A high-precision master-slave camera vision positioning method

Abstract

This invention discloses a high-precision master-slave camera visual positioning method, aiming to provide a low-cost, low-cost, and master-slave camera visual positioning method that meets the accuracy requirements of a local inspection area, overcomes hardware limitations, and is applicable to various sizes. The method comprises the following steps: a. establishing a first matrix mapping relationship M1 between the image coordinates of the master camera 3 and the mechanical axis coordinates of the XY-axis mobile platform 1; b. establishing a second matrix mapping relationship M2 between the image coordinates of the master camera 3 and a second calibration plate 6; and a third matrix mapping relationship M3 between the image coordinates of the slave camera 4 and the second calibration plate 6; and c. converting the matrix mapping relationships into mechanical coordinates. The present invention has application in the field of visual positioning technology for gripping equipment.

Description

Technical Field

[0001] The present invention relates to a visual positioning technology, and in particular to a high-precision master-slave camera visual positioning method. Background Art

[0002] Visual positioning technology is a key enabler of intelligent industrial machinery and equipment. With technological advancements, there's an increasing demand for positioning with a wider field of view and higher precision. Industrial cameras with large fields of view and high pixel counts are expensive. For example, Chinese patent publication CN211630241U discloses a visual positioning device for electroplating a front film on a mobile phone back cover, which utilizes a 10-megapixel camera lens to improve positioning accuracy. However, current industrial cameras have limited fields of view and pixel counts, making them unattainable in certain scenarios, even without considering the cost. Therefore, when hardware cannot meet these requirements, a less hardware-intensive method is needed to achieve high-precision visual positioning.

[0003] In some demand scenarios, when the product to be positioned is large and the positioning accuracy is required to be at the micron level, the traditional approach is to use industrial cameras with larger target sizes and more pixels and telecentric lenses with higher field of view magnification to meet the needs. It is well known that industrial cameras with larger target sizes and more pixels are more expensive. When hundreds of millions of pixels are required, the price of an industrial camera is more than 100,000 yuan. A device generally uses many industrial cameras, which leads to excessively high equipment research and development costs and is difficult for the market to accept. On the other hand, when the product to be positioned is larger, there are no industrial cameras that meet the requirements on the market, resulting in a situation where there is a price but no market. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the existing technology and provide a high-precision master-slave camera visual positioning method that can meet the accuracy requirements of the local detection area, break through the hardware limitations and is applicable to various sizes and low cost.

[0005] The technical solution adopted by the present invention is as follows: the grasping device used in the present invention includes an XY axis moving platform 1, a camera fixing bracket 2, a master camera 3 and a slave camera 4, the master camera 3 and the slave camera 4 are both fixedly arranged on the camera fixing bracket 2, and the master camera 3 and the slave camera 4 are both located above the XY axis moving platform 1. The present invention includes the following steps:

[0006] a. Establishing a first matrix mapping relationship M1 between the image coordinates of the main camera 3 and the mechanical axis coordinates of the XY-axis mobile platform 1: A first calibration plate 5 has a solid circular pattern on it. The first calibration plate 5 is placed on the XY-axis mobile platform 1 and is at the center of the field of view of the main camera 3. The XY-axis mobile platform 1 is then moved nine times according to the nine-point calibration method. The image coordinates of these nine points and the corresponding mechanical axis XY coordinates of the XY-axis mobile platform 1 are obtained respectively, thereby establishing the first matrix mapping relationship M1 between the main camera 3 and the mechanical XY-axis coordinates;

[0007] b. Establish a second matrix mapping relationship M2 between the image coordinates of the main camera 3 and the second calibration plate 6, and establish a third matrix mapping relationship M3 between the image coordinates of the slave camera 4 and the second calibration plate 6: the calibration plate array is filled with solid circular patterns, and the XY spacing between two adjacent circles is 1 mm. It is placed flat in the field of view of the main camera 3 and the slave camera 4, and ensure that there are at least nine circles in the field of view of the main camera 3 and the slave camera 4; select nine circles in the field of view of the main camera 3 and the slave camera 4 respectively, and generate a Cartesian coordinate system with the upper left corner of the nine circles of the main camera 3 as the origin. The second matrix mapping relationship M2 between the main camera 3 and the calibration plate can be obtained by calculating the pixel coordinates of the nine circles in the field of view of the main camera 3 and combining them with the coordinates of the newly generated Cartesian coordinate system. Subsequently, the coordinates of the nine circles in the field of view of the slave camera 4 are obtained by analogy with the Cartesian coordinate system just generated by the main camera 3, and the third matrix mapping relationship M3 between the slave camera 4 and the second calibration plate 6 is obtained by combining the pixel coordinates of the nine circles in the field of view of the slave camera 4;

[0008] c. The main camera 3 identifies and locates the feature point A of the product to obtain the image coordinate PA, and the slave camera 4 identifies and locates the feature point B of the product to obtain the image coordinate PB. The image coordinate PB is converted to the calibration plate coordinate system WB using the third matrix mapping relationship M3, and then the pixel coordinate PB of the calibration plate coordinate system WB in the main camera 3 is obtained using the inverse matrix of the second matrix mapping relationship M2. 1 At this time, the pixel result of the slave camera 4 is converted to the master camera 3, which is equivalent to the feature point B of the product also being within the field of view of the master camera 3. At this time, the image coordinate PA and the pixel coordinate PB are mapped using the first matrix mapping relationship M1. 1 Convert to the mechanical axis coordinate system to obtain the mechanical XY axis coordinates HA and HB; use the mechanical XY axis coordinates HA and HB to obtain the angle and center point coordinates of the product in the mechanical axis, and ultimately achieve positioning.

[0009] Furthermore, the pixels of the master camera 3 and the slave camera 4 are both in the millions.

[0010] Furthermore, the pixels of the master camera 3 and the slave camera 4 are both 5 million.

[0011] Furthermore, a plurality of slave cameras 4 may be arranged on the camera fixing bracket 2 .

[0012] The beneficial effects of the present invention are: the present invention adopts a combination of two cameras for visual positioning, which can be applied to various positioning and grasping equipment. It only needs to select an industrial camera that meets the accuracy requirements of the local detection area, and adopts two low-pixel master-slave cameras to combine for visual positioning. It can break through the hardware limitations and be applicable to products of various sizes, greatly reducing the equipment cost. For example, a camera with 5 million pixels only costs two to three thousand yuan. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a schematic diagram of the hardware layout of the present invention;

[0014] Figure 2 It is the mechanical XY movement path;

[0015] Figure 3 This is a schematic diagram of master-slave camera calibration;

[0016] Figure 4 is a schematic diagram of the calibration plate coordinates;

[0017] Figure 5 It is a schematic diagram of product testing;

[0018] Figure 6 It is a schematic diagram of matrix transformation relationship;

[0019] Figure 7 It is a schematic diagram of coordinate transformation relationship;

[0020] Figure 8 It is the equipment operation flow chart. DETAILED DESCRIPTION

[0021] Summary of the present invention: 1. Establish the matrix mapping relationship between the main camera image coordinates and the mechanical axis coordinates (see Figure 1 ); 2. Establish the coordinate mapping relationship between the master camera and the slave camera (see Figure 3 ); 3. Use the results of 1 and 2 matrix mapping to locate the product during runtime (see Figure 5 ).

[0022] like Figure 1 As shown, there is a solid circular pattern on the first calibration plate 5, which is placed on the mechanical axis XY platform and in the center of the main camera's field of view, and then the mechanical XY axis is adjusted according to Figure 2 The path moves nine times, and the image coordinates of these nine points and the corresponding mechanical axis XY coordinates are obtained respectively, so as to establish the coordinate mapping relationship M1 between the main camera and the mechanical XY axis.

[0023] like Figure 3 As shown, the second calibration plate 6 is filled with solid circular patterns, with the XY spacing between adjacent circles being 1mm. They are placed flatly within the field of view of the master and slave cameras, ensuring that there are at least nine circles within the field of view of the master and slave cameras. Nine circles are selected from each of the master and slave cameras, and a Cartesian coordinate system is generated with the upper left corner of the nine circles of the master camera as the origin, as shown in the following example: Figure 4 As shown in the diagram, the K and N values are based on the actual coordinate values obtained with the upper left corner of the nine points of the main camera as the origin. By calculating the pixel coordinates of the nine points in the main camera's field of view and combining them with the newly generated Cartesian coordinate system coordinates, we can obtain the mapping matrix M2 between the main camera and the calibration plate. Similarly, we can obtain the mapping matrix M3 between the slave camera and the calibration plate.

[0024] like Figure 5 As shown, the main camera identifies and locates the feature point A to obtain the image coordinates PA, and the slave camera identifies and locates the feature point B to obtain the image coordinates PB. The matrix M3 is used to transform PB to the calibration plate coordinate system WB, and then the inverse matrix M2 is used to obtain the pixel coordinates PB of WB in the main camera. 1 At this time, the pixel results of the slave camera are converted to the main camera, which is equivalent to the feature point B also being within the field of view of the main camera. At this time, the matrix M1 is used to convert PA and PB 1 Convert to the mechanical axis coordinate system to obtain the coordinates HA and HB (see Figure 6 ). HA and HB can be used to obtain the angle and center coordinates of the product in the mechanical axis, ultimately achieving positioning.

[0025] The operation process of the grabbing device used in the present invention is as follows:

[0026] 1. Start;

[0027] 2. Preparation 1: Establish the matrix relationship M1 between the main camera and the mechanical axis;

[0028] 3. Preparation 2: Establish the matrix relationship M2, M3 between the master and slave cameras and the calibration plate;

[0029] 4. The equipment PLC tells the vision to perform positioning processing;

[0030] 5. Visual positioning of the product;

[0031] 6. Visual matrix M1, M2, M3 is converted into the final mechanical axis coordinate system;

[0032] 7. PLC obtains coordinates for deviation correction;

[0033] 8. End.

[0034] Compared with the prior art, the present invention has the following advantages:

[0035] 1. The positioning accuracy of the present invention is 3 μm, which is not limited by the size of the product and reduces the accuracy. This is fundamentally different from the traditional method of relying on the number of camera pixels to improve positioning accuracy. The present invention has a wider applicability.

[0036] 2. This invention breaks through the limitations of camera hardware. For certain demanding occasions, industrial cameras with ultra-large field of view and high positioning accuracy are not available on the market.

[0037] 3. The present invention greatly saves hardware costs. An industrial camera with hundreds of millions of pixels costs tens of thousands or even hundreds of thousands of yuan, while the present invention only requires a very low-pixel camera, which costs only a few thousand yuan each.

[0038] Although the embodiments of the present invention are described with practical solutions, they do not limit the meaning of the present invention. For those skilled in the art, it is obvious to modify the implementation scheme and combine it with other solutions based on this description.

Claims

1. A high-precision master-slave camera visual positioning method, wherein the grasping device used comprises an XY axis moving platform (1), a camera fixing bracket (2), a master camera (3) and a slave camera (4), wherein the master camera (3) and the slave camera (4) are both fixedly arranged on the camera fixing bracket (2), and the master camera (3) and the slave camera (4) are both located above the XY axis moving platform (1), and is characterized in that: The high-precision master-slave camera visual positioning method comprises the following steps: a. Establishing a first matrix mapping relationship (M1) between the image coordinates of the main camera (3) and the mechanical axis coordinates of the XY axis mobile platform (1): a first calibration plate (5) has a solid circular pattern on it, the first calibration plate (5) is placed on the XY axis mobile platform (1) and is located at the center of the field of view of the main camera (3), and then the XY axis mobile platform (1) is moved nine times according to the nine-point calibration method, and the image coordinates of the nine points and the corresponding mechanical axis XY coordinates of the XY axis mobile platform (1) are respectively obtained, thereby establishing the first matrix mapping relationship (M1) between the main camera (3) and the mechanical XY axis coordinates; b. Establish a second matrix mapping relationship (M2) between the image coordinates of the master camera (3) and the second calibration plate (6), and establish a third matrix mapping relationship (M3) between the image coordinates of the slave camera (4) and the second calibration plate (6): the calibration plate array is filled with solid circular patterns, and the XY spacing between two adjacent circles is 1 mm. They are placed flatly in the field of view of the master camera (3) and the slave camera (4), and it is ensured that there are at least nine circles in the field of view of the master camera (3) and the slave camera (4); nine circles are selected in the field of view of the master camera (3) and the slave camera (4), respectively, with the main camera (3) and the slave camera (4) as the first circle. The upper left corner of the nine dots of the main camera (3) is used as the origin to generate a Cartesian coordinate system. By calculating the pixel coordinates of the nine dots in the field of view of the main camera (3) and combining them with the newly generated Cartesian coordinate system coordinates, a second matrix mapping relationship (M2) between the main camera (3) and the calibration plate can be obtained. Then, the coordinates of the nine dots in the field of view of the slave camera (4) are obtained by analogy with the Cartesian coordinate system just generated by the main camera (3). The third matrix mapping relationship (M3) between the slave camera (4) and the second calibration plate (6) is obtained by combining the pixel coordinates of the nine dots in the field of view of the slave camera (4). c. The main camera (3) identifies and locates the feature point A of the product to obtain the image coordinate PA, and the slave camera (4) identifies and locates the feature point B of the product to obtain the image coordinate PB, and uses the third matrix mapping relationship (M3) to convert the image coordinate PB into the calibration plate coordinate system WB, and then uses the inverse matrix of the second matrix mapping relationship (M2) to obtain the pixel coordinate PB of the calibration plate coordinate system WB in the main camera (3). 1 At this time, the pixel result of the slave camera (4) is converted to the main camera (3), which is equivalent to the feature point B of the product being within the field of view of the main camera (3). At this time, the image coordinate PA and the pixel coordinate PB are mapped using the first matrix mapping relationship (M1). 1 Convert to the mechanical axis coordinate system to obtain the mechanical XY axis coordinates HA and HB; use the mechanical XY axis coordinates HA and HB to obtain the angle and center point coordinates of the product in the mechanical axis, and ultimately achieve positioning.

2. A high-precision master-slave camera visual positioning method according to claim 1, characterized in that: The pixels of the master camera (3) and the slave camera (4) are both in the millions.

3. The high-precision master-slave camera visual positioning method according to claim 2, characterized in that: The pixels of the master camera (3) and the slave camera (4) are both 5 million.

4. The high-precision master-slave camera visual positioning method according to claim 1, characterized in that: A plurality of slave cameras (4) may be arranged on the camera fixing bracket (2).

Images