Construction method of simulation model of structured light diffuse reflection pattern in spatial frequency domain imaging

Abstract

The invention relates to the technical field of simulation model construction methods, and specifically discloses a construction method for a structured light diffuse reflection pattern simulation model in spatial frequency domain imaging, which specifically comprises the following steps: step 1, using the partial current boundary condition (PCBC) formula described by Farrell and Patterson to calculate the distribution of any diffuse reflection pattern of a sample; step 2, in spatial frequency domain imaging (SFDI), calculating the diffuse reflection coefficient based on optical characteristics at the fringe spatial frequency; in the invention, the advantages and disadvantages of PCBC and SFDI in simulating diffuse reflection patterns in spatial frequency domain imaging are analyzed, and the two are integrated through theoretical calculation, and a new method for simulating fringe diffuse reflection patterns with large field of view, high precision and real-time is proposed, and combined with CNN, it is realized that under the conditions of large fluctuations in sample optical characteristics, large fluctuations in height, different defocus levels, etc., the corresponding 8-bit result can still be well predicted according to a 1-bit experiment.

Description

Technical Field

[0001] The invention relates to the technical field of simulation model construction methods, in particular to a construction method of a structured light diffuse reflection pattern simulation model in spatial frequency domain imaging. Background Art

[0002] Spatial frequency domain imaging (SFDI) is a cutting-edge optical detection method whose basic principle is rooted in Monte Carlo simulation or diffusion approximation (DA) theory. This technique uses spatially modulated light projection to quantitatively analyze the optical properties of a sample, including absorption and reduced scattering coefficients. As a label-free and non-contact detection method, SFDI has been widely used to quantify the optical properties of strongly scattering media such as biological tissues. Although the imaging precision and accuracy of SFDI have been verified by numerous studies, the traditional three-phase demodulation method requires the sample to remain stationary when acquiring three phase-shifted images for demodulation calculation to avoid introducing artifacts, which limits its application in real-time online detection.

[0003] In order to increase the detection speed and thus improve its applicability in real-time monitoring, researchers in the prior art have explored two-phase and single-phase demodulation methods, aiming to improve imaging efficiency by reducing the number of required phase images. In addition, single-shot imaging technology aims to extract more information from a single projection pattern, thereby reducing the number of required patterns and increasing imaging speed, while maintaining the integrity of sample information as much as possible. In terms of projection technology, imaging speeds in the kilohertz range are achieved by using halftone (1-Bit) technology. Compared with traditional continuous tone (8-Bit) SFDI, 1-Bit SFDI uses a 1-Bit digital light pattern (binary stripes), which significantly reduces the amount of projection processing data and greatly increases the projection speed, achieving high-speed imaging. This technology can increase the projection speed of SFDI by about two orders of magnitude.

[0004] Although the halftone SFDI technique has made significant progress in processing speed, in-depth research on the underlying mechanism and construction of theoretical models are still necessary. Especially in the inversion process of optical properties, it is currently difficult to directly provide analytical solutions for reducing the scattering coefficient and absorption coefficient due to the computational complexity of SFDI. Summary of the invention

[0005] In view of the technical problems raised in the background technology, the present invention provides a method for constructing a simulation model of structured light diffuse reflection patterns in spatial frequency domain imaging.

[0006] The technical solution adopted by the present invention is: a method for constructing a simulation model of structured light diffuse reflection pattern in spatial frequency domain imaging, which specifically includes the following steps:

[0007] Step 1, calculate the arbitrary diffuse reflectance distribution of the sample using the partial current boundary condition (PCBC) formula described by Farrell and Patterson;

[0008] Step 2, in spatial frequency domain imaging (SFDI), the fringe spatial frequency Diffuse reflectance based on optical properties The calculation formula is as follows:

[0009] ,

[0010] In the formula, is the effective attenuation coefficient, is the extinction coefficient, is the internal reflection parameter, is the sample diffuse reflectance, is the transmission albedo;

[0011] Step 3, simulating the diffuse reflection pattern and calculating and establishing the PCBC-SFD fusion model;

[0012] Step 4: Set the sample parameters wavelength λ and absorption coefficient of the diffuse reflection pattern to be predicted , reduced scattering coefficient , input the PCBC-SFD fusion model to obtain the diffuse reflection simulation results corresponding to the projection pattern;

[0013] Step 5: The host computer generates 5 pictures, including a completely black picture, a completely white picture, and a selected spatial frequency. f x Three fringe patterns with different initial phases α under the ;

[0014] Step 6: Use the error diffusion method or the halftone fringe generation method to generate the continuous 8-bit fringe pattern into the corresponding 1-bit halftone fringe pattern, and the host computer controls the projector to project the fringe pattern onto the sample;

[0015] Step 7: The CCD camera synchronously collects the reflected light intensity distribution of the sample under the stripe pattern illumination mode and saves the image data.

[0016] The present invention is further configured such that, in step 1, the diffuse reflection distribution of the sample is calculated using the partial current boundary condition (PCBC) formula described by Farrell and Patterson, as follows:

[0017] ,

[0018] ,

[0019] in,r is the radial distance to the light source, is the distance to the light source, is the distance to the image source, is the effective attenuation coefficient, is the reduced scattering coefficient;

[0020] , , , , , , is the internal reflection parameter, is a parameter related to the tissue refractive index, ,and is the transmission albedo, is the reflectivity at a distance r from the light source, is the sample diffuse reflectance, is the extinction coefficient.

[0021] The present invention is further configured as follows: in step 3, the PCBC-SFD fusion model calculation process is as follows:

[0022] First, the calculation formula for the projection pattern light intensity distribution is as follows: is the amplitude envelope of the DC component, is the amplitude envelope of the AC component, is the intensity distribution of the three phase corresponding patterns, and the calculation formula is as follows:

[0023] ,

[0024] ,

[0025] The diffuse reflectance of the sample is determined by the ratio with the standard reflector ,

[0026] For two different samples, the corresponding relationship between them can be established

[0027] , , ,

[0028] Assignment: ,

[0029] The light intensity distribution of the projection, standard panel, and sample can be expressed as:

[0030] ,

[0031] ,

[0032] ,

[0033] The light intensity distribution of the sample diffuse reflection pattern and the sample , And the relationship between the standard plate and the projection pattern strength is as follows:

[0034] ,

[0035] The intensity of the projected fringe is expressed as 0.5+0.5cos(x), so the value of A is usually set to 0.5.

[0036] ,

[0037] Therefore, the sample is calculated The PCBC-SFD fusion model of the actual light intensity distribution is as follows: is the diffuse reflectance of the AC component under the projected fringe group with a spatial frequency of k calculated using SFDI, The pattern calculated by PCBC, To calculate the diffuse reflectance of the DC component represented by the frequency 0 using SFDI, is the diffuse reflectivity of the standard white plate:

[0038] .

[0039] The beneficial effects of the present invention are as follows: in the present invention, the advantages and disadvantages of PCBC and SFDI in simulating diffuse reflection patterns in spatial frequency domain imaging are analyzed, and the two are integrated through theoretical calculations, and a new method for simulating stripe diffuse reflection patterns with large field of view, high precision and real-time is proposed, and combined with CNN, it is realized that under the conditions of large fluctuations in sample optical properties, large fluctuations in height, different degrees of defocus, etc., the corresponding 8-Bit results can still be well predicted according to the 1-Bit experiment. This is beneficial to subsequent online detection in actual industries. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of an implementation process of the present invention.

[0041] Figure 2 It is a schematic diagram of the experimental device and experimental materials of the present invention.

[0042] Figure 3 It is a schematic diagram of the process of proposing the model of the present invention.

[0043] Figure 4 It is a schematic diagram of the convolutional neural network structure of the present invention.

[0044] Figure 5 It is a schematic diagram of the present invention using a convolutional neural network to predict continuous tone results based on a halftone pattern. DETAILED DESCRIPTION

[0045] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] In order to solve the problems existing in the background technology, the present application proposes the following technical solution: a method for constructing a simulation model of structured light diffuse reflection pattern in spatial frequency domain imaging, which specifically includes the following steps:

[0047] Step 1, calculate the arbitrary diffuse reflectance distribution of the sample using the partial current boundary condition (PCBC) formula described by Farrell and Patterson;

[0048] Among them, the partial current boundary condition (PCBC) described by Farrell and Patterson can provide more accurate diffuse reflection transmission simulation results and is widely used in various light transmission simulation experiments. Therefore, we use the PCBC formula proposed by the two to calculate the diffuse reflection distribution of the sample.

[0049] In step 1, the diffuse reflectance distribution of the sample is calculated using the partial current boundary condition (PCBC) formula described by Farrell and Patterson, as follows:

[0050] ,

[0051] ,

[0052] in, r is the radial distance to the light source, is the distance to the light source, is the distance to the image source, is the effective attenuation coefficient, is the reduced scattering coefficient;

[0053] , , , , , , is the internal reflection parameter, is a parameter related to the tissue refractive index, ,and is the transmission albedo, is the reflectivity at a distance r from the light source, is the sample diffuse reflectance, is the extinction coefficient.

[0054] Step 2, in spatial frequency domain imaging (SFDI), the fringe spatial frequency Diffuse reflectance based on optical properties The calculation formula is as follows:

[0055] ,

[0056] In the formula, is the effective attenuation coefficient, is the extinction coefficient, is the internal reflection parameter, is the sample diffuse reflectance, is the transmission albedo;

[0057] Step 3, simulating the diffuse reflection pattern and calculating and establishing the PCBC-SFD fusion model;

[0058] The SFDI formula can accurately calculate the diffuse reflectance of the sample, but it cannot give the distribution of the light source projected into the medium at a pixel point. PCBC can give a solution for the radial distance of the light source, but there will be a large error when the radial distance is too small, resulting in errors in the predicted intensity of the halftone stripe pattern in spatial frequency domain imaging. Therefore, the present invention achieves a fast and accurate simulation of the stripe diffuse reflection pattern by calculating the fusion model PCBC-SFD of the two.

[0059] First, the calculation formula for the projection pattern light intensity distribution is as follows: is the amplitude envelope of the DC component, is the amplitude envelope of the AC component, is the intensity distribution of the three phase corresponding patterns, and the calculation formula is as follows:

[0060] ,

[0061] ,

[0062] The diffuse reflectance of the sample is determined by the ratio with the standard reflector ,

[0063] For two different samples, the corresponding relationship between them can be established

[0064] ,

[0065] ,

[0066] ,

[0067] Assignment: ,

[0068] The light intensity distribution of the projection, standard panel, and sample can be expressed as:

[0069] ,

[0070] ,

[0071] ,

[0072] The light intensity distribution of the sample diffuse reflection pattern and the sample , And the relationship between the standard plate and the projection pattern strength is as follows:

[0073] ,

[0074] The intensity of the projected fringe is expressed as 0.5+0.5cos(x), so the value of A is usually set to 0.5.

[0075] ,

[0076] Therefore, the sample calculation The PCBC-SFD fusion model of the actual light intensity distribution is as follows: is the diffuse reflectance of the AC component under the projected fringe group with a spatial frequency of k calculated using SFDI, The pattern calculated by PCBC, To calculate the diffuse reflectance of the DC component represented by the frequency 0 using SFDI, is the diffuse reflectivity of the standard white plate:

[0077] .

[0078] Step 4: Set the sample parameters wavelength λ and absorption coefficient of the diffuse reflection pattern to be predicted , reduced scattering coefficient , input the PCBC-SFD fusion model to obtain the diffuse reflection simulation results corresponding to the projection pattern;

[0079] Step 5: The host computer generates 5 pictures, including a completely black picture, a completely white picture, and a selected spatial frequency.f x Three fringe patterns with different initial phases α under the ;

[0080] Step 6: Use the error diffusion method or the halftone fringe generation method to generate the continuous 8-bit fringe pattern into the corresponding 1-bit halftone fringe pattern, and the host computer controls the projector to project the fringe pattern onto the sample;

[0081] Step 7: The CCD camera synchronously collects the reflected light intensity distribution of the sample under the stripe pattern illumination mode and saves the image data.

[0082] Experimental Example 1:

[0083] like Figure 1 In this example, the samples used were plaster sculptures and polytetrafluoroethylene balls as well as fat emulsion solutions of different concentrations, such as Figure 2 As shown, distilled water and fat emulsion solution are mixed to prepare fat emulsion solutions of different volume concentrations representing samples with different optical properties, including the following steps:

[0084] Step 1, generating continuous tone projection stripes;

[0085] Step 2, generating corresponding halftone stripes using an error diffusion method according to the generated continuous tone stripes;

[0086] Step 3: Input the absorption and reduced scattering coefficients of the medium to be simulated into the PCBC-SFD fusion model to obtain the simulation results of the diffuse reflection pattern, such as Figure 3 As shown;

[0087] Step 4: Provide a reference for the experimental setup based on the simulation results.

[0088] Experimental Example 2:

[0089] The following steps are included:

[0090] Step 1: A large amount of theoretical data is obtained by combining halftones and their corresponding continuous tone stripes and inputting different absorption and reduced scattering coefficients into PCBC-SFD;

[0091] Step 2: Construct a convolutional neural network architecture and use the generated theoretical data as a data set for training to obtain a model that predicts continuous tone experimental results based on halftone data, such as Figure 4 As shown;

[0092] Step 3: The host computer controls the projector to project the fringe pattern onto the sample; the CCD camera synchronously collects the reflected light intensity distribution of the sample under the fringe pattern illumination mode and saves the image data;

[0093] Step 4: Input the collected halftone pattern into the trained network to obtain the prediction result of the corresponding continuous tone experiment, such as Figure 5 shown.

[0094] As another embodiment, this embodiment provides another technical solution:

[0095] The specific steps include:

[0096] Step 1: Input the 1-bit and 8-bit projection patterns and the data of different absorption and reduced scattering coefficient samples into the PCBC-SFD fusion model, and simulate and obtain the corresponding simulated diffuse reflection images under 1-bit and 8-bit projection under different optical characteristic samples;

[0097] Step 2. Establish a CNN network, input the 1-Bit and corresponding 8-Bit pattern datasets generated by PCBC-SFD for training, and obtain the correspondence between the two. The network structure design of CNN is specifically to start with a single-channel input layer with a size of 1080×1920 pixels, and use convolution layers with 3×3 kernels and ReLU activation functions to extract and refine the features of binary inputs. This configuration introduces nonlinearity and helps prevent gradient disappearance during training. The subsequent MaxPooling2D operation reduces the dimension of the feature map and enhances the generalization ability of the network. In order to capture a wider range of patterns, a convolution layer with an 11×11 kernel is introduced. The UpSampling2D layer combines additional convolutions to upscale the feature map to reconstruct the 8-bit pattern. The network ends with a final convolution layer that uses a sigmoid activation function to normalize the pixel values ​​between 0 and 1, effectively simulating an 8-bit pattern;

[0098] Step 3, projecting 1-bit and 8-bit patterns, and the CCD camera synchronously collects the reflected light intensity distribution of the sample under the stripe pattern illumination mode, and saves the image data;

[0099] Step 4: Input 1-bit experimental data and use CNN network to obtain the corresponding 8-bit pattern prediction results;

[0100] This embodiment analyzes the advantages and disadvantages of PCBC and SFDI in simulating diffuse reflection patterns in spatial frequency domain imaging, and integrates the two through theoretical calculations, and proposes a new method that can simulate stripe diffuse reflection patterns in a large field of view, high precision, and real time, and combines CNN to achieve the corresponding 8-Bit results can still be well predicted based on 1-Bit experiments when the sample optical properties fluctuate greatly, the height fluctuates greatly, and the degree of defocus is different. This is beneficial to subsequent online detection in actual industries.

[0101] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for constructing a simulation model of structured light diffuse reflection pattern in spatial frequency domain imaging, characterized in that: The specific steps include: Step 1, calculate the arbitrary diffuse reflectance distribution of the sample using the partial current boundary condition (PCBC) formula described by Farrell and Patterson; The diffuse reflection distribution of the sample is calculated using the partial current boundary condition formula described by Farrell and Patterson, as follows: ， ， in, r is the radial distance to the light source, is the distance to the light source, is the distance to the image source, is the effective attenuation coefficient, is the reduced scattering coefficient; , , , , , , is the internal reflection parameter, is a parameter related to the tissue refractive index, ,and is the transmission albedo, is the reflectivity at a distance r from the light source, is the sample diffuse reflectance, is the extinction coefficient; Step 2, in spatial frequency domain imaging (SFDI), the fringe spatial frequency Diffuse reflectance based on optical properties The calculation formula is as follows: ， In the formula, is the effective attenuation coefficient, is the extinction coefficient, is the internal reflection parameter, is the sample diffuse reflectance, is the transmission albedo; Step 3, simulating the striped diffuse reflection pattern and calculating and establishing a PCBC-SFD fusion model; Among them, the PCBC-SFD fusion model is as follows: ， In the formula, is the diffuse reflectance of the AC component under the projected fringe group with a spatial frequency of k calculated using SFDI, The pattern calculated by PCBC, To calculate the diffuse reflectance of the DC component represented by the frequency 0 using SFDI, is the diffuse reflectivity of the standard white plate: Step 4: Set the sample parameters wavelength λ and absorption coefficient of the diffuse reflection pattern to be predicted , reduced scattering coefficient , input the PCBC-SFD fusion model to obtain the diffuse reflection simulation results corresponding to the projection pattern; Step 5: The host computer generates 5 pictures, including a completely black picture, a completely white picture, and a selected spatial frequency. f x Three fringe patterns with different initial phases α under the ; Step 6: Use the halftone fringe generation method to generate the continuous 8-bit fringe pattern into the corresponding 1-bit halftone fringe pattern, and the host computer controls the projector to project the fringe pattern onto the sample; Step 7: The CCD camera synchronously collects the reflected light intensity distribution of the sample under the stripe pattern illumination mode and saves the image data.

2. The method for constructing a simulation model of structured light diffuse reflection pattern in spatial frequency domain imaging according to claim 1, characterized in that: in, In step 3, the PCBC-SFD fusion model calculation process is as follows: First, the calculation formula for the projection pattern light intensity distribution is as follows: is the amplitude envelope of the DC component, is the amplitude envelope of the AC component, is the intensity distribution of the three phase corresponding patterns, and the calculation formula is as follows: ， ， The diffuse reflectance of the sample is determined by the ratio with the standard reflector , For two different samples, the corresponding relationship between them can be established ， ， ， Assignment: , The light intensity distribution of the projection, standard panel, and sample can be expressed as: ， ， ， The light intensity distribution of the sample diffuse reflection pattern and the sample , And the relationship between the standard plate and the projection pattern strength is as follows: ， The intensity of the projected fringe is expressed as 0.5+0.5cos(x), so the value of A is set to 0.

5. 。

Images