration method.
The existing calibration method mainly utilizes a peak value extraction algorithm to extract a wavelength value (hereinafter referred to as peak wavelength) corresponding to the peak value of a spectral response signal, and establishes a mapping relation between the peak wavelength and a measurement position. The existing calibration method is as follows:
1. the linear relation is used for describing the mapping relation between the peak wavelength and the measured position, however, the testing device has errors in the processes of design, processing, assembly and the like, so that the peak wavelength and the measured position are not in the linear mapping relation, and therefore, the traditional linear relation calibration method is difficult to construct the mapping relation characteristic;
2. the united states patent US7876456 discloses a method for calibrating the mapping relationship between the peak wavelength and the measurement position by using the nonlinear relationship of a polynomial, however, in the actual measurement, factors such as noise level, signal level and light intensity non-uniformity directly affect the accuracy of the mapping relationship, and the accurate mapping relationship cannot be obtained simply by using the nonlinear relationship of the polynomial;
3. the mapping relation between the peak wavelength and the measurement position is calibrated by adopting a piecewise linear relation, however, factors such as noise level, signal level, light intensity nonuniformity and the like in actual measurement directly influence the accuracy of the mapping relation, and meanwhile, certain errors existing in the extracted peak wavelength also influence the accuracy of the mapping relation.
Disclosure of Invention
Aiming at the defects or improvement requirements in the prior art, the invention provides a calibration method of a chromatic dispersion confocal measuring device based on a neural network, which comprises the steps of dividing all new light intensity sequences of normalized spectral response signals and corresponding measuring positions into a training set and a testing set by using the neural network, adjusting the calibration value of the chromatic dispersion confocal measuring device by using the training set, and further testing the accuracy of the calibration value by using the testing set, thereby improving the accuracy of the calibration value of the chromatic dispersion confocal measuring device.
In order to achieve the above object, according to an aspect of the present invention, there is provided a calibration method for a chromatic dispersion confocal measurement apparatus based on a neural network, comprising the following steps:
s1, measuring the measuring position of the piezoelectric ceramic by using a dispersion confocal measuring device to obtain spectral response signals corresponding to different positions, and filtering the light intensity sequence of the spectral response signals to obtain a new light intensity sequence of normalized spectral response signals;
s2, forming a light intensity set by the light intensity sequences of all the new normalized spectral response signals, forming a measurement position set by the measurement positions corresponding to the light intensity sequences of all the new normalized spectral response signals, and randomly dividing data in the light intensity set and the measurement position set into a training set and a test set;
s3, inputting the light intensity sequence of the training set into a neural network, and obtaining a predicted position value corresponding to the light intensity signal after the weight coefficient matrix operation of the neural network;
s4, calculating an error E between the light intensity sequence predicted position value of the training set and the actual measured value, adjusting each weight coefficient in the weight coefficient matrix according to the error E, recalculating the light intensity sequence predicted position values of the training set according to the step S3 until the light intensity sequence predicted position values of all the training sets meet preset conditions, and when the light intensity sequence predicted position values of all the test sets meet the preset conditions, each weight coefficient of the weight coefficient matrix is the calibration value of the dispersion confocal measuring device.
As a further improvement of the present invention, in step S1, an Alpha statistical filtering method is used to filter the light intensity sequence of the spectral response signal to obtain a new light intensity sequence of the normalized spectral response signal, which specifically includes:
s1.1 starts with i ═ 1, and calculates the light intensity sequence of the spectral response signalTaken out of the reaction vessel with IiA light intensity sequence of length 2n-1 as a start bit
Figure BDA0001895487260000022
Light intensity sequence of spectral response signal
Figure BDA0001895487260000023
The expression of (a) is:
Figure BDA0001895487260000024
I1、Im-1and ImRespectively representing the light intensity of the 1 st, the m-1 st and the m-th sampling points of the spectral response signal;
Figure BDA0001895487260000025
the mathematical expression is:
Figure BDA0001895487260000026
Iiand Ii+2n-2Respectively representing the light intensity of the ith and the (i +2 n-1) th sampling points of the spectral response signal;
S1.2
Figure BDA0001895487260000027
the middle 2n-1 light intensity values are sorted in a descending manner, the front m larger values and the rear m smaller values are removed to obtain an average value IvUpdating the light intensity value I of the I + n sampling points of the spectral response signali+nIs Iv
S1.3 making i equal to i +1, and sequentially and iteratively calculating the light intensity sequence
Figure BDA0001895487260000031
Until i is m-2n +2, and normalizing the filtered light intensity sequence to obtain a new light intensity sequence of the normalized spectral response signal
As a further improvement of the present invention, step S3 specifically includes:
the neural network comprises an input layer, a first hidden layer, a second hidden layer and an output layer, wherein the input layer is the light of the ith new normalized spectral response signalStrong sequence
Figure BDA0001895487260000033
The output layer is the predicted position value corresponding to the ith new normalized spectral response signalThe weight coefficient matrixes between the input layer and the first hidden layer, between the first hidden layer and the second hidden layer, and between the second hidden layer and the output layer are respectively Wmj、WjkAnd WkWherein, the operation relationship is a matrix multiplication relationship, namely:
Figure BDA0001895487260000035
as a further improvement of the present invention, step S4 specifically includes:
the error E is calculated as:
Figure BDA0001895487260000036
wherein d isiActual measurement values corresponding to the ith new normalized spectral response signal;
the light intensity sequence predicted position values of all the training sets meet the preset condition:
Figure BDA0001895487260000037
the light intensity sequence predicted position values of all the test sets meet the preset condition that:
Figure BDA0001895487260000038
as a further improvement of the present invention, adjusting each weight coefficient in the weight coefficient matrix according to the error E specifically includes: and expressing the error E as a function of each weight coefficient in the weight coefficient matrix, solving the partial derivative of the error E on each weight coefficient in the weight coefficient matrix, and obtaining the optimized step length of each weight coefficient in the weight coefficient matrix according to the partial derivative value.
Generally, compared with the prior art, the above technical solution conceived by the present invention has the following beneficial effects:
the invention relates to a calibration method of a chromatic dispersion confocal measuring device based on a neural network, which divides light intensity sequences of all new normalized spectral response signals and corresponding measuring positions into a training set and a testing set by using the neural network, adjusts calibration parameters of the chromatic dispersion confocal measuring device by using the training set, and further tests the accuracy of the calibration parameters by using the testing set, thereby not needing to extract peak values of the spectral response signals, having no peak value extraction error, and having higher accuracy of the calibration parameters of the chromatic dispersion confocal measuring device compared with the traditional calibration method.
According to the calibration method of the dispersion confocal measurement device based on the neural network, the Alpha statistical filtering method is utilized to obtain a new light intensity sequence of the normalized spectral response signal, so that salt and pepper noise of the spectral response signal is filtered, and the accuracy of the calibration value of the dispersion confocal measurement device is further improved.
The calibration method of the dispersion confocal measurement device based on the neural network adjusts the optimized step length of each weight coefficient in the weight coefficient matrix of the neural network by utilizing the light intensity sequence of the new normalized spectral response signal and the error value of the corresponding measurement position, thereby further improving the accuracy of the calibration value of the dispersion confocal measurement device.
Drawings
Fig. 1 is a schematic diagram of an Alpha statistical filtering method of a calibration method of a dispersion confocal measurement apparatus based on a neural network according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a neural network of a calibration method of a dispersion confocal measurement apparatus based on the neural network according to an embodiment of the present invention;
fig. 3 is a schematic diagram of the error comparison between the technical solution of the embodiment of the present invention and the technical solution in the prior art.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other. The present invention will be described in further detail with reference to specific embodiments.
A calibration method of a dispersion confocal measuring device based on a neural network comprises the following steps:
s1, measuring the measuring position of the piezoelectric ceramic by using a dispersion confocal measuring device to obtain spectral response signals corresponding to different positions, and filtering the light intensity sequence of the spectral response signals to obtain a new light intensity sequence of the normalized spectral response signals.
Spectral response signal representation of spectrometer output in dispersive confocal measurement apparatus as implicit function
Figure BDA0001895487260000041
Wherein the content of the first and second substances,is a sequence of wavelengths of the spectral response signal,
Figure BDA0001895487260000043
λ1、λm-1and λmRespectively representing the wavelengths of the 1 st, m-1 st and m-th sampling points;
Figure BDA0001895487260000044
is a sequence of light intensities of the spectral response signal,
Figure BDA0001895487260000045
I1、Im-1and ImRespectively representing the light intensity of the 1 st, m-1 st and m-th sampling points.
Since a large amount of noise is doped in the spectral signal output by the spectrometer, if the spectral signal is directly used for calibration, the accuracy of establishing the mapping relationship between the peak wavelength and the measured position is affected, and therefore, the spectral response signal needs to be filtered.
Fig. 1 is a schematic diagram of an Alpha statistical filtering method of a calibration method of a dispersion confocal measurement apparatus based on a neural network according to an embodiment of the present invention. As shown in fig. 1, as a preferred embodiment of the present invention, an Alpha statistical filtering method is used to filter the spectral response signal, specifically:
s1.1 starts with i ═ 1, and calculates the light intensity sequence of the spectral response signalTaken out of the reaction vessel with IiA light intensity sequence of length 2n-1 as a start bit
Figure BDA0001895487260000052
Figure BDA0001895487260000053
The mathematical expression is:
Figure BDA0001895487260000054
S1.2
Figure BDA0001895487260000055
the middle 2n-1 light intensity values are sorted in a descending manner, the front m larger values and the rear m smaller values are removed to obtain an average value IvUpdating the light intensity value I of the I + n sampling pointsi+nIs Iv(ii) a Thereby removing the obvious salt and pepper noise in the light intensity sequence;
s1.3 making i equal to i +1, and sequentially and iteratively calculating the light intensity sequenceUntil i is m-2n +2, and normalizing the filtered light intensity sequence to obtain a new light intensity sequence of the normalized spectral response signal
Figure BDA0001895487260000057
S2, forming a light intensity set I by using all the new light intensity sequences of the normalized spectral response signals, forming a measurement position set D by using the measurement positions corresponding to all the new light intensity sequences of the normalized spectral response signals, wherein the specific expression is as follows:
Figure BDA0001895487260000058
wherein the content of the first and second substances,
Figure BDA0001895487260000059
and
Figure BDA00018954872600000510
light intensity sequences of new normalized spectral response signals obtained from the 1 st, 2 nd, p-1 st and p-th sampling, respectively, d1、d2、dp-1And dpRespectively the measurement positions obtained by sampling at the 1 st time, the 2 nd time, the p-1 st time and the p th time.
The data in the set of light intensities I and the set of measurement locations D are randomly divided into a training set and a test set. The training set trains and optimizes the weight coefficient matrix in the neural network, and the test set checks the mapping relation description capacity of the optimized neural network.
S3, concentrating the training into a light intensity sequence
Figure BDA00018954872600000511
Inputting the signal into a neural network, and obtaining a predicted position value corresponding to the light intensity signal after hidden layer operation
Figure BDA00018954872600000513
The method specifically comprises the following steps:
fig. 2 is a schematic structural diagram of a neural network of a calibration method of a chromatic dispersion confocal measurement apparatus based on the neural network according to an embodiment of the present invention. As shown in FIG. 2, the neural network consists of inputsThe input layer, the double hidden layers and the output layer. Wherein the input layer is the light intensity sequence of the ith new normalized spectral response signal
Figure BDA0001895487260000061
The output layer is the predicted position value corresponding to the ith new normalized spectral response signal
Figure BDA0001895487260000062
The weight coefficient matrixes between the input layer and the hidden layer, between the hidden layer and the hidden layer, and between the hidden layer and the output layer are respectively Wmj、WjkAnd WkWherein, the operation relationship is a matrix multiplication relationship, namely:
Figure BDA0001895487260000063
s4, calculating an error E between the light intensity sequence predicted position value of the training set and an actual measured value, adjusting each weight coefficient in the weight coefficient matrix according to the error E, recalculating the predicted position value according to the step S3 until the light intensity sequence predicted position values of all the training sets meet preset conditions, and when the light intensity sequence predicted position values of all the test sets meet the preset conditions, each weight coefficient of the weight coefficient matrix is a calibration value; the method specifically comprises the following steps:
calculating the light intensity sequence predicted position values of all training sets
Figure BDA0001895487260000064
And the actual measured value diE, adjusting the weight coefficient matrix W according to the error Emj、WjkAnd WkAnd recalculates the predicted position value in accordance with step S3Until the light intensity sequence of all training sets predicts the position value
Figure BDA0001895487260000066
Meet the preset conditions and calculate all tests simultaneouslyWhen the predicted position value of the collected light intensity sequence meets the preset condition, the weight coefficient W of the neural network can be completedmj、WjkAnd WkWherein the error is calculated by the following formula:
Figure BDA0001895487260000067
wherein d isiActual measurement values corresponding to the ith new normalized spectral response signal;
the light intensity sequence predicted position values of all the training sets meet the preset condition:
the light intensity sequence predicted position values of all the test sets meet the preset condition that:
if the predicted light intensity sequence position values of the test set do not meet the preset conditions, recalculating errors E between the predicted light intensity sequence position values of all the training sets and the actual measured values, adjusting the weight coefficients in the weight coefficient matrix according to the errors E, recalculating the predicted position values according to the step S3 until the predicted light intensity sequence position values of all the training sets meet the preset conditions, recalculating whether the predicted light intensity sequence position values of all the test sets meet the preset conditions, and training again until the predicted light intensity sequence position values of all the training sets and the test sets meet the preset conditions if the predicted light intensity sequence position values of all the training sets and the test sets do not meet the preset conditions.
Adjusting the weight coefficient matrix W according to the error Emj、WjkAnd WkThe weight coefficients in (1) adopt a chain type derivation rule, and specifically comprise: and expressing the error E as a function of each weight coefficient in the weight coefficient matrix, solving the partial derivative of the error E on each weight coefficient in the weight coefficient matrix, and obtaining the optimized step length of each weight coefficient in the weight coefficient matrix according to the partial derivative value.
Fig. 3 is a schematic diagram of the error comparison between the technical solution of the embodiment of the present invention and the technical solution in the prior art. As shown in fig. 3, the technical solution in the prior art includes linear relationship fitting, polynomial relationship fitting, and piecewise linear relationship fitting, and it can be seen that the total error obtained by the calibration method of the chromatic dispersion confocal measurement apparatus based on the neural network of the present invention is smaller than that obtained by the linear relationship fitting, the polynomial relationship fitting, and the piecewise linear relationship fitting in the prior art, which indicates that the calibration method of the chromatic dispersion confocal measurement apparatus based on the neural network of the present invention is more accurate in calibration of the chromatic dispersion confocal measurement apparatus.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (4)

1. A calibration method of a dispersion confocal measuring device based on a neural network is characterized by comprising the following specific steps:
s1, measuring the measuring position of the piezoelectric ceramic by using a dispersion confocal measuring device to obtain spectral response signals corresponding to different positions, and filtering the light intensity sequence of the spectral response signals to obtain a new light intensity sequence of normalized spectral response signals;
step S1 is to filter the light intensity sequence of the spectral response signal by an Alpha statistical filtering method to obtain a new light intensity sequence of the normalized spectral response signal, which specifically comprises:
s1.1 starts with i ═ 1, and calculates the light intensity sequence of the spectral response signal
Figure FDA0002264843480000011
Taken out of the reaction vessel with IiA light intensity sequence of length 2n-1 as a start bit
Figure FDA0002264843480000012
Light intensity sequence of spectral response signal
Figure FDA0002264843480000013
The expression of (a) is:I1、Im-1and ImRespectively representing the light intensity of the 1 st, the m-1 st and the m-th sampling points of the spectral response signal;
Figure FDA0002264843480000015
the mathematical expression is:Iiand Ii+2n-2Respectively representing the light intensity of the ith and the (i +2 n-1) th sampling points of the spectral response signal;
S1.2
Figure FDA0002264843480000017
the middle 2n-1 light intensity values are sorted in a descending manner, the front m larger values and the rear m smaller values are removed to obtain an average value IvUpdating the light intensity value I of the I + n sampling points of the spectral response signali+nIs Iv
S1.3 making i equal to i +1, and sequentially and iteratively calculating the light intensity sequence
Figure FDA0002264843480000019
Until i is m-2n +2, and normalizing the filtered light intensity sequence to obtain a new light intensity sequence of the normalized spectral response signal
S2, forming a light intensity set by the light intensity sequences of all the new normalized spectral response signals, forming a measurement position set by the measurement positions corresponding to the light intensity sequences of all the new normalized spectral response signals, and randomly dividing data in the light intensity set and the measurement position set into a training set and a test set;
s3, inputting the light intensity sequence of the training set into a neural network, and obtaining a light intensity sequence predicted position value corresponding to the light intensity signal after the weight coefficient matrix operation of the neural network;
s4, calculating an error E between the light intensity sequence predicted position value of the training set and the actual measured value, adjusting each weight coefficient in the weight coefficient matrix according to the error E, recalculating the light intensity sequence predicted position values of the training set according to the step S3 until the light intensity sequence predicted position values of all the training sets meet preset conditions, and when the light intensity sequence predicted position values of all the test sets meet the preset conditions, each weight coefficient of the weight coefficient matrix is the calibration value of the dispersion confocal measuring device.
2. The calibration method of the dispersive confocal measurement device based on the neural network as claimed in claim 1, wherein the step S3 specifically comprises:
the neural network comprises an input layer, a first hidden layer, a second hidden layer and an output layer, wherein the input layer is the light intensity sequence of the ith new normalized spectral response signal
Figure FDA0002264843480000021
The output layer is the predicted position value corresponding to the ith new normalized spectral response signal
Figure FDA0002264843480000022
The weight coefficient matrixes between the input layer and the first hidden layer, between the first hidden layer and the second hidden layer, and between the second hidden layer and the output layer are respectively Wmj、WjkAnd WkWherein, the operation relationship is a matrix multiplication relationship, namely:
Figure FDA0002264843480000023
3. the calibration method of the dispersive confocal measurement device based on the neural network as claimed in claim 2, wherein the step S4 specifically comprises:
the error E is calculated as:
Figure FDA0002264843480000024
wherein d isiActual measurement values corresponding to the ith new normalized spectral response signal;
the light intensity sequence predicted position values of all the training sets meet the preset condition:
Figure FDA0002264843480000025
the light intensity sequence predicted position values of all the test sets meet the preset condition that:
Figure FDA0002264843480000026
4. the calibration method of the dispersion confocal measurement device based on the neural network as claimed in claim 3, wherein the adjusting of each weight coefficient in the weight coefficient matrix according to the error E is specifically: and expressing the error E as a function of each weight coefficient in the weight coefficient matrix, solving the partial derivative of the error E on each weight coefficient in the weight coefficient matrix, and obtaining the optimized step length of each weight coefficient in the weight coefficient matrix according to the partial derivative value.
CN201811490138.9A 2018-12-06 2018-12-06 Calibration method of dispersion confocal measuring device based on neural network Active CN109612686B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811490138.9A CN109612686B (en) 2018-12-06 2018-12-06 Calibration method of dispersion confocal measuring device based on neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811490138.9A CN109612686B (en) 2018-12-06 2018-12-06 Calibration method of dispersion confocal measuring device based on neural network

Publications (2)

Publication Number Publication Date
CN109612686A CN109612686A (en) 2019-04-12
CN109612686B true CN109612686B (en) 2020-01-14

Family

ID=66007405

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811490138.9A Active CN109612686B (en) 2018-12-06 2018-12-06 Calibration method of dispersion confocal measuring device based on neural network

Country Status (1)

Country Link
CN (1) CN109612686B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112484968B (en) * 2020-11-20 2021-12-21 上海复享光学股份有限公司 Method, system, computing device and storage medium for optical metrology
CN114659465B (en) * 2022-05-20 2022-08-23 华中科技大学 Method and device for rapidly measuring micro-nano cross-scale surface structure

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101949826A (en) * 2010-09-02 2011-01-19 西安交通大学 Positive model and inverse model-based quantitative spectrometric analysis and calibration method of multi-component gas
US7876456B2 (en) * 2009-05-11 2011-01-25 Mitutoyo Corporation Intensity compensation for interchangeable chromatic point sensor components
CN105067617A (en) * 2015-07-21 2015-11-18 上海理工大学 Cell recognition apparatus and method based on phase contrast image and confocal scattering microspectrum
CN108020171A (en) * 2017-12-25 2018-05-11 东莞市三姆森光电科技有限公司 Mobile phone curved surface housing profile measuring method and its measuring apparatus based on Spectral Confocal technology

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7876456B2 (en) * 2009-05-11 2011-01-25 Mitutoyo Corporation Intensity compensation for interchangeable chromatic point sensor components
CN101949826A (en) * 2010-09-02 2011-01-19 西安交通大学 Positive model and inverse model-based quantitative spectrometric analysis and calibration method of multi-component gas
CN105067617A (en) * 2015-07-21 2015-11-18 上海理工大学 Cell recognition apparatus and method based on phase contrast image and confocal scattering microspectrum
CN108020171A (en) * 2017-12-25 2018-05-11 东莞市三姆森光电科技有限公司 Mobile phone curved surface housing profile measuring method and its measuring apparatus based on Spectral Confocal technology

Also Published As

Publication number Publication date
CN109612686A (en) 2019-04-12

Similar Documents

Publication Publication Date Title
JP6091493B2 (en) Spectroscopic apparatus and spectroscopy for determining the components present in a sample
CN109596543B (en) Spectral reflectivity reconstruction method for particle swarm optimization multi-core support vector regression
CN110006829B (en) Micro spectrometer calibration method based on least square method
CN110687072B (en) Calibration set and verification set selection and modeling method based on spectral similarity
CN109612686B (en) Calibration method of dispersion confocal measuring device based on neural network
CN104089911A (en) Spectral model transmission method based on unary linear regression
CN109766909B (en) Analysis method for aging behavior of shore environment microplastic based on spectrogram fusion
CN112462349A (en) Wavelength calculation method, system, server and storage medium for spectrum confocal displacement sensor
CN111879709B (en) Lake water body spectral reflectivity inspection method and device
WO2020186844A1 (en) Self-adaptive surface absorption spectrum analysis method and system, storage medium, and device
CN110926609A (en) Spectrum reconstruction method based on sample feature matching
CN113506235A (en) Adaptive weighted spectrum reconstruction method for resisting exposure change
CN115420806A (en) Nondestructive ultrasonic detection method based on neural network and image fusion
CN105699356A (en) Method for judging fluorescence quenching degree of Raman spectrum through information entropy
CN117664906B (en) Water body antibiotic pollution detection method based on spectral characteristics
CN109142251B (en) LIBS quantitative analysis method of random forest auxiliary artificial neural network
KR100769566B1 (en) Method and apparatus for measuring thin film thickness using neural network and record media therefor
EP1853898B1 (en) Method and system for physicochemical analysis using a laser pulsed ablation
CN116911183A (en) Karst wetland water chlorophyll a inversion method, system, equipment and medium
CN116879181A (en) Method for measuring optical characteristics and geometric characteristics of film material
CN104236710A (en) Spectrum super-resolution method of handheld light source color illumination photometry meter
CN111044504B (en) Coal quality analysis method considering uncertainty of laser-induced breakdown spectroscopy
CN110889368A (en) Method for eliminating Raman spectrum background by adopting arithmetic mean and geometric mean difference
KR20190091067A (en) Spectroscopic apparatus and spectroscopic method
CN114279566B (en) Method, device and system for correcting non-linearity of spectrometer radiance response

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