A feature point sorting method for circular stripe array targets

Abstract

The invention discloses a method for sorting feature points for circular stripe array targets, comprising: step S1: a camera sequentially collects a plurality of circular stripe array images; step S2: extracting a mask image of the circular stripe array image, and obtaining a region of interest of each circular stripe through a connected domain marker; step S3: comparing the plurality of circular stripe array images to obtain a binary image, and performing morphological hole filling to obtain a filled image; step S4: counting the area proportion of each filled image in the region of interest of each circular stripe, and calculating a coding value of each circular stripe; step S5: determining a phase shift of each circular stripe and an order of a circle center feature point; calculating a phase distribution of the circular stripe array image, and extracting a circle center feature point; step S6: establishing a one-to-one mapping relationship between a world coordinate and an image coordinate of a circle center feature point for subsequent camera calibration; determining the order of a circle center feature point by encoding circular stripes with different phase shifts, and having strong robustness and applicability.

Description

Technical Field

[0001] The present invention belongs to the field of visual measurement technology, and in particular, relates to a feature point sorting method for a circular stripe array target. Background Art

[0002] Camera calibration is a key step in the field of visual measurement, and its calibration accuracy directly affects the measurement accuracy of the entire system. Usually, camera calibration needs to be completed with the help of a specific target. First, the feature points in the target image are detected, and then the projection equation of the camera is established using the three-dimensional world coordinates and two-dimensional image coordinates of the feature points, and finally the intrinsic and extrinsic parameters of the camera are solved. Intrinsic parameters generally include parameters such as focal length, principal point, distortion, etc., which are closely related to the inherent characteristics of the camera; extrinsic parameters generally include parameters such as rotation matrix and translation vector, which are closely related to the position and posture of the camera. Traditional targets, such as chessboards and circular arrays, detect feature points from the image intensity distribution, and their detection accuracy is greatly affected by image blur. Phase targets, such as orthogonal stripes and circular stripe arrays, detect feature points from the image phase distribution, and their detection accuracy is less affected by image blur, which is also suitable for the calibration of defocused cameras.

[0003] Feature point sorting is crucial for establishing the projection equation of the camera. The general idea is to find the feature points located at the four corners, calculate the homography matrix between the target plane and the image plane, estimate their image coordinates based on the world coordinates of the feature points, and then determine the one-to-one mapping relationship between the world coordinates and the image coordinates. However, the above method is more dependent on the detection results of the corner feature points. When the corner feature points are detected incorrectly, it will lead to a large calibration error. Therefore, designing a feature point sorting method suitable for phase targets has important practical significance. Summary of the invention

[0004] The present invention provides a feature point sorting method for a circular stripe array target to solve the problems existing in the above-mentioned background technology.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is: a method for sorting feature points of a circular stripe array target, which specifically includes the following steps:

[0006] Step S1: fixing the camera and the LCD screen, the LCD screen sequentially displays a plurality of circular stripe array images, and the camera sequentially collects a plurality of circular stripe array images;

[0007] Step S2: Calculate the background intensity of the circular stripe array image, use the Otsu threshold algorithm to extract the mask image of the circular stripe array image, and obtain the region of interest of each circular stripe through connected domain marking;

[0008] Step S3: obtaining a binary image corresponding to each circular stripe array image by comparing the multiple circular stripe array images pixel by pixel, and performing morphological hole filling to obtain a corresponding filled image;

[0009] Step S4: Counting the area ratio of each filled image in the region of interest of each circular stripe, and calculating the coding value of each circular stripe in different circular stripe array images according to the area ratio, and the coding value of each circular stripe in different circular stripe array images constitutes a coding sequence;

[0010] Step S5: according to the coding sequence of each circular stripe, determine the phase shift of each circular stripe and the order of the corresponding circle center feature points; calculate the phase distribution of the circular stripe array image by using the phase shift method, and extract the circle center feature points;

[0011] Step S6: According to the order of all circle center feature points, a one-to-one mapping relationship between the world coordinates and the image coordinates of the circle center feature points is established for subsequent camera calibration.

[0012] Preferably, in step S1, the circular fringe array image captured by the camera can be expressed as:

[0013] I n (x,y)=A(x,y)+B(x,y)cos[φ(x,y)+2πC m,n / N];

[0014] Where: m = 1, 2, ..., M, M represents the number of circular fringes in the circular fringes array image; n = 1, 2, ..., N, N represents the number of phase shift steps, that is, the number of circular fringes array images; (x, y) represents the pixel coordinates; I n (x, y) represents the intensity distribution of the nth circular stripe array image; A(x, y) represents the background intensity; B(x, y) represents the modulation intensity; φ(x, y) represents the phase distribution; δ m,n =2πC m,n / N represents the phase shift of the mth circular stripe in the nth circular stripe array image, C m,n represents the code value of the mth circular stripe in the nth circular stripe array image, and its value range is 1, 2, ..., N; note that each circular stripe corresponds to N code values ​​in N circular stripe array images, and each code value is different; N code values ​​constitute a code sequence C m,1 C m ,2…C m,N , the coding sequence C corresponding to each circular stripe m,1 C m,2 …C m,N Each is different.

[0015] Preferably, in step S2, the calculation formula for the background intensity of the circular fringe array image is as follows:

[0016]

[0017] Preferably, in step S3, the binary image corresponding to each circular stripe array image can be expressed as:

[0018]

[0019] When the intensity value I of a pixel in the circular stripe array image n (x, y) is greater than the intensity value I of the other N-1 circular stripe array images at the corresponding pixel k (x, y), k≠n, then the corresponding pixel in the binary image is set to B n (x, y) = 1; otherwise, the corresponding pixel in the binary image is set to B n (x,y)=0.

[0020] Preferably, in step S4, the calculation formula for the area ratio of each filled image in the region of interest of each circular stripe is as follows:

[0021]

[0022] Where: L m (x, y) represents the region of interest of the mth circular stripe; F n (x,y) represents the binary image B n (x, y) hole filling image; because the phase shift of the mth circular stripe in the N circular stripe array images is δ m,n and the coded value C m,n Different, resulting in area ratio S m,1 ,S m,2 ,…,S m,N There are differences; refer to the area proportion S corresponding to the circular stripes with different phase shifts 2πn / N in the ideal case n , the coding sequence of the mth circular stripe can be deduced.

[0023] Preferably, in step S5, the calculation formula of the phase distribution of the circular fringe array image is as follows:

[0024]

[0025] According to the phase distribution of the circular stripe array image, the equal phase point of each circular stripe is extracted, and the equal phase circle of each circular stripe is obtained through the ellipse fitting algorithm, so that the center feature point can be extracted.

[0026] The beneficial effects of adopting the above technical solution are:

[0027] 1. The present invention provides a method for sorting feature points of circular stripe array targets, which determines the order of each center feature point by encoding circular stripes with different phase shift amounts, and has strong robustness and applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 is the circular stripe array image I 1 (x,y),I 2 (x,y),I 3 (x,y) and I 4 (x,y);

[0029] Figure 2 is the mapping relationship between the order of circular stripes and the coding sequence;

[0030] Figure 3 (a) is the mask image of the circular stripe array image, and (b) is the connected domain label image;

[0031] Figure 4 Binary image B 1 (x,y),B 2 (x,y),B 3 (x,y) and B 4 (x,y) and the filled image F 1 (x,y),F 2 (x,y),F 3 (x,y) and F 4 (x,y);

[0032] Figure 5 (a) is the phase distribution of the circular stripe array image, (b) is the coding sequence of each circular stripe, and (c) is the order of each circular stripe or circle center feature point; DETAILED DESCRIPTION

[0033] The specific implementation methods of the present invention are further explained in detail below by describing the embodiments with reference to the accompanying drawings, with the aim of helping those skilled in the art to have a more complete, accurate and in-depth understanding of the concept and technical solution of the present invention and facilitating its implementation.

[0034] like Figures 1 to 5 As shown, the present invention is a feature point sorting method for circular stripe array targets, which determines the order of each circle center feature point by encoding circular stripes with different phase shift amounts, and has strong robustness and applicability.

[0035] The specific working method is described below with specific embodiments:

[0036] Embodiment 1:

[0037] Step S1: The camera and the LCD screen are fixed, and the LCD screen sequentially displays N=4 circular stripe array images, and the camera sequentially collects N=4 circular stripe array images; Figure 1 Displays N = 4 circular stripe array images I 1 (x,y),I 2 (x,y),I 3 (x,y) and I 4 (x, y), each circular stripe array image includes M = 4 × 4 circular stripes; the phase shift of the mth circular stripe in N = 4 circular stripe array images δ m,n and the coded value C m,n For example, the m=1th circular stripe in the circular stripe array image I 1 (x,y),I 2 (x,y),I 3 (x,y) and I 4 The phase shifts in (x,y) are δ 1,1 =π / 2,δ 1,2 =π,δ 1,3 =3π / 2 and δ 1,4 =2π, the corresponding code values ​​are C 1,1 =1,C 1,2 =2,C 1,3 =3 and C 1,4 =4, so the coding sequence of the m=1th circular stripe is C 1,1 C 1,2 C 1,3 C 1,4 = 1234. Generally, the number of circular stripes that can be encoded by N = 4 circular stripe array images is 24, Figure 2 The mapping relationship between the order of circular stripes and the coding sequence is shown.

[0038] Step S2: Calculate the background intensity of the circular stripe array image The Otsu threshold algorithm is used to extract the mask image of the circular stripe array image, such as Figure 3 As shown in (a), the region of interest L of each circular stripe is obtained by connecting the connected domains. m (x,y), such as Figure 3 (b) as shown.

[0039] Step S3: by comparing N=4 circular stripe array images pixel by pixel, obtaining a binary image corresponding to each circular stripe array image, and performing morphological hole filling to obtain a corresponding filled image; Figure 4 The binary image B is shown in 1 (x,y),B 2 (x,y),B 3 (x,y) and B 4 (x,y), and the filled image F1 (x,y),F 2 (x,y),F 3 (x,y) and F 4 (x,y).

[0040] Step S4: Count the area proportion of each filled image in the region of interest of each circular stripe, and infer the coding value of each circular stripe in different circular stripe array images according to the area proportion, and the coding value of each circular stripe in different circular stripe array images constitutes a coding sequence; Figure 4 It can be seen that the area proportions corresponding to the circular stripes with different coding values ​​or phase shifts are different, and they are sorted from small to large according to the area proportions: coding value C = 3 or phase shift δ = 3π / 2, coding value C = 2 or phase shift δ = π, coding value C = 1 or phase shift δ = π / 2, coding value C = 4 or phase shift δ = 2π. For example, by sorting the area proportions corresponding to the m = 1th circular stripe in the N = 4 circular stripe array images from small to large, it can be deduced that the coding sequence of the m = 1th circular stripe is 1234.

[0041] Step S5: According to the coding sequence of each circular stripe, Figure 2 The mapping relationship between the order of circular stripes and the coding sequence is used to determine the phase shift of each circular stripe and the order of the corresponding center feature points; the phase shift method is used to calculate the phase distribution of the circular stripe array image, and then the equal phase point of each circular stripe is extracted, and the equal phase circle of each circular stripe is obtained through the ellipse fitting algorithm, so that the center feature point can be extracted. Figure 5 (a) shows the phase distribution of the circular fringe array image. Figure 5 (b) shows the coding sequence of each circular stripe. Figure 5 (c) shows the order of each circular stripe or circle center feature point.

[0042] Step S6: According to the order of all circle center feature points, a one-to-one mapping relationship between the world coordinates and the image coordinates of the circle center feature points is established for subsequent camera calibration.

[0043] The present invention is described above by way of example in conjunction with the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-mentioned method. As long as various non-substantial improvements are made using the method concept and technical solution of the present invention; or the above-mentioned concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. A feature point sorting method for circular stripe array targets, Features: The specific steps include: Step S1: fixing the camera and the LCD screen, the LCD screen sequentially displays a plurality of circular stripe array images, and the camera sequentially collects a plurality of circular stripe array images; Step S2: Calculate the background intensity of the circular stripe array image, use the Otsu threshold algorithm to extract the mask image of the circular stripe array image, and obtain the region of interest of each circular stripe through connected domain marking; Step S3: obtaining a binary image corresponding to each circular stripe array image by comparing the multiple circular stripe array images pixel by pixel, and performing morphological hole filling to obtain a corresponding filled image; Step S4: Counting the area ratio of each filled image in the region of interest of each circular stripe, and calculating the coding value of each circular stripe in different circular stripe array images according to the area ratio, and the coding value of each circular stripe in different circular stripe array images constitutes a coding sequence; Step S5: according to the coding sequence of each circular stripe, determine the phase shift of each circular stripe and the order of the corresponding circle center feature points; calculate the phase distribution of the circular stripe array image by using the phase shift method, and extract the circle center feature points; Step S6: According to the order of all circle center feature points, a one-to-one mapping relationship between the world coordinates and the image coordinates of the circle center feature points is established for subsequent camera calibration; In step S1, the circular stripe array image captured by the camera can be expressed as: I n (x,y)=A(x,y)+B(x,y)cos[φ(x,y)+2πC m,n / N]; Where: m = 1, 2, L, M, M represents the number of circular fringes in the circular fringes array image; n = 1, 2, L, N, N represents the number of phase shift steps, that is, the number of circular fringes array images; (x, y) represents the pixel coordinates; I n (x, y) represents the intensity distribution of the nth circular stripe array image; A(x, y) represents the background intensity; B(x, y) represents the modulation intensity; φ(x, y) represents the phase distribution; δ m,n =2πC m,n / N represents the phase shift of the mth circular stripe in the nth circular stripe array image, C m,n represents the code value of the mth circular stripe in the nth circular stripe array image, and its value range is 1, 2, L, N; note that each circular stripe corresponds to N code values ​​in N circular stripe array images, and each code value is different; N code values ​​constitute a code sequence C m,1 C m,2 LC m,N , the coding sequence C corresponding to each circular stripe m,1 C m,2 LC m,N Each is different.

2. A method for sorting feature points for circular stripe array targets according to claim 1, Features: In step S2, the calculation formula of the background intensity of the circular stripe array image is as follows:

3. A method for sorting feature points for circular stripe array targets according to claim 1, Features: In step S3, the binary image corresponding to each circular stripe array image can be expressed as: When the intensity value I of a pixel in the circular stripe array image n (x, y) is greater than the intensity value I of the other N-1 circular stripe array images at the corresponding pixel k (x, y), k≠n, then the corresponding pixel in the binary image is set to B n (x, y) = 1; otherwise, the corresponding pixel in the binary image is set to B n (x,y)=0.

4. A method for sorting feature points for circular stripe array targets according to claim 1, Features: In step S4, the calculation formula for the area ratio of each filled image in the region of interest of each circular stripe is as follows: Where: L m (x, y) represents the region of interest of the mth circular stripe; F n (x,y) represents the binary image B n (x, y) hole filling image; because the phase shift of the mth circular stripe in the N circular stripe array images is δ m,n and the coded value C m,n Different, resulting in area ratio S m,1 ,S m,2 ,L,S m,N There are differences; refer to the area proportion S corresponding to the circular stripes with different phase shifts 2πn / N in the ideal case n , the coding sequence of the mth circular stripe can be deduced.

5. A method for sorting feature points for circular stripe array targets according to claim 1, Features: In step S5, the calculation formula of the phase distribution of the circular fringe array image is as follows: According to the phase distribution of the circular stripe array image, the equal phase point of each circular stripe is extracted, and the equal phase circle of each circular stripe is obtained through the ellipse fitting algorithm, so that the center feature point can be extracted.

Images