CN107179058B - The two step phase shift algorithms based on the optimization of structure optical contrast ratio - Google Patents

Abstract

A kind of two step phase shift algorithms based on the optimization of structure optical contrast ratio, optimize two processes including two step phase shift phase calculations and contrast, by initial phase calculating process, reflectivity distribution is first acquired by two spoke line diagram datas, two solutions of wrapped phase are calculated again, by the differentiation of solution, constructs initial phase and be distributed and mark sensitive pixels point；Linear character based on absolute phase obtains structure optical contrast ratio more accurate at mark point by the combined optimization of phase and contrast；Revised structure optical contrast ratio is subjected to global diffusion, and carries out phase calculation again to obtain more accurate phase information.This method brings the influence of environment light in the variation of structure optical contrast ratio into, simplify the unknown quantity in phase shift, reduce throwing and adopts the time required for phase shifted images, obtain more accurate phase information, contrast optimization is only carried out in local neighborhood, has taken into account the global nonuniformity and computational efficiency of contrast.

Description

Two-step phase shift algorithm based on structured light contrast optimization

technical field

The invention relates to a two-step phase shift algorithm based on structured light contrast optimization for three-dimensional imaging and measurement technology, and belongs to the field of three-dimensional imaging and measurement.

Background technique

In recent years, phase-assisted structured light 3D imaging technology has been widely used in many fields such as industrial production, culture and art, medical imaging, and virtual reality due to its advantages of non-contact, high precision, and easy implementation. Among them, the phase information, as the characteristic information in 3D imaging, provides unique constraints for matching the corresponding points of the same object point in different images. Obtaining the phase information of the object surface quickly and accurately is the key to this kind of 3D imaging technology.

The phase shift algorithm is a classic algorithm for phase reconstruction. By projecting to the target and collecting multiple fringe images with the same frequency and different phase shifts, the phase information of the target surface can be obtained. The traditional phase-shift algorithm needs to project at least three phase-shift fringe patterns to achieve phase reconstruction. However, the acquisition of multiple phase shift images requires a certain amount of image acquisition time.

Therefore, the present invention attempts to reconstruct phase information through two phase-shifted fringe patterns, so as to shorten the time of phase reconstruction. In order to make up for the decrease of the phase reconstruction accuracy caused by the reduction of the phase shift fringe pattern, the present invention introduces a contrast optimization algorithm to reduce the phase reconstruction error.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a two-step phase-shift algorithm based on structured light contrast optimization, so as to reduce the number of phase-shifted images that are cast and collected, shorten the phase reconstruction time, and obtain more accurate phase information.

The two-step phase shift algorithm based on the structured light contrast optimization of the present invention includes two processes of two-step phase shift phase calculation and contrast optimization:

(1) Two-step phase-shift phase calculation process, which processes the collected two phase-shift fringe images and calculates the phase information on the corresponding pixel points; including the following steps:

① Filter the collected images to reduce the influence of noise;

② The background light intensity is represented by the contrast change, and the light intensity distribution model about the two-step phase shift is obtained. From the light intensity distribution model of the two-step phase shift combined with the square sum relationship of the trigonometric function, the one-dimensional quadratic about the surface reflectance R of the object is obtained. equation, solve the equation to obtain two solutions for the reflectivity R;

③ Substitute the solution of the obtained reflectance R back into the light intensity distribution model, and obtain the information about the folding phase the two

④According to the characteristic of absolute phase increment, filter out the incremental part from the two folded phase solutions to form the initial solution of the folded phase, and mark the contrast-sensitive pixels at the same time;

(2) The contrast optimization process is to select the pixels in the area sensitive to the contrast of structured light according to the linear increase of the absolute phase, make linear assumptions in its neighborhood, and achieve the modified contrast through the joint optimization of the contrast and the phase. Purpose; specifically includes the following steps:

①According to the linear characteristic of absolute phase, make a linear assumption on the phase value on the horizontal neighborhood of the marked pixel point, take 5-7 points on the horizontal neighborhood of the marked pixel point, and distribute the light intensity of these 5-7 points Simultaneous model, through the joint optimization of contrast and absolute phase, to achieve the purpose of optimizing regional contrast;

②According to the characteristics of similar contrast between adjacent pixels, the whole is interpolated and filled by the contrast of the local area, and the optimized contrast distribution of structured light in the entire projection range is obtained.

In the contrast optimization process, the optimized contrast distribution is substituted into the light intensity distribution model in the two-step phase shift phase calculation process, and then through steps ②-④ in the phase calculation process, more accurate reflectance distribution and phase information are obtained. .

The phase difference between the two phase-shift fringe patterns is 3π/2.

The light intensity distribution model of the two-step phase shift is:

I ₁ (x, y)=R(x, y){1+A _m cos[φ(x, y)]},

I ₂ (x, y)=R(x, y){1+A _m sin[φ(x, y)]},

where R is the reflectivity of the object surface, _Am is the initial contrast of the projected fringes, and φ(x, y) is the phase value of the corresponding pixel on the image.

The one-dimensional quadratic equation of the reflectivity R is:

(2-A _m ² )R ² -2(I ₁ +I ₂ )R+I ₁ ² +I ₂ ² =0,

_Am is the initial contrast of the projected fringes, and I ₁ and I ₂ are the light intensity distributions of the two collected images.

The expression of the folding phase is: I ₁ and I ₂ are the light intensity distributions of the two collected images, and R ₁ and R ₂ are the two solutions of the quadratic equation of the reflectivity R.

In the step 4. of the described two-step phase shift phase calculation process, the process of distinguishing and marking the pixel point is:

①Calculate the first-order difference of the two solutions in the horizontal direction to obtain the position matrix of the incremental region

②Calculate the discriminant matrix of marked pixels

③ If P=1, the position is a reliable pixel point, corresponding to two solutions, and the position matrix is If P=2, the position is a sensitive pixel, and the corresponding position matrix is S;

④ Determine the pixel at the corresponding pixel of the S matrix The part smaller than π is denoted as S ₁ , and the part larger than π is denoted as S ₂ ;

⑤ Calculate the initial estimate of the folding phase:

The present invention incorporates the influence of ambient light into the change of structured light contrast, constructs a two-step phase-shifted light intensity distribution model with phase and surface reflectivity variables to be determined, and obtains the initial distribution of the phase. Based on the linear feature of absolute phase, the regional phase information and structured light contrast are jointly optimized to obtain more accurate structured light contrast. Apply the corrected structured light contrast to obtain more precise phase information.

The present invention has the following characteristics:

(1) Compared with the traditional phase shift algorithm (no less than three steps), it can reduce the time required for the acquisition of phase shift images and achieve faster phase reconstruction;

(2) Incorporate the influence of background light into the change of contrast, and obtain more accurate phase information by optimizing and correcting the distribution of contrast;

(3) Contrast optimization is only performed in local neighborhoods, taking into account the global inconsistency of contrast and computational efficiency.

Description of drawings

Fig. 1 is the overall flow chart of the algorithm of the present invention.

FIG. 2 is a flow chart of phase calculation in the present invention.

FIG. 3 is a flow chart of contrast optimization in the present invention.

Figure 4 is two solution profiles of the folding phase in the present invention.

FIG. 5 is a schematic diagram of a region of contrast optimization in the present invention.

Detailed ways

FIG. 1 shows the overall flow chart of the two-step phase shift algorithm based on structured light contrast optimization of the present invention, including the flow chart of the phase calculation process and the flow chart of contrast optimization. The whole process starts from the collected data, goes through the phase calculation process, the contrast optimization process, and then goes through a phase calculation process to obtain more accurate phase information.

1. The process of phase calculation is shown in Figure 2. According to the physical process of fringe mining, a two-step phase shift light intensity distribution model is established, and then the reflectivity distribution and folding phase distribution of the surface of the object are calculated. is 3π/2, the square relationship of the trigonometric function is introduced as an auxiliary condition for solving the folded phase, and the quadratic equation about the folded phase is obtained. Contrast-sensitive areas.

The specific implementation process is as follows:

(1) Filter the collected image to reduce the influence of noise.

In this example, TI's DLP LightCrafter 4500 is used to project two sinusoidal fringes with a phase difference of 3π/2, a contrast setting of 0.8, a frequency of 1/36, and a resolution of 912×1140, which are captured by a Basler acA1300-60gm camera. image with a resolution of 1280×1024 and a signal-to-noise ratio of 39.8db. A two-step phase-shifted image of a white diffuse plate was taken in a dark environment.

(2) The background light intensity is represented by the contrast change, and the light intensity distribution model for the two-step phase shift is obtained:

I ₁ (x, y)=R(x, y){1+A _m cos[φ(x, y)]}

I ₂ (x, y)=R(x, y){1+A _m sin[φ(x, y)]}

where I ₁ and I ₂ are the light intensity distributions of the two collected images, R is the reflectivity of the object surface, _Am is the initial contrast of the projected fringes, and φ(x, y) is the phase value of the corresponding pixel on the image. Combining the two-step phase-shifted intensity distribution model with the sum-of-squares relationship of trigonometric functions, the quadratic equation of one variable for the reflectance R is obtained:

(2-A _m ² )R ² -2(I ₁ +I ₂ )R+I ₁ ² +I ₂ ² =0

Solving this equation obtains two solutions R ₁ , R ₂ of the surface reflectance R of the object.

(3) Substitute the solution of the obtained reflectance R back into the model to obtain the information about the folding phase two solutions of and

The expression for the folded phase is: The distributions of the two solutions are obviously different, as shown in Figure 4.

(4) According to the characteristic of absolute phase increment, the incremental part is selected from the two folded phase solutions to form the initial solution of the folded phase, and the pixels that are sensitive to contrast are marked at the same time. The process of distinguishing and labeling pixels is:

①Calculate the first-order difference of the two solutions in the horizontal direction to obtain the position matrix of the incremental region

②Calculate the discriminant matrix of marked pixels

③ If P=1, the position is a reliable pixel point, corresponding to two solutions, and the position matrix is If P=2, the position is a sensitive pixel, and the corresponding position matrix is S;

④ Determine the pixel at the corresponding pixel of the S matrix The part smaller than π is denoted as S ₁ , and the part larger than π is denoted as S ₂ ;

⑤ Calculate the initial estimate of the folding phase:

2. The process of contrast optimization, as shown in Figure 3, is based on the characteristic of linear increase of absolute phase, selects the pixel points in the area sensitive to the contrast of structured light, makes linear assumptions in its neighborhood, and through the joint optimization of contrast and phase, To achieve the purpose of correcting the contrast. The algorithm incorporates the influence of the background light into the change of the contrast of the structured light, and corrects the influence of the background light by correcting the contrast. The specific implementation process is as follows:

(1) According to the linear characteristics of the phase, make a linear assumption on the phase value on the horizontal neighborhood of the marked pixel, take 5-7 points on the horizontal neighborhood of the marked pixel, and calculate the light intensity of these 5-7 points. The distribution models are combined to achieve the purpose of optimizing regional contrast through the joint optimization of contrast and phase.

In this embodiment, a neighborhood composed of 7 pixels in the horizontal direction is selected as the selection window, assuming that the linear incremental step size is Φ _t , the initial phase value of the marked pixel point is Φ _c , and the objective function is as follows:

(2) According to the characteristics of similar contrast of adjacent pixels, the whole is interpolated and filled by the contrast of the local area, and the corrected contrast distribution of the structured light on the entire projected area is obtained. In this embodiment, an algorithm of interpolation of adjacent points is used to diffuse the contrast until the projection area is filled.

3. Substitute the corrected contrast distribution into the light intensity distribution model, and then go through steps (2)-(4) in the phase calculation process to obtain more accurate reflectance distribution and phase information.

Figure 4 presents two solutions for the folded phase on one line of the image The distribution of , where the monotonically increasing area shown by the ellipse constitutes the initial phase value in one cycle; the bright spots in Figure 5 are the marked contrast-optimized pixels.

Claims (6)

1. a method for phase reconstruction based on a two-step phase-shift algorithm based on structured light contrast optimization, is characterized in that: comprising two processes of two-step phase-shift phase calculation and contrast optimization: (1) Two-step phase-shift phase calculation process, which processes the two collected phase-shift fringe images and calculates the phase information on the corresponding pixel points; including the following steps: ① Filter the collected images to reduce the influence of noise; ② The background light intensity is represented by the contrast change, and the light intensity distribution model about the two-step phase shift is obtained. From the light intensity distribution model of the two-step phase shift combined with the square sum relationship of the trigonometric function, the one-dimensional quadratic about the surface reflectance R of the object is obtained. equation, solve the equation to obtain two solutions for the reflectivity R; ③ Substitute the solution of the obtained reflectance R back into the light intensity distribution model, and obtain the information about the folding phase the two ④According to the characteristic of absolute phase increment, filter out the incremental part from the two folded phase solutions to form the initial solution of the folded phase, and mark the contrast-sensitive pixels at the same time; (2) The contrast optimization process is to select the pixels in the area sensitive to the contrast of structured light according to the linear increase of the absolute phase, make linear assumptions in its neighborhood, and achieve the modified contrast through the joint optimization of the contrast and the phase. Purpose; specifically includes the following steps: ①According to the linear characteristic of absolute phase, make a linear assumption on the phase value on the horizontal neighborhood of the marked pixel point, take 5-7 points on the horizontal neighborhood of the marked pixel point, and distribute the light intensity of these 5-7 points Simultaneous model, through the joint optimization of contrast and absolute phase, to achieve the purpose of optimizing regional contrast; ②According to the characteristics of similar contrast of adjacent pixels, the whole is interpolated and filled by the contrast of the local area, and the optimized contrast distribution of structured light in the entire projection range is obtained; The optimized contrast distribution is substituted into the light intensity distribution model in the two-step phase-shift phase calculation process, and then through steps ②-④ in the phase calculation process, more accurate reflectance distribution and phase information are obtained. 2 . The method for phase reconstruction using a two-step phase shift algorithm based on structured light contrast optimization according to claim 1 , wherein the phase difference between the two phase shift fringe patterns is 3π/2. 3 . 3. the method for phase reconstruction based on the two-step phase shift algorithm of structured light contrast optimization according to claim 1, is characterized in that: the light intensity distribution model of described two-step phase shift is: I ₁ (x, y)=R(x, y){1+A _m cos[φ(x, y)]}, I ₂ (x, y)=R(x, y){1+A _m sin[φ(x, y)]}, where R is the reflectivity of the object surface, _Am is the initial contrast of the projected fringes, and φ(x, y) is the phase value of the corresponding pixel on the image. 4. The method for phase reconstruction based on a two-step phase shift algorithm based on structured light contrast optimization according to claim 1, wherein the quadratic equation of the reflectivity R is: (2-A _m ² )R ² -2(I ₁ +I ₂ )R+I ₁ ² +I ₂ ² =0, _Am is the initial contrast of the projected fringes, and I ₁ and I ₂ are the light intensity distributions of the two collected images. 5. The method for phase reconstruction based on a two-step phase shift algorithm based on structured light contrast optimization according to claim 1, wherein the expression of the folded phase is: I ₁ and I ₂ are the light intensity distributions of the two collected images, and R ₁ and R ₂ are the two solutions of the quadratic equation of the reflectivity R. 6. the method for phase reconstruction based on the two-step phase shift algorithm of structured light contrast optimization according to claim 1, it is characterized in that: in the step 4. of described two-step phase shift phase calculation process, distinguish and mark pixel The point process is: ①Calculate the first-order difference of the two solutions in the horizontal direction to obtain the position matrix of the incremental region ②Calculate the discriminant matrix of marked pixels ③ If P=1, the position is a reliable pixel point, corresponding to two solutions, and the position matrix is If P=2, the position is a sensitive pixel, and the corresponding position matrix is S; ④ Determine the pixel at the corresponding pixel of the S matrix The part smaller than π is denoted as S ₁ , and the part larger than π is denoted as S ₂ ; ⑤ Calculate the initial estimate of the folding phase: