A light field phase image generation method and device based on light intensity and iterative optimization

Abstract

The application relates to the technical field of phase image generation, and discloses a light field phase image generation method based on light intensity and iterative optimization, which comprises the following steps: collecting a light intensity image and a corresponding phase image of a to-be-tested light field as a sample light intensity image and a sample phase image; generating a conversion set by using the sample light intensity image and the sample phase image; collecting a real-time light intensity image of the to-be-tested light field, and generating a corresponding real-time phase image by using the conversion set; wherein the same photosensitive equipment is used for collecting the sample light intensity image and collecting the real-time light intensity image. The application uses a common camera (a light intensity sensor or a light intensity measurer or an industrial camera) to acquire the intensity image of the light field, and can generate the phase image of the light field by using a light field phase image generation algorithm, so that the operation is easy, the calculation is simple, and the hardware cost is extremely low.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of phase imaging technology, and in particular to a light field phase image generation method and device based on light intensity and iterative optimization. BACKGROUND

[0002] Phase is an important characteristic parameter of light field information, which reflects the state of light wave at a certain position in transmission, and it has a wide range of applications in the fields of optical interference, optical imaging and optical communication. Measuring the phase information of the light beam has become an important research problem.

[0003] Quantitative phase imaging technology is generally divided into two categories. The first category is phase imaging technology based on interference technology and interferometer architecture. For example, the Michelson interferometer shown in (a) of Figure 1 and the Mach-Zehnder interferometer shown in (b) of Figure 1 These two are double-path interferometers. In addition, there are coaxial single-path interferometers, such as the Mireau interferometer shown in (c) of Figure 1 and the Fizeau interferometer shown in (d) of Figure 1 The second category of quantitative phase imaging technology is based on phase iterative algorithm without interferometer architecture. There are two main technical routes. The first is to use a phase measurement sensor such as a Shack-Hartmann wavefront sensor, which is designed as shown in Figure 2 , (a) of Figure 2 is the imaging principle diagram; Figure 2 (b) is the detector surface diagram; Figure 2 (c) is the reconstructed phase result; because it uses a microlens array design, it discards the traditional interferometer architecture, so that the Shack-Hartmann wavefront sensor is successfully applied in the field of astronomy and wavefront phase detection technology. These two methods both rely on hardware devices and require special design (optical device design, etc.) by professional optical researchers to obtain the phase image of the light field, and the operation is complex. SUMMARY

[0004] The purpose of the present application is to solve the problems in the prior art.

[0005] The technical scheme adopted by the present application to solve its technical problems is to provide a light field phase image generation method based on light intensity and iterative optimization, comprising the following steps:

[0006] Collecting the light intensity image and the corresponding phase image of the light field to be measured as a sample light intensity image and a sample phase image;

[0007] Generating a conversion set using the sample light intensity image and the sample phase image;

[0008] Collect the real-time intensity image of the light field to be measured, and generate the corresponding real-time phase image using the transformation set;

[0009] The same photosensitive device is used for collecting the sample light intensity image and the real-time light intensity image;

[0010] The method of generating a transformation set by using the sample light intensity image and the sample phase image comprises the following steps:

[0011] The conversion step converts the sample intensity image and the sample phase image into matrices to obtain the intensity matrix I and the phase matrix ;

[0012] The calculation set generation step generates a calculation set A, which contains N random number matrices A with the same size as the light intensity matrix I. n , the range is between [0,1]; n represents the random number matrix sequence in the calculation set A;

[0013] In the source matrix selection step, three matrices A1, A2, and A3 in the calculation set A are selected as source matrices based on adaptability;

[0014] T matrix generation step: generate a three-value random number matrix T with the same size as the light intensity matrix I, and the elements in the matrix T are all 0, 0.5 or 1; calculate matrices T1, T2 and T3 based on the three-value matrix T;

[0015] In the transformation set generation step, the source matrix and matrices T1, T2, and T3 are combined to calculate the new calculation set A' = A1·T1+A2·T2+A3·T3; x values ​​in A' are selected for mutation to obtain the calculation set A'';

[0016] Judgment step, calculate the new phase matrix ´=I·A″, thus calculating the phase matrix With the new phase matrix If the difference is less than the threshold, the loop ends and A″ is output as the transformation set; otherwise, the calculation set update step is entered;

[0017] In the calculation set updating step, the adaptability of the N calculation sets in A and the N calculation sets in A″ is calculated, and based on the adaptability, the N calculation sets are selected as the new calculation set A, and the process returns to the source matrix selection step.

[0018] Preferably, the source matrix selection step includes the following steps:

[0019] Adaptability uses difference measure, The smaller it is, the greater the adaptability;

[0020] Based on adaptability, three matrices A1, A2 and A3 in the calculation set A are selected as source matrices; random number matrix An The greater the adaptability, the greater the probability of being selected.

[0021] Preferably, the matrices T1, T2 and T3 are calculated based on the ternary matrix T, specifically:

[0022] ;

[0023] ;

[0024] ;

[0025] Here, i and j represent the row index and column index of the matrix respectively.

[0026] Preferably, x values ​​in A' are selected for mutation to obtain the calculation set A", where x is a preset value within the range of:

[0027] x=[0.01*M, 0.1*M];

[0028] Wherein, M represents the number of elements in the light intensity matrix I.

[0029] Preferably, the calculation set updating step includes the following steps:

[0030] Compute the adaptability of the N computation sets in A and the N computation sets in A″, and use the difference measure, The larger it is, the smaller the adaptability is; m represents the serial number of the random number matrix in the union of the calculation sets A and A″;

[0031] Select the N computation sets with the greatest adaptability as the new computation set A;

[0032] Return to the source matrix selection step.

[0033] Preferably, the real-time light intensity image of the light field to be measured is collected, and the corresponding real-time phase image is generated using a transformation set, specifically:

[0034] Converting the collected light intensity image into a light intensity matrix;

[0035] The phase matrix is ​​obtained by performing a dot multiplication operation between the transformation set A″ and the light intensity matrix;

[0036] Convert the phase matrix to a phase image.

[0037] The present invention also provides a light field phase image generation device based on light intensity and iterative optimization, comprising:

[0038] A sample acquisition module acquires a light intensity image and a corresponding phase image of the light field to be measured as a sample light intensity image and a sample phase image;

[0039] a conversion set calculation module, which generates a conversion set by using the sample light intensity image and the sample phase image;

[0040] a phase image generation module, which acquires a real-time light intensity image of the light field to be measured, and generates a corresponding real-time phase image by using the conversion set;

[0041] wherein the same light-sensing device is used to acquire the sample light intensity image and the real-time light intensity image;

[0042] The conversion set is generated by using the sample light intensity image and the sample phase image, and includes the following steps:

[0043] a conversion step, which converts the sample light intensity image and the sample phase image into matrices to obtain a light intensity matrix I and a phase matrix ;

[0044] a calculation set generation step, which generates a calculation set A, wherein the calculation set A contains N random number matrices A n with the same size as the light intensity matrix I and ranges between 0 and 1; n represents the serial number of the random number matrix in the calculation set A;

[0045] a source matrix selection step, which selects three matrices A1, A2 and A3 in the calculation set A as source matrices based on adaptability;

[0046] a T matrix generation step, which generates a three-value random number matrix T with the same size as the light intensity matrix I, wherein the elements in the matrix T are all 0, 0.5 or 1; and the matrices T1, T2 and T3 are calculated according to the three-value matrix T;

[0047] a conversion set generation step, which calculates a new calculation set A´= A1·T1+A2·T2+A3·T3 by combining the source matrices and the matrices T1, T2 and T3; selects x values in A´ to mutate to obtain a calculation set A″;

[0048] a judgment step, which calculates a new phase matrix ´=I·A″, and then calculates the difference between the phase matrix and the new phase matrix ´; if the difference is less than a threshold value, the cycle is ended and A″ is output as the conversion set; otherwise, the calculation set updating step is entered;

[0049] the calculation set updating step, which calculates the adaptability of the N calculation sets in A and the N calculation sets in A″, selects N calculation sets as a new calculation set A based on the adaptability, and returns to the source matrix selection step.

[0050] The present application has the following beneficial effects:

[0051] (1) The present invention does not require the design of an optical path, the selection of optical components, or the design of optical components. It only needs to directly photograph the light field to obtain the intensity information of the light field, and then obtain the phase image of the light field through subsequent calculation;

[0052] (2) The present invention does not require hardware devices such as spatial filters, lenses, beam splitters, reflectors, half-wave plates or microlens arrays. It only requires a camera (light intensity sensor or light intensity meter or industrial camera) to obtain light field phase information, and the hardware cost is low.

[0053] (3) The present invention obtains the conversion set through sample calculation, which can realize all the subsequent light intensity phase conversion operations without complex calculations.

[0054] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments, but the present invention is not limited to the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 Schematic diagram of quantitative phase imaging using a classical interferometer in the prior art;

[0056] Figure 2 The phase wavefront sensing principle of Shakhadaman wavefront sensor;

[0057] Figure 3 A diagram showing the steps of a method according to an embodiment of the present invention;

[0058] Figure 4 A flowchart of transformation set calculation according to an embodiment of the present invention;

[0059] Figure 5 is a structural diagram of a device according to an embodiment of the present invention;

[0060] Figure 6 To verify the intensity image and phase image collected at the first moment in the experiment;

[0061] Figure 7 To verify the light intensity image collected at the second moment in the experiment, the measured phase image, and the calculated phase image calculated using the method of the embodiment of the present invention. DETAILED DESCRIPTION

[0062] See also Figure 3 FIG. 1 is a diagram showing steps of a method according to an embodiment of the present invention, comprising the following steps:

[0063] S301, collecting a light intensity image and a corresponding phase image of a light field to be measured as a sample light intensity image and a sample phase image;

[0064] S302, generating a transformation set using the sample intensity image and the sample phase image;

[0065] S303, collecting a real-time light intensity image of the light field to be measured, and generating a corresponding real-time phase image using a transformation set;

[0066] The sample light intensity image and the real-time light intensity image are collected using the same photosensing device, such as a light intensity sensor, a light intensity meter, or an industrial camera of the same model.

[0067] Specifically, the detailed process of S302 is shown in Figure 4 As shown, the following steps are included:

[0068] The conversion step converts the sample intensity image and the sample phase image into matrices to obtain the intensity matrix I and the phase matrix ;

[0069] The calculation set generation step generates a calculation set A, which contains N random number matrices A with the same size as the light intensity matrix I. n , the range is between [0,1];

[0070] The source matrix selection step selects three matrices A1, A2, and A3 in the calculation set A as source matrices based on adaptability; adaptability uses difference measure, The smaller the value, the greater the adaptability. Based on the adaptability, three matrices A1, A2 and A3 in the calculation set A are selected as source matrices. The random number matrix A n The greater the adaptability, the greater the probability of being selected;

[0071] The T matrix generation step generates a three-value random number matrix T with the same size as the light intensity matrix I, where all elements in the matrix T are 0, 0.5, or 1; and the matrices T1, T2, and T3 are calculated based on the three-value matrix T. Specifically:

[0072] ;

[0073] ;

[0074] ;

[0075] In the transformation set generation step, the source matrix and matrices T1, T2, and T3 are combined to calculate a new calculation set A' = A1·T1+A2·T2+A3·T3. X values ​​in A' are selected for mutation to obtain the calculation set A", where x is a preset value in the range of x = [0.01*M, 0.1*M], and M represents the number of elements in the intensity matrix I.

[0076] Judgment step, calculate the new phase matrix ´=I·A″, thus calculating the phase matrix With the new phase matrix If the difference is less than the threshold σ, the loop ends and outputs A″ as the conversion set; otherwise, it enters the calculation set update step; the threshold σ is set manually. If the calculation result needs to be more accurate, σ should be selected as a smaller value; if the calculation speed needs to be fast, σ should be selected as a larger value;

[0077] In the calculation set updating step, the adaptability of the N calculation sets in A and the N calculation sets in A″ is calculated, and based on the adaptability, the N calculation sets are selected as the new calculation set A, and the process returns to the source matrix selection step.

[0078] Specifically, the calculation set updating step includes the following steps:

[0079] Compute the adaptability of the N computation sets in A and the N computation sets in A″, and use the difference measure, The larger it is, the smaller the adaptability is; m represents the serial number of the random number matrix in the union of the calculation sets A and A″;

[0080] Select the N computation sets with the greatest adaptability as the new computation set A;

[0081] Return to the source matrix selection step.

[0082] Specifically, the step of collecting the real-time light intensity image of the light field to be measured and generating the corresponding real-time phase image using the transformation set is as follows:

[0083] Converting the collected light intensity image into a light intensity matrix;

[0084] The phase matrix is ​​obtained by performing a dot multiplication operation between the transformation set A″ and the light intensity matrix;

[0085] Convert the phase matrix to a phase image.

[0086] For details, see Figure 5 FIG. 1 is a structural diagram of a device according to an embodiment of the present invention, comprising:

[0087] The sample acquisition module 501 acquires the light intensity image and the corresponding phase image of the light field to be measured as a sample light intensity image and a sample phase image;

[0088] A transformation set calculation module 502 generates a transformation set using the sample intensity image and the sample phase image;

[0089] Phase image generation module 503, collects the real-time light intensity image of the light field to be measured, and generates the corresponding real-time phase image using the transformation set;

[0090] The step of generating a transformation set using the sample intensity image and the sample phase image comprises the following steps:

[0091] The conversion step converts the sample intensity image and the sample phase image into matrices to obtain the intensity matrix I and the phase matrix ;

[0092] The calculation set generation step generates a calculation set A, which contains N random number matrices A with the same size as the light intensity matrix I. n , the range is between [0,1];

[0093] In the source matrix selection step, three matrices A1, A2, and A3 in the calculation set A are selected as source matrices based on adaptability;

[0094] T matrix generation step: generate a three-value random number matrix T with the same size as the light intensity matrix I, and the elements in the matrix T are all 0, 0.5 or 1; calculate matrices T1, T2 and T3 based on the three-value matrix T;

[0095] In the transformation set generation step, the source matrix and matrices T1, T2, and T3 are combined to calculate the new calculation set A' = A1·T1+A2·T2+A3·T3; x values ​​in A' are selected for mutation to obtain the calculation set A'';

[0096] Judgment step, calculate the new phase matrix ´=I·A″, thus calculating the phase matrix With the new phase matrix If the difference is less than the threshold, the loop ends and A″ is output as the transformation set; otherwise, the calculation set update step is entered;

[0097] In the calculation set updating step, the adaptability of the N calculation sets in A and the N calculation sets in A″ is calculated, and based on the adaptability, the N calculation sets are selected as the new calculation set A, and the process returns to the source matrix selection step.

[0098] The embodiment of the present invention is verified by experiment. The light field adopts a light field with a laser as the light source and the light sensing device adopts a camera. At the first moment, the light intensity image and the phase image are collected as the sample light intensity image and the sample phase image. Figure 6 As shown. The sample intensity image and the sample phase image are used to calculate the transformation set B. Then, the intensity image of the light field can be collected at any time using the transformation set B to calculate the phase image at the corresponding moment. For example, at the second moment, the intensity image of the light field is collected as follows: Figure 7 As shown, the calculated phase image at the second moment is obtained by using the transformation set B and the light intensity image at the second moment. The calculated phase image at the second moment is different from the measured phase image as shown in Figure 7 As shown, it can be seen that the calculated phase image is very close to the measured phase image, and the process of obtaining the calculated phase image only requires obtaining a phase image by actual measurement at the beginning, and then the phase image can be quickly generated without the cumbersome hardware equipment (the hardware equipment only requires a camera, which is easy to operate).

[0099] It can be seen that the light field phase image generation method and device based on light intensity and iterative optimization proposed in the present invention use an ordinary camera (light intensity sensor or light intensity meter or industrial camera) to obtain the intensity image of the light field, and use the light field phase image generation algorithm to generate the phase image of the light field. It is easy to operate, simple to calculate and has extremely low hardware cost.

[0100] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A light field phase image generation method based on light intensity and iterative optimization, characterized in that: The following steps are involved: Collecting a light intensity image and a corresponding phase image of the light field to be measured as a sample light intensity image and a sample phase image; generating a transformation set using a sample intensity image and a sample phase image; Collect the real-time intensity image of the light field to be measured, and generate the corresponding real-time phase image using the transformation set; The same photosensitive device is used for collecting the sample light intensity image and the real-time light intensity image; The method of generating a transformation set by using the sample light intensity image and the sample phase image comprises the following steps: a conversion step, converting the sample intensity image and the sample phase image into matrices to obtain the intensity matrix I and the phase matrix ψ; The calculation set generation step generates a calculation set A, which contains N random number matrices A with the same size as the light intensity matrix I. n , ranging from [0,1]; n represents the random number matrix sequence number in the calculation set A; In the source matrix selection step, three matrices A1, A2, and A3 in the calculation set A are selected as source matrices based on adaptability; T matrix generation step: generate a three-value random number matrix T with the same size as the light intensity matrix I, and the elements in the matrix T are all 0, 0.5 or 1; calculate matrices T1, T2 and T3 based on the three-value matrix T; In the transformation set generation step, the source matrix and matrices T1, T2, and T3 are combined to calculate a new calculation set A′=A1·T1+A2·T2+A3·T3; x values ​​in A′ are selected for mutation to obtain the calculation set A″; In the judgment step, the new phase matrix ψ′=I·A″ is calculated, thereby calculating the difference between the phase matrix ψ and the new phase matrix ψ′. If the difference is less than the threshold, the loop ends and A″ is output as the transformation set; otherwise, the calculation set update step is entered; a calculation set updating step, calculating the adaptability of the N calculation sets in A and the N calculation sets in A″, selecting the N calculation sets as the new calculation set A based on the adaptability, and returning to the source matrix selection step; The calculation set updating step includes the following steps: Compatibility is calculated for the N computation sets in A and the N computation sets in A″. The compatibility is measured by the difference Δ2 = |I·Am-ψ|. The larger Δ2 is, the smaller the compatibility is. m represents the serial number of the random number matrix in the union of computation sets A and A″. Select the N computation sets with the greatest adaptability as the new computation set A; Return to the source matrix selection step; The real-time light intensity image of the light field to be measured is collected, and the corresponding real-time phase image is generated using the transformation set, specifically: Converting the collected light intensity image into a light intensity matrix; The phase matrix is ​​obtained by performing a dot multiplication operation between the transformation set A″ and the light intensity matrix; Convert the phase matrix to a phase image.

2. The light field phase image generation method based on light intensity and iterative optimization according to claim 1, characterized in that: The source matrix selection step comprises the following steps: Adaptability uses the difference Δ1=|I·A n -ψ|measurement, the smaller Δ1 is, the greater the adaptability is; Based on adaptability, three matrices A1, A2 and A3 in the calculation set A are selected as source matrices; random number matrix A n The greater the adaptability, the greater the probability of being selected.

3. The light field phase image generation method based on light intensity and iterative optimization according to claim 1, characterized in that: The matrices T1, T2 and T3 are calculated based on the ternary matrix T, specifically: Here, i and j represent the row index and column index of the matrix respectively.

4. The light field phase image generation method based on light intensity and iterative optimization according to claim 1, characterized in that: The calculation set A″ is obtained by selecting x values ​​in A′ for mutation, where x is a preset value within the range of: x = [0.01*M, 0.1*M]; Wherein, M represents the number of elements in the light intensity matrix I.

5. A light field phase image generation device based on light intensity and iterative optimization, used to implement the light field phase image generation method based on light intensity and iterative optimization according to any one of claims 1 to 4, characterized in that: include: A sample acquisition module acquires a light intensity image and a corresponding phase image of the light field to be measured as a sample light intensity image and a sample phase image; A transformation set calculation module generates a transformation set using a sample intensity image and a sample phase image; The phase image generation module collects the real-time light intensity image of the light field to be measured and generates the corresponding real-time phase image using the transformation set; The same photosensitive device is used for collecting the sample light intensity image and the real-time light intensity image; The method of generating a transformation set by using the sample light intensity image and the sample phase image comprises the following steps: a conversion step, converting the sample intensity image and the sample phase image into matrices to obtain the intensity matrix I and the phase matrix ψ; The calculation set generation step generates a calculation set A, which contains N random number matrices A with the same size as the light intensity matrix I. n , ranging from [0,1]; n represents the random number matrix sequence number in the calculation set A; In the source matrix selection step, three matrices A1, A2, and A3 in the calculation set A are selected as source matrices based on adaptability; T matrix generation step: generate a three-value random number matrix T with the same size as the light intensity matrix I, and the elements in the matrix T are all 0, 0.5 or 1; calculate matrices T1, T2 and T3 based on the three-value matrix T; In the transformation set generation step, the source matrix and matrices T1, T2, and T3 are combined to calculate a new calculation set A′=A1·T1+A2·T2+A3·T3; x values ​​in A′ are selected for mutation to obtain the calculation set A″; In the judgment step, the new phase matrix ψ′=I·A″ is calculated, thereby calculating the difference between the phase matrix ψ and the new phase matrix ψ′. If the difference is less than the threshold, the loop ends and A″ is output as the transformation set; otherwise, the calculation set update step is entered; In the calculation set updating step, the adaptability of the N calculation sets in A and the N calculation sets in A″ is calculated, and based on the adaptability, the N calculation sets are selected as the new calculation set A, and the process returns to the source matrix selection step.

Images