CN101504770A - Structural light strip center extraction method - Google Patents

Abstract

The invention discloses a method for extracting the center of a structured light strip, comprising: obtaining an initial column of light strip center points; and obtaining structured light through multiple iterations between the center point of the light strip and the normal direction at the center point of the light strip The center of the light bar: performing a smoothing operation on the obtained center of the structured light bar to extract the final center of the structured light bar. Using the method for extracting the center of the structured light strip of the present invention, based on the extraction of the center of the one-dimensional light strip, through multiple iterations between the center point column of the light strip and the normal direction at the center point of the light strip, the high-precision light strip center can be realized The extraction does not involve a large number of addition and multiplication operations of image filtering, so it can reduce the amount of calculation and improve the processing speed while ensuring the detection accuracy; in addition, the normal direction of each center point of the light strip can be better calculated through iteration, The shape of the light strip image has little influence on the extraction precision, and the invention has good universality and strong anti-interference ability.

Description

A method for extracting the center of a structured light strip

technical field

The invention relates to a structured light vision measurement technology, in particular to a method for extracting the center of a structured light strip.

Background technique

In structured light vision measurement, fast and high-precision extraction of structured light strips is one of the key technologies. The high-speed dynamic structured light vision measurement system requires that the extraction speed of the light strip center be within a few milliseconds, and in order to ensure the accuracy and stability of the measurement system, the accuracy of the light strip center extraction is required to be at the sub-pixel level.

München，1998.27～81；D.Eberly，R.Gardner，Ridges for image analysis.Journal ofMathematical Imaging and Vision，1994，4：353～373；胡坤，一种快速结构光条纹中心亚像素精度提取方法，仪器仪表学报，2006，27(10)，1326～1329；周富强，结构光光条提取的混合图像处理方法，光电子与激光，2008，19(11)，1534～1537)。The existing sub-pixel level light bar center extraction algorithms mainly include: light bar section fitting method (MAG, Sub-pixel measurement of 3D surfaces by laser scanning, Sensors and Actuators A: Physical, 1999, 76 (1-3): 1 ~8; McIvor A M, Substripe localization for improved structured light system performance, Proc Of DICTA/IVCNZ97. New Zealand: Massey University, 1997, 309~314), multi-directional template method (Hu Bin, detection of structured light stripe center based on directional template Methods, Computer Engineering and Application, 2002, 38(11): 59-60, 109; Lei Haijun, A fast detection method for structured light fringe center, Journal of Huazhong University of Science and Technology (Natural Science Edition), 2003, 31(1): 74-76) and the Hessian matrix method (Carsten Steger, Unbiased Extraction of Curvilinear Structures from 2D and 3DImages, Germany: Technische

München, 1998.27～81; D.Eberly, R.Gardner, Ridges for image analysis.Journal of Mathematical Imaging and Vision, 1994, 4:353～373; Hu Kun, a fast method for extracting sub-pixel accuracy in the center of structured light stripes, Instrument Journal of Instrumentation, 2006, 27(10), 1326~1329; Zhou Fuqiang, Hybrid image processing method for structured light strip extraction, Optoelectronics and Laser, 2008, 19(11), 1534~1537).

The above-mentioned light strip cross-section fitting method performs Gaussian or parabolic fitting on the light strip cross-section for images with little change in the normal direction of the light strip, and then obtains the sub-pixel position in the center of the light strip by finding its extreme point. This method relies on Due to the shape of the light strip, the versatility is not good; for the light strip with a large change in the normal direction, the above-mentioned multi-directional template method can be used to detect the center position of the sub-pixel, but this method has a large amount of calculation; the above-mentioned Hessian matrix method is essentially It belongs to the fitting interpolation algorithm, which uses the Hessian matrix of the image and has good versatility, but it needs to perform convolution operations on the image repeatedly, and the processing speed is relatively slow.

Contents of the invention

In view of this, the main purpose of the present invention is to provide a method for extracting the center of a structured light strip, which can reduce the amount of calculation and improve the processing speed while ensuring the detection accuracy, and has good versatility and strong anti-interference ability.

In order to achieve the above object, technical solution of the present invention is achieved in that way:

A method for extracting the center of a structured light strip, comprising:

Obtain the initial bar center point column of the structured light bar;

Obtaining the center of the structured light light bar through multiple iterations between the center point of the light bar in the initial light bar center point column and the normal direction at the center point of the light bar;

A smoothing operation is performed on the acquired center of the structured light bar to extract the final center of the structured light bar.

The acquiring the initial light bar center point column includes: after determining the initial normal direction of the structured light light bar, obtaining the initial light bar center point column according to the determined initial normal line direction of the light bar.

The initial normal direction of the determined light strip is:

binarized light strip image;

Calculate the one-dimensional projection matrix of the binary image in the horizontal and vertical directions respectively;

Determines the direction corresponding to the one-dimensional projection matrix with fewer non-zero values as the initial normal direction of the light bar.

The column of obtaining the initial light bar center point according to the determined initial normal direction of the light bar is:

The initial normal direction is the horizontal direction, and the initial light bar center point column is obtained by scanning in rows; or, the initial normal direction is in the vertical direction, and the initial light bar center point column is obtained by column scanning.

The multiple iterations between the center point of the light bar and the normal direction at the center point of the light bar, and extracting the center of the structured light bar include:

a. Extracting multiple light bar center points in the obtained initial light bar center point column, and calculating the normal direction of each light bar center point;

b. Calculate a new center point of the light bar according to the normal direction of the center point of the light bar;

c. Judging whether to stop the iteration, if so, performing the smoothing operation; otherwise, calculating the normal direction of the new central point of the light strip in step b, and returning to step b.

The judgment whether to stop the iteration is: judging $\sqrt{\underset{k=0}{\overset{k=\mathrm{no}}{\mathrm{\&Sigma;}}}({(x_k^t-x_k^{t-1})}^2+{({\mathrm{the\; y}}_k^t-{\mathrm{the\; y}}_k^{t-1})}^2)/\mathrm{no}}<\mathrm{\&epsiv;}$ Whether it is true or not, wherein, n is the number of center points of light strips extracted in step a, and ε is a small positive number.

The method further includes: setting an upper limit T of the number of iterations;

Correspondingly, the judging whether to stop the iteration is: judging whether the number of iterations t=T holds true.

The method for extracting the center of the structured light strip in the present invention is based on the extraction method of the center of the one-dimensional light strip, and realizes the extraction of the center of the light strip with high precision through multiple iterations between the center point column of the light strip and the normal direction of the light strip. Since the present invention does not involve a large number of addition and multiplication operations of image filtering, it can reduce the amount of calculation and improve the processing speed while ensuring the detection accuracy.

In addition, since the present invention adopts an iterative idea, the normal direction of each point of the light strip can be better calculated, so the shape of the light strip image has little influence on the extraction accuracy of the present invention. In other words, the present invention has good versatility and anti-interference strong ability.

Description of drawings

Fig. 1 is a flow chart of the method for extracting the center of the structured light strip of the present invention;

Fig. 2 is a schematic diagram of the distribution function of the gray scale of the light bar in an ideal state;

Fig. 3 is a schematic diagram of the light bar gray scale distribution function in practical applications;

Fig. 4 is the flow chart of the method for obtaining the center point column of the light bar with higher precision in the present invention;

Fig. 5 is a schematic diagram of an ideal light strip image generated by a computer with a gray scale distribution as shown in Fig. 3;

Fig. 6 is the schematic diagram of the binarized light strip image obtained by adopting the adaptive threshold method;

FIG. 7 is a schematic diagram of a partial extraction effect of simulating the light bar shown in FIG. 5 .

Detailed ways

The basic idea of the present invention is: based on the extraction method of the one-dimensional light strip center, through multiple iterations between the light strip center point column and the light strip normal direction, the high-precision light strip center extraction is realized.

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail by citing the following embodiments and referring to the accompanying drawings.

Fig. 1 is a flow chart of the method for extracting the center of the structured light strip of the present invention. As shown in Fig. 1, the method for extracting the center of the structured light strip of the present invention mainly includes the following steps:

Step 11: Determine the initial normal direction of the structured light strip.

There are only two possibilities for the initial normal direction of the light bar: horizontal or vertical, which depends on the projection length of the light bar image in the horizontal and vertical directions. If the projection length in a certain direction is long, the direction is considered to be closer to the light bar. direction, so that another direction perpendicular to it is determined as the initial normal direction of the light bar. For example, if the projection length of the light strip image in the horizontal direction is longer, it is considered that the direction of the light strip is more similar to the horizontal direction, and its initial normal direction is the vertical direction. Therefore, the initial normal direction of the light strip can be obtained by calculating the one-dimensional projection matrix of the binary image in the horizontal direction and the vertical direction, and judging the number of non-zero values of the one-dimensional projection matrix.

Here, firstly, the adaptive threshold method is used to binarize the light strip image, and then the one-dimensional projection matrix of the binary image in the horizontal direction and the vertical direction is calculated respectively, and the number of non-zero values in the two one-dimensional projection matrices is compared, and there are more A one-dimensional projection matrix with non-zero values corresponds to the direction of the light strip, and its corresponding vertical direction is the initial normal direction (n _x , n _y ) of the light strip, that is, a one-dimensional projection matrix with fewer non-zero values corresponds to the light The initial normal direction of the bar. The above-mentioned binarization of light stripe images using adaptive threshold method (Otsu NA threshold selection method from gray level histograms [J]. Transaction on System Man and Cybernetic, 1979, 9 (1): 62-66) is a prior art, here No detailed introduction.

Step 12: According to the initial normal direction of the structured light strip determined in step 11, calculate the initial light strip center point column of the structured light strip.

Ideally, a one-dimensional light strip image can be approximated as a line-like structure with a Gaussian grayscale distribution along the normal direction of the centerline of the light strip, as shown in Figure 2. Taking the center point of the one-dimensional light strip as the origin, the one-dimensional mathematical model of image gray level distribution along the normal direction of the light strip can be expressed as $I\left(x\right)=h\&CenterDot;e^{-\mathrm{\&lambda;}{x}^2}$ , where h and λ are both non-zero positive real numbers, and the center of the light bar is the highest point of the function, which satisfies the formula (1):

$\mathrm{<span\; class="notranslate">\; 2</span>}{<span\; class="notranslate">\; \&Integral;</span>}_{\mathrm{<span\; class="notranslate">\; a</span>}}^{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; dx</span>}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; \&Integral;</span>}_{\mathrm{<span\; class="notranslate">\; a</span>}}^{\mathrm{<span\; class="notranslate">\; b</span>}}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; dx</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

Among them, x ₀ is the coordinate of the center of the light strip, and a, b are the two boundaries of the one-dimensional light strip. Since the integrals of I(x) on the left and right sides of x ₀ are equal, x ₀ can also be called the energy center. In practical applications, due to the limitation of camera gray sampling depth, the gray distribution function of the light bar will be shown in Figure 3, and it can also obtain the center point x ₀ of the light bar through formula (1). Obviously, the energy centers obtained by the two gray distribution functions through the formula (1) are the same.

Here, considering the discreteness of the image, for a point (x, y) and its normalized initial normal direction (n _x , n _y ), multiple sub-pixel coordinates can be obtained according to the step size l, such as 2m+1 sub-pixel coordinates : (x+k×l×n _x , y+k×l×n _y ), (k∈{-m,-m+1,...,m}), where the physical meaning of the step size l is a discrete point The distance, generally between 0.2 and 0.4, m represents the distance between the sub-pixel points in the (n _x , n _y ) direction away from the (x, y) point, and these points are located at the light bar passing point (x, y) with an interval l ), formula (1) becomes formula (2):

$\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; m</span>}}{\overset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; m</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; \&times;</span><span\; class="notranslate">\; [</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; m</span>}}{\overset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&Delta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&Delta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>$

(2)

Among them, k, x ₀ ∈{-m,...,m}, I(x+k×l×n _x , y+k×l×n _y ) represents the gray level of discrete pixels, by using the known Integer coordinate grayscale interpolation is obtained, Δ∈(0, 1), then (x+(x ₀ +Δ)×l×n _x , y+(x ₀ +Δ)×l×n _y ) is what is obtained for the light strip section center of light bar.

According to step 11, the initial normal direction can only be horizontal or vertical, so by scanning the image by row or column, using formula (2), you can get n initial light bar center points, that is, the initial light bar center point column .

Step 13: Through multiple iterations between the center point column of the light bar and the normal direction at the center point of the light bar, obtain a center point column of the structured light bar with higher precision.

The specific implementation process of the multiple iterations described in step 13 is shown in Figure 4, including the following steps:

Step 131: Extract multiple light bar center points in the initial light bar center point column obtained in step 12.

Here, the extraction refers to: for any point (x _i , y _i ) in the obtained light bar center point column, take its nearest 2q light bar center points in the light bar center point column, where , q is a positive integer.

Step 132: Calculate the normal direction corresponding to the central point of the light strip obtained in step 131.

Here, according to formulas (3)-(5), calculate the normalized normal direction of each light bar center point taken in step 131, and obtain: (n _x1 , n _y1 ), ......, ( n _xn , n _yn ). Wherein, (n _xn , n _yn ) represents the normal direction at the central point (x _n , y _n ) of the light strip.

${\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; q</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; q</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\sqrt{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

Step 133: Utilize the normal direction of each light bar center point calculated in step 132 to calculate a new light bar center point column by formula (2): $(x_1^t,{\mathrm{the\; y}}_1^t),......,(x_{\mathrm{no}}^t,{\mathrm{the\; y}}_{\mathrm{no}}^t),$ Among them, t represents the number of iterations.

Step 134: Determine whether to stop the iteration, if yes, the processing flow of step 13 ends; otherwise, execute step 135.

Here, there are multiple options for judging whether to stop the iteration. For example, by judging $\sqrt{\underset{k=0}{\overset{k=\mathrm{no}}{\mathrm{\&Sigma;}}}({(x_k^t-x_k^{t-1})}^2+{({\mathrm{the\; y}}_k^t-{\mathrm{the\; y}}_k^{t-1})}^2)/\mathrm{no}}<\mathrm{\&epsiv;}$ Whether to set up, to judge whether to stop iteration, here, n is the number that described step 131 extracts the light bar center point, and ε is a small positive number, generally gets 0.05, $\sqrt{\underset{k=0}{\overset{k=\mathrm{no}}{\mathrm{\&Sigma;}}}({(x_k^t-x_k^{t-1})}^2+{({\mathrm{the\; y}}_k^t-{\mathrm{the\; y}}_k^{t-1})}^2)/\mathrm{no}}<\mathrm{\&epsiv;}$ If it is established, it means that the result obtained by using the iterative method can basically not be changed, and it is meaningless to continue the iterative operation, so the iteration is stopped, and the processing flow of step 13 ends; otherwise, step 135 is executed. Of course, the upper limit T of the number of iterations can also be set. If the number of iterations t=T, the iteration is stopped, and the processing flow in step 13 ends; otherwise, t=t+1, step 135 is executed to continue the iteration. In practical applications, T usually takes 5-10.

通过公式(3)～(5)，计算新的光条中心点列的法线方向后，返回步骤133。Step 135: According to the light bar center point column calculated in step 133

Calculate the normal direction of the new light strip center point column by formulas (3)~(5) After that, return to step 133.

Step 14: Perform a smoothing operation on the relatively high-precision light bar center point sequence obtained in step 13 to extract the final center of the structured light light bar.

Record the energy center points of the light bar obtained through iteration in step 13 as: (x ₁ , y ₁ ), (x ₂ , y ₂ ), . . . (x _n , y _n ).

Based on formula (6):

$<span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}}_{\mathrm{<span\; class="notranslate">\; m</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}_{\mathrm{<span\; class="notranslate">\; m</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}}_{\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; ,</span>{\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}_{\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}}_{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}_{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

After smoothing the above point sequence, the point sequence is obtained $({\stackrel{\&OverBar;}{x}}_1^1,{\stackrel{\&OverBar;}{\mathrm{the\; y}}}_1^1),({\stackrel{\&OverBar;}{x}}_2^1,{\stackrel{\&OverBar;}{\mathrm{the\; y}}}_2^1).......,({\stackrel{\&OverBar;}{x}}_{\mathrm{no}-1}^1,{\stackrel{\&OverBar;}{\mathrm{the\; y}}}_{\mathrm{no}-1}^1).$ For the newly obtained point sequence, the above smoothing operation is repeated, and after p times of smoothing, the final result point sequence is obtained: $({\stackrel{\&OverBar;}{x}}_1^p,{\stackrel{\&OverBar;}{\mathrm{the\; y}}}_1^p),({\stackrel{\&OverBar;}{x}}_2^p,{\stackrel{\&OverBar;}{\mathrm{the\; y}}}_2^p),......,({\stackrel{\&OverBar;}{x}}_{\mathrm{no}-p}^p,{\stackrel{\&OverBar;}{\mathrm{the\; y}}}_{\mathrm{no}-p}^p).$ Generally, p takes 5-10.

The following is a simulation example about the present invention:

First, the computer generates an ideal light strip image with the grayscale distribution shown in Figure 3, as shown in Figure 5, with a resolution of 768×576, which is used to simulate the ideal spatial distribution of the grayscale of the light strip, and then add the mean is 0, and the Gaussian distribution noise with standard deviation σ is used to simulate the actual image of spatial light stripes. The specific process of extracting the light strip image shown in Figure 5 is as follows:

According to the above step 11, the binarized light strip image is obtained by using the adaptive threshold method, as shown in Figure 6, and then the one-dimensional projection matrix of the binary image in the horizontal and vertical directions is calculated as follows:

One-dimensional projection matrix in the horizontal direction: [0, ..., 0, 2, 3, 4, 6, 7, 7, ..., 127, 109, 89, 63, 1, 0 ,0,...,0], its size is 1×768, and the number of non-zero values is 104.

One-dimensional projection matrix in the vertical direction: [5, 5, 4, 5, 5, 5, 4, 4, ..., 5, 5, 4, 5, 5, 5, 4]', whose size is 576×1, and the number of non-zero values is 576.

Since the one-dimensional projection matrix in the horizontal direction has fewer non-zero values, the initial normal direction of the light bar is the horizontal direction.

According to the above step 12, since the initial normal direction of the light strip is the horizontal direction, it is considered that each row of pixels is a one-dimensional section of the light strip, thereby scanning the image by row, using formula (2), and selecting parameters l=0.3 and m=15 , 576 initial light bar center points are calculated: (400.068, 0), (400.497, 1), (401.248, 2), ..., (416.95, 574), (416.094, 575).

According to step 13, select all the calculated initial center points of light strips for iteration. During the iteration process, the relevant parameters are selected as: q=5, l=0.3, m=15, T=10, ε=0.05, and the final iteration contains 575 points of light bar center point column: (400.146, -0.0557085), (400.619, 0.912298), (401.13, 2.06592), ..., (417.975, 571.868), (416.959, 572.947), (416.581 , 573.811).

According to step 14, after smoothing the above point sequence 10 times using formula (6), the final result point sequence containing 565 points is obtained: (407.07, 10.1059), (407.763, 11.0993), (408.454, 12.0798), .. ...., (424.24, 561.87), (423.619, 562.859), (423.01, 563.86).

Figure 7 is a schematic diagram of the partial extraction effect of the simulated light strip shown in Figure 5, and Table 1 is the extraction accuracy data of the simulated image plus Gaussian noise light strip center:

Table 1

In this simulation example, the computer configuration is: CPU is Intel Core 2.0G, memory 1G, operating system is Windows XP, compiling tool is VC 6.0, the size of the image to be processed is 768×576, and the structured light strip of the present invention is adopted The center extraction method consumes 6.5 milliseconds.

It can be seen from the above that the present invention does not involve a large number of addition and multiplication operations of image filtering at all. For an image of 768×576 pixels, it only consumes 6.5 milliseconds, while using the improved filtering method based on the Hessian matrix-based light bar center sub-pixel extraction algorithm , using a processor with a main frequency of 2.4G needs to consume hundreds of milliseconds. If image preprocessing methods are used in combination to extract part of the image area where the light bar is located for processing, the same Hessian matrix method will consume about 40 milliseconds, so The processing speed of the present invention is fast; in terms of extraction accuracy, for the image shown in Figure 4, the extraction accuracy of the present invention is better than 0.12 pixels, which can fully meet the requirements of the structured light vision measurement system. In terms of versatility, since the present invention adopts an iterative idea, the normal direction of each point of the light strip can be better calculated, so the shape of the light strip image has little influence on the extraction accuracy of the present invention. In addition, it can be seen from the experimental results that the present invention has better noise anti-interference ability for images. Therefore, the present invention can reduce the calculation amount and improve the processing speed while ensuring the detection accuracy, and has good versatility and strong anti-interference ability.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention.

Claims (7)

1. A method for extracting the center of a structured light strip, characterized in that the method comprises: Obtain the initial bar center point column of the structured light bar; Obtaining the center of the structured light light bar through multiple iterations between the center point of the light bar in the initial light bar center point column and the normal direction at the center point of the light bar; A smoothing operation is performed on the acquired center of the structured light bar to extract the final center of the structured light bar. 2. The extraction method according to claim 1, characterized in that the list of obtaining the center point of the initial light bar is as follows: after determining the initial normal direction of the structured light light bar, according to the determined initial method of the light bar Line direction to get the initial bar center point column. 3. The extraction method according to claim 2, wherein the determination of the initial normal direction of the light strip is: binarized light strip image; Calculate the one-dimensional projection matrix of the binary image in the horizontal and vertical directions respectively; Determines the direction corresponding to the one-dimensional projection matrix with fewer non-zero values as the initial normal direction of the light bar. 4. The extraction method according to claim 3, wherein the initial light bar center point obtained according to the determined initial normal direction of the light bar is listed as: The initial normal direction is the horizontal direction, and the initial light bar center point column is obtained by scanning in rows; or, the initial normal direction is in the vertical direction, and the initial light bar center point column is obtained by column scanning. 5. The extraction method according to claim 1, wherein the extraction of the center of the structured light strip through multiple iterations between the center point of the light strip and the normal direction at the center point of the light strip comprises: a. Extracting multiple light bar center points in the obtained initial light bar center point column, and calculating the normal direction of each light bar center point; b. Calculate a new center point of the light bar according to the normal direction of the center point of the light bar; c. Judging whether to stop the iteration, if so, performing the smoothing operation; otherwise, calculating the normal direction of the new central point of the light strip in step b, and returning to step b. 6. The extraction method according to claim 5, wherein said judging whether to stop iteration is: judging $\sqrt{\underset{k=0}{\overset{k=\mathrm{no}}{\mathrm{\&Sigma;}}}({(x_k^t-x_k^{t-1})}^2+{({\mathrm{the\; y}}_k^t-{\mathrm{the\; y}}_k^{t-1})}^2)/\mathrm{no}}<\mathrm{\&epsiv;}$ Whether it is true or not, wherein, n is the number of center points of light strips extracted in step a, and ε is a small positive number. 7. The extraction method according to claim 5, further comprising: setting an upper limit T of the number of iterations; Correspondingly, the judging whether to stop the iteration is: judging whether the number of iterations t=T holds true.

Images