A method for measuring the diffuse reflectance of spherical fruit surfaces in spatial frequency domain imaging

Abstract

The present application relates to a kind of spherical fruit surface diffuse reflectance correction measurement method in spatial frequency domain imaging, solve when measuring the diffuse reflection of spherical fruit in spatial frequency domain imaging system is influenced by sample surface height, reflection angle and light incidence angle.This method includes the following steps: one, projection obtains structured light fringe diagram;Two, fringe diagram whiteboard correction;Three, obtain fruit diffuse reflectance and three-dimensional profile;Four, height correction is carried out to diffuse reflectance, and the reflection angle of spherical fruit surface is obtained from height;Five, angle correction is carried out to the diffuse reflectance after height correction, obtain the final diffuse reflectance of spherical fruit;Six, the optical property of diffuse reflectance is reversed to obtain absorption coefficient and reduced scattering coefficient.This method proposes a kind of diffuse reflection correction method suitable for spherical fruit, corrects the diffuse reflectance measurement error caused by fruit surface height and angle, and the diffuse reflectance of corrected fruit and its reversed optical property are more in line with the characteristics of fruit texture uniformity.

Description

Technical Field

[0001] This invention relates to the field of optical detection of biological tissues, and specifically to a method for correcting and measuring the diffuse reflectance of the surface of a spherical fruit in spatial frequency domain imaging. Background Technology

[0002] Rapid non-destructive testing (RDBMT) technology offers significant advantages and wide applicability for online detection of fruit defects. Researchers have investigated several imaging modalities, such as broadband, multispectral, and hyperspectral imaging, for detecting surface defects. However, detecting subsurface defects faces challenges due to limitations of visual observation and the similarity between defective surfaces and surrounding healthy tissue. Spatial frequency domain imaging (SFDI) is a non-contact, wide-field imaging technique where low-frequency and high-frequency illumination penetrate deep and shallow biological tissues, respectively, attributed to the tissue acting as a low-pass filter. Therefore, SFDI can be used to assess subsurface defects, i.e., early bruising occurring within half an hour, by quantifying the optical properties of fruit tissue (absorption coefficient μa and reduced scattering coefficient μ'). s SFDI is widely used in medical and agricultural engineering due to its non-contact properties, high specificity, and abundant measurement data. However, SFDI is sensitive to surface profiles. Researchers report that demodulating diffuse reflectance (R0) is crucial. d The surface profile of non-flat samples (such as animal tumors and spherical fruits) is strongly influenced by the surface contour. Therefore, one of the main challenges of using SFDI to detect subsurface defects is that SFDI cannot adjust the R-value according to the height of the image surface. d .

[0003] Recent studies have introduced methods to minimize this effect. In 2009, Giovux et al. proposed a correction method based on surface profile and a Lambertian reflection model. Since the projection of structured light onto an object is the basis of phase profilometry (PMP) and SFDI, this method combines the two techniques on the same instrument to obtain the three-dimensional profile of the object and simultaneously extract its optical properties. Finally, the correction model was applied to human fingers to reduce edge imaging artifacts with surface angles up to 40° (Gioux S, Mazhar A, Cuccia DJ, et al. Three-dimensional surface profile intensity correction for spatially modulated imaging[J]. Journal of biomedical optics, 2009, 14(3): 034045.). In 2012, Nguyen et al. proposed a correction method based on a 3D model. This method uses precise 3D imaging and 3D printing technology to construct a three-dimensional model using acrylonitrile-butadiene-styrene (ABS) plastic. The correction method can detect and improve curvature effects on surfaces with slopes of 30 to 80 degrees and height variations greater than 4 mm, and can correct complex structures such as the eyeball and mice (Nguyen TT, Le HN, Vo M, et al. Three-dimensional phantoms for curvature correction in spatial frequency domain imaging[J]. Biomedical optics express, 2012, 3(6): 1200-1214.). In 2016, Zhao et al. adapted the Minnaert correction method of satellite images to propose an improved Lambert correction (MLC), and compared and analyzed the correction effects of uncorrected, Lambert corrected, and MLC on a hemispherical model. The results showed that MLC had the best correction effect, correcting the tissue surface angle to 75°. In the test of small animal tumor models, the corrected tumor μ aThe total hemoglobin value is physiologically reasonable (Zhao Y, Tabassum S, Piracha S, et al. Angle correction for small animal tumor imaging with spatial frequency domain imaging (SFDI)[J]. Biomedical opticsexpress,2016,7(6):2373-2384.). In 2021, He et al. proposed a profile-based diffuse reflectance correction method, used Monte Carlo (MC) simulation to determine the Rd correction formula, and applied the proposed Rd correction method to the detection of pear bruising (He X, Hu D, Fu X, et al. Spatial frequency domain imaging for determining absorption and scattering properties of bruised pears based on profilecorrected diffused reflectance[J]. Postharvest Biology and Technology,2021,179:111570.).

[0004] The correction methods proposed above can be divided into two types based on the correction model: correction based on physical 3D models and correction based on PMP profile measurement. The former, however, involves very time-consuming 3D model creation; therefore, most researchers adopt correction methods based on PMP profile measurement. Profile-based correction is based on the Lambertian model, and accurately calculating the surface angles of the object is crucial for Lambertian correction. Currently, SFDI correction is mostly applied in the medical field, with very few methods proposed for fruit detection. However, the surface curvature of fruit has a significant impact on the optical properties of SFDI detection. Summary of the Invention

[0005] The purpose of this invention is to solve the problem that when measuring the diffuse reflectance of spherical fruits in a spatial frequency domain imaging system, the surface height, reflection angle, and incident angle of the light are all factors that affect the calculation of the surface reflection angle when performing angle correction on the sample. This invention provides a method for correcting and measuring the diffuse reflectance of spherical fruits in spatial frequency domain imaging, so as to obtain a more accurate diffuse reflectance of spherical fruits, which can be used to invert the optical properties of tissues and evaluate their quality.

[0006] The technical solution adopted by this invention to solve its technical problem is: a method for correcting and measuring the diffuse reflectance of the surface of a spherical fruit in spatial frequency domain imaging, characterized by including the following steps:

[0007] Step 1: The host computer generates 26 images, including one completely black image, one completely white image, and 8 spatial frequencies f. n =0, 0.01, 0.05, 0.1, 0.15, 0.20, 0.25, 0.3mm -1 Below, there are three fringe patterns with different initial phase α at each spatial frequency, among which... The host computer controls the projector to project a stripe pattern onto the fruit to obtain a diffuse reflection image of the fruit;

[0008] Step 2: Project each fringe pattern sequentially onto a standard diffuse reflector plate. Use a CCD camera to capture the intensity distribution I of the reference light field on the standard diffuse reflector plate, and perform whiteboard correction on the fringe patterns. Whiteboard correction is achieved using Equation 1:

[0009]

[0010] Among them, I corrected I0 is the corrected relative image intensity, and I0 is the original image intensity. dark I represents the intensity of the dark field image. white The image intensity is used as a reference whiteboard.

[0011] Step 3: Demodulate the modulated images of the fruit from Step 1 and the standard diffuse reflector from Step 2 using a three-phase demodulation algorithm to obtain different spatial frequencies f. n The demodulated image M AC (f n ) and M AC,ref (f n To correct system response and eliminate system errors, different frequencies f are obtained. n Diffuse reflection image of the fruit below R initial (f n The three-dimensional profile of the fruit's surface height h was obtained using phase profilometry.

[0012] Step 4: Perform height correction based on the obtained height map and diffuse reflection map of the fruit to obtain the corrected diffuse reflection map R. height (f n Eliminate the variation in diffuse reflectance caused by different surface heights; determine the surface reflection angle diagram based on the fruit's height diagram;

[0013] Step 5: Based on the improved Lambert correction principle, considering the incident angle of light, the measurement area is divided into four equal-sized regions in a cross shape. To minimize the influence of the incident angle on the measurement results, the diffuse reflectance after height correction is angularly corrected to eliminate errors caused by changes in surface curvature. Finally, the corrected diffuse reflectance R is obtained. final (f n );

[0014] Step 6: Use a reverse algorithm to solve for the absorption coefficient μ of the fruit tissue. a Reduced scattering coefficient μ' s .

[0015] As a preferred option, in step three: the three-phase demodulation algorithm is implemented using equation two:

[0016]

[0017] Among them, M AC (f n ) is the demodulated diffuse reflection image, I1(f n ), I2(f n ) and I3(f n The initial phases after whiteboard correction are shown below. The relative image intensity below;

[0018] The system response is achieved through Equation 3:

[0019]

[0020] Among them, M AC,ref (f n ) and M AC (f n M represents the diffuse reflectance images of the reference sample and the test sample after three-phase demodulation. AC,ref (f n R is obtained through equation two. ref (f n ) and R inital (f n R represents the systematically corrected diffuse reflectance images of the reference sample and the sample to be tested; ,ref (f n =0.97 is the diffuse reflectance value of a standard whiteboard.

[0021] As a preferred option, the phase profilometry in step three specifically involves: using a three-step phase shifting method on three fringe patterns at one frequency to obtain a truncated phase; using the branching method to expand the truncated phase to obtain the expanded phase; and finally, based on the principle of triangulation, measuring the expanded phase of a standard diffuse reflector every 2 mm within a height range of 0–50 mm, and performing linear fitting between the height and phase to obtain a height calculation formula; substituting the expanded phase of the fruit into the formula yields the three-dimensional profile h of the fruit.

[0022] The three-step phase shift algorithm is implemented using Equation 4:

[0023]

[0024] in, The truncated phase is the obtained phase, where I1, I2, and I3 are the initial phases after whiteboard correction at a single frequency. The relative image intensity is shown below; the branch cutting method is used to... Phase expansion is performed to obtain the expanded phase.

[0025] Phase profiling is achieved through Formula 5:

[0026]

[0027] Where h is the surface height, and the values ​​of x and y are determined by the system hardware and obtained through calibration during the experiment. It is the difference between the unfolded phase of the desired height position and the unfolded phase of the reference plane h=0.

[0028] Preferably, in step 4, the diffuse reflection value of the standard diffuse reflection plate is measured every 2 mm within the height range of 0–50 mm, and a linear fit is performed on the height and diffuse reflection value to obtain the height correction formula for the diffuse reflection value; the parameters of the diffuse reflection height correction formula are achieved through the following equations six and seven:

[0029] ΔR(f n )=a(f n )·h+b(f n )

[0030] R height (f n ) = R initial (f n )+ΔR(f n )

[0031] Among them, R initial (f n ), R height (f n ) are the diffuse reflection images before and after height correction, respectively, ΔR(f n ) is the diffuse reflection deviation caused by altitude, a(f n b(f) n The slope and intercept are obtained from the calibration, respectively; then the cosine value of the angle on the fruit surface is obtained, which is achieved through Equation 8:

[0032]

[0033] Where θ is the angle between the reflected ray and the camera's optical axis, i.e., the reflection angle of the fruit surface; h and r are the height and radius of the sample surface, respectively, and the maximum value h is obtained by phase profilometry. max Using the diameter of the fruit as an example, the radius of the fruit is calculated to be r = h. max / 2.

[0034] Preferably, in step 5, the correction coefficient is calculated using Equation 9:

[0035]

[0036] Where, k j (j = 1, 2, 3, 4) are the correction coefficients for the four regions, and i and z are the pixel index and the total number of pixels in the measurement region, respectively; R height,ref (f n ) j It is the average value of diffuse reflection in the low-angle region of the measurement area, where the reflection angle is less than 10°; the demodulation values ​​in the low-angle region are almost uncorrected, so their average value is used as the "gold standard" to calculate the k value with the minimum variance.

[0037] The diffuse reflection angle correction employs a modified Lambert correction, achieved through Equation 10:

[0038]

[0039] Among them, R final (f n R is the diffuse reflection map after angle correction. height (f n 1. R height (f n )2、R height (f n 3. R height (f n )4 is the height-corrected diffuse reflection map of the four regions.

[0040] Preferably, in step 6, nonlinear curve fitting based on diffuse approximation theory is used to solve for μ of the sublayer tissue of the sample under test. a With μ' s The analytical solution of the diffuse approximation theory is shown in Equation 11:

[0041]

[0042] Where A is a constant with respect to the refractive index of the tissue, μ tr =μ a +μ′ s ,μ' eff =(3μ a μ tr +(2πf n ) 2 ) 1 / 2 Equation 11 forms two or more analytical expressions based on different frequencies, yielding the fitted solution μ. a 、μ' s The optical properties of the fruit to be tested.

[0043] This invention combines phase profilometry and spatial frequency domain imaging technology. Using the proposed method for calculating the surface reflection angle, it can calculate the surface height and reflection angle of spherical fruits and correct for diffuse reflection variations caused by the surface height and reflection angle. Furthermore, considering that the incident angle of light divides the measurement area into four equal regions, it minimizes the influence of the incident angle on the measurement results. The corrected diffuse reflection image and the optical characteristic image retrieved from it eliminate errors caused by height and angle, better reflecting the uniformity of the fruit's texture and ensuring more accurate evaluation results in subsequent work. Attached Figure Description

[0044] The invention will be further described below with reference to the accompanying drawings.

[0045] Figure 1 This is a schematic diagram of one implementation process of the present invention.

[0046] Figure 2 The apple surface profile is calculated using phase profilometry in this invention.

[0047] Figure 3 This invention provides the surface reflection angle obtained by measuring the height of the apple's surface.

[0048] Figure 4 From left to right is the invention f n =0.15mm -1 The original diffuse reflection image of the apple, the height-corrected diffuse reflection image, and the angle-corrected diffuse reflection image.

[0049] Figure 5 Comparison of absorption coefficient μa before and after correction in this invention.

[0050] Figure 6 A comparison of the reduced scattering coefficient μ's before and after correction in this invention. Detailed Implementation

[0051] The present invention will be further described below with reference to specific embodiments and accompanying drawings.

[0052] Example: A method for correcting and measuring the diffuse reflectance of a spherical fruit surface in spatial frequency domain imaging, such as... Figure 1 As shown. In this example, apples are used as the fruit, and the specific steps are as follows:

[0053] Step 1: The host computer generates 26 images: one completely black image, one completely white image, and three initial phase images for each spatial frequency. There are 8 different fringe patterns with a total of 8 spatial frequencies f. n (0, 0.01, 0.05, 0.1, 0.15, 0.20, 0.25, 0.3mm -1), 24 stripe patterns. The host computer controls the projector to project the stripe patterns onto the apple.

[0054] Step 2: Project the fringe pattern onto a standard diffuse plate (reflectivity 0.97), and use a CCD camera to collect the distribution I of the reference light field illumination intensity on the standard diffuse plate to perform whiteboard correction (uniformity correction) on the fringe pattern.

[0055] Whiteboard calibration is achieved through formula one:

[0056]

[0057] Among them, I corrected I0 is the corrected relative image intensity, and I0 is the original image intensity. dark I represents the intensity of the dark field image. white The image intensity is used as a reference whiteboard.

[0058] Step 3: Demodulate the modulated images of the apple from Step 1 and the standard diffuse reflector from Step 2 using a three-phase demodulation algorithm to obtain different spatial frequencies f. n The demodulated image M AC (f n ) and M AC,ref (f n A reference sample (standard diffuse reflector) and a test sample (apple) are set up. System response correction is performed to eliminate system errors and obtain different frequencies f. n The diffuse reflection image of the apple below R initial (f n Phase profilometry: A truncated phase is obtained by using a three-step phase shift on three fringe patterns at one frequency. The truncated phase is then expanded using a branching method to obtain the expanded phase. Finally, based on the principle of triangulation, the expanded phase of a standard diffuse reflector is measured every 2 mm within a height range of 0–50 mm. A linear fit is then performed on the height and phase to obtain the height calculation formula. The expanded phase of the apple is then substituted into the formula to obtain the three-dimensional profile h of the apple.

[0059] The three-phase demodulation algorithm is implemented using Equation 2:

[0060]

[0061] Among them, M AC (f n ) is the demodulated diffuse reflection image, I1(f n ), I2(f n ) and I3(f n The initial phases after whiteboard correction are shown below. The relative image intensity.

[0062] The system response is achieved through Equation 3:

[0063]

[0064] Among them, M AC,ref (f n ) and M AC (f n M represents the diffuse reflectance images of the reference sample and the test sample after three-phase demodulation. AC,ref (f n R is obtained through equation two. ref (f n ) and R inital (f n (R) represents the systematically corrected diffuse reflectance images of the reference sample and the sample to be tested. ,ref (f n =0.97 is the diffuse reflectance value of a standard whiteboard.

[0065] The three-step phase shift algorithm is implemented using Equation 4:

[0066]

[0067] in, These are the cutoff phases obtained, where I1, I2, and I3 are the frequencies at 0.15 mm. -1 Initial phase after lower whiteboard correction The relative image intensity is determined by the branching method. Phase expansion is performed to obtain the expanded phase.

[0068] Phase profiling is achieved through Formula 5:

[0069]

[0070] Among them, such as Figure 2 As shown, h is the surface height, and the values ​​of x and y are determined by the system hardware and obtained through calibration during the experiment. It is the difference between the unfolded phase of the desired height position and the unfolded phase of the reference plane (h=0).

[0071] Step 4: Within a height range of 0–50 mm, measure the diffuse reflection value of the standard diffuse reflector every 2 mm, and perform linear fitting on the height and diffuse reflection values ​​to obtain the height correction formula for the diffuse reflection value. Based on the obtained apple height map and diffuse reflection map, perform height correction to obtain the corrected diffuse reflection map R. height (f n This eliminates the variation in diffuse reflectance caused by different surface heights. The surface reflection angle diagram is derived from the apple's height diagram.

[0072] The parameters of the diffuse reflection height correction formula are achieved through equations six and seven: ΔR(fn )=a(f n )·h+b(f n )

[0073] R height (f n ) = R initial (f n )+ΔR(f n )

[0074] Among them, R initial (f n ), R height (f n ) are the diffuse reflection images before and after height correction, respectively, ΔR(f n ) is the diffuse reflection deviation caused by altitude, a(f n b(f) n ) are the slope and intercept obtained from calibration, respectively.

[0075] The cosine value of the surface angle is achieved through Equation 8:

[0076]

[0077] Where θ is the angle between the reflected ray and the camera's optical axis, i.e., the reflection angle of the apple's surface, such as... Figure 3 As shown. h and r are the surface height and radius of the sample to be tested, respectively. The maximum value h obtained by phase profilometry is used. max Using the highest point of the apple's outline as the apple's diameter, the apple's radius is calculated to be r = h. max / 2.

[0078] Step 5: Based on the improved Lambert correction principle, considering the incident angle of light, the measurement area is divided into four equal-sized regions in a cross shape. To minimize the influence of the incident angle on the measurement results, the diffuse reflectance after height correction is angularly corrected to eliminate errors caused by changes in surface curvature. Finally, the corrected diffuse reflectance R is obtained. final (f n ).

[0079] The correction coefficient k is achieved through Equation 9:

[0080]

[0081] Where, k j (j = 1, 2, 3, 4) are the correction coefficients in the four regions, and i and z are the pixel index and the total number of pixels in the measurement region, respectively. R height,ref (f n ) jIt is the average of the diffuse reflectance values ​​of the low-angle region (less than 10°) within the measurement area. The demodulated values ​​of the low-angle region are almost uncorrected, so their average is used as the "gold standard" to calculate the k value with the minimum variance.

[0082] Diffuse reflection angle correction (improved Lambert correction) is achieved through Equation 10:

[0083]

[0084] Among them, R final (f n R is the diffuse reflection map after angle correction. height (f n 1. R height (f n )2、R height (f n 3. R height (f n )4 is the diffuse reflection map after height correction for the four regions.

[0085] Step 6: Solve for the absorption coefficient μ of apple tissue using a reverse algorithm. a Reduced scattering coefficient μ' s .

[0086] The μ of the sublayer tissue of the test sample is solved by nonlinear curve fitting based on the diffuse approximation theory. a With μ' s The analytical solution of the diffuse approximation theory is shown in Equation 11:

[0087]

[0088] Where A is a constant with respect to the refractive index of the tissue, μ tr =μ a +μ′ s ,μ' eff =(3μ a μ tr +(2πf n ) 2 ) 1 / 2 Equation 11 forms two or more analytical expressions based on different frequencies, yielding the fitted solution μ. a 、μ' s As an optical property of apples.

[0089] Figure 4 From left to right is the invention f n =0.15mm -1 The standard deviations of the original diffuse reflection image, the height-corrected diffuse reflection image, and the angle-corrected diffuse reflection image of the apple are 0.0362, 0.0338, and 0.0206, respectively. Figure 5 and Figure 6 Distribution of absorption coefficient μ a The comparison chart of the reduced scattering coefficient μ's before and after correction shows that the standard deviations before correction were 0.0089 and 0.1890, and after correction they were 0.0021 and 0.1097, representing improvements of 76% and 42% respectively. The corrected optical characteristics image is more consistent with the uniform texture of apples and can more accurately evaluate the quality of the fruit pulp.

Claims

1. A method for correcting and measuring the diffuse reflectance of a spherical fruit surface in spatial frequency domain imaging, characterized in that: Includes the following steps, Step 1: The host computer generates 26 images, including one completely black image, one completely white image, and 8 spatial frequencies f. n =0, 0.01, 0.05, 0.1, 0.15, 0.20, 0.25, 0.3mm -1 Below, there are three fringe patterns with different initial phase α at each spatial frequency, among which... The host computer controls the projector to project a stripe pattern onto the fruit to obtain a diffuse reflection image of the fruit; Step 2: Project each fringe pattern sequentially onto a standard diffuse reflector plate. Use a CCD camera to capture the intensity distribution I of the reference light field on the standard diffuse reflector plate, and perform whiteboard correction on the fringe patterns. Whiteboard correction is achieved using Equation 1: Among them, I corrected I0 is the corrected relative image intensity, and I0 is the original image intensity. dark I represents the intensity of the dark field image. white The image intensity is used as a reference whiteboard. Step 3: Demodulate the modulated images of the fruit from Step 1 and the standard diffuse reflector from Step 2 using a three-phase demodulation algorithm to obtain different spatial frequencies f. n The demodulated image M AC (f n ) and M AC,ref (f n To correct system response and eliminate system errors, different frequencies f are obtained. n Diffuse reflection image of the fruit below R initial (f n The three-dimensional profile of the fruit's surface height h was obtained using phase profilometry. Step 4: Perform height correction based on the obtained height map and diffuse reflection map of the fruit to obtain the corrected diffuse reflection map R. height (f n Eliminate the variation in diffuse reflectance caused by different surface heights; determine the surface reflection angle diagram based on the fruit's height diagram; In step 4, the diffuse reflection value of the standard diffuse reflector is measured every 2 mm within the height range of 0~50 mm, and a linear fit is performed on the height and diffuse reflection value to obtain the height correction formula for the diffuse reflection value; the parameters of the diffuse reflection height correction formula are achieved through the following equations six and seven: Among them, R initial (f n ), R height (f n These are diffuse reflection images before and after height correction, respectively. The diffuse reflection deviation caused by altitude, a(f n b(f) n The slope and intercept are obtained from the calibration, respectively; then the cosine value of the angle on the fruit surface is obtained, which is achieved through Equation 8: Where θ is the angle between the reflected ray and the camera's optical axis, i.e., the reflection angle of the fruit surface; h and r are the height and radius of the sample surface, respectively, and the maximum value h is obtained by phase profilometry. max Using the diameter of the fruit as an example, the radius of the fruit is calculated to be r = h. max / 2; Step 5: Based on the improved Lambert correction principle, considering the incident angle of light, the measurement area is divided into four equal-sized regions in a cross shape. To minimize the influence of the incident angle on the measurement results, the diffuse reflectance after height correction is angularly corrected to eliminate errors caused by changes in surface curvature. Finally, the corrected diffuse reflectance R is obtained. final (f n ); In step 5, the correction coefficient is calculated using Equation 9: Where, k j (j=1, 2, 3, 4) are the correction coefficients in the four regions, and i and z are the pixel index and the total number of pixels in the measurement region, respectively; R height,ref (f n ) j It is the average value of diffuse reflection in the low-angle region of the measurement area, where the reflection angle is less than 10°; the demodulation values ​​in the low-angle region are almost uncorrected, so their average value is used as the "gold standard" to calculate the k value with the smallest variance; The diffuse reflection angle correction employs a modified Lambert correction, achieved through Equation 10: Among them, R final (f n R is the diffuse reflection map after angle correction. height (f n 1. R height (f n )2、R height (f n 3. R height (f n )4 is the height-corrected diffuse reflection map of the four regions; Step 6: Use a reverse algorithm to solve for the absorption coefficient μ of the fruit tissue. a Reduced scattering coefficient μ' s .

2. The method for correcting and measuring the diffuse reflectance of a spherical fruit surface in spatial frequency domain imaging according to claim 1, characterized in that: In step three: the three-phase demodulation algorithm is implemented using equation two: Among them, M AC (f n ) is the demodulated diffuse reflection image, I1(f n ), I2(f n ) and I3(f n The initial phases after whiteboard correction are shown below. The relative image intensity below; The system response is achieved through Equation 3: Among them, M AC,ref (f n ) and M AC (f n M represents the diffuse reflectance images of the reference sample and the test sample after three-phase demodulation. AC,ref (f n R is obtained through equation two. ref (f n ) and R inital (f n R represents the systematically corrected diffuse reflectance images of the reference sample and the sample to be tested; ,ref (f n =0.97 is the diffuse reflectance value of a standard whiteboard.

3. The method for correcting and measuring the diffuse reflectance of a spherical fruit surface in spatial frequency domain imaging according to claim 1, characterized in that: Step three, the phase profilometry, specifically involves: using a three-step phase shifting method on three fringe patterns at one frequency to obtain a truncated phase; using the branching method to expand the truncated phase to obtain the expanded phase; and finally, based on the principle of triangulation, measuring the expanded phase of a standard diffuse reflector every 2 mm within a height range of 0-50 mm, and performing linear fitting between the height and phase to obtain the height calculation formula; substituting the expanded phase of the fruit into the formula yields the three-dimensional profile h of the fruit. The three-step phase shift algorithm is implemented using Equation 4: in, The truncated phase is the obtained phase, where I1, I2, and I3 are the initial phases after whiteboard correction at a single frequency. The relative image intensity is shown below; the branch cutting method is used to... Phase expansion is performed to obtain the expanded phase. ; Phase profiling is achieved through Formula 5: Where h is the surface height, and the values ​​of x and y are determined by the system hardware and obtained through calibration during the experiment. It is the difference between the unfolded phase of the desired height position and the unfolded phase of the reference plane h=0.

4. The method for correcting and measuring the diffuse reflectance of a spherical fruit surface in spatial frequency domain imaging according to claim 1, characterized in that: In step 6, nonlinear curve fitting based on diffuse approximation theory is used to solve for μ of the sublayer tissue of the sample under test. a With μ' s The analytical solution of the diffuse approximation theory is shown in Equation 11: Where A is a constant with respect to the refractive index of the tissue, μ tr =μ a +μ′ s ,μ' eff =(3μ a μ tr +(2πf n ) 2 ) 1 / 2 Equation 11 forms two or more analytical expressions based on different frequencies, yielding the fitted solution μ. a μ' s The optical properties of the fruit to be tested.

Images