Coding target real-time decoding and positioning method of gap measurement system

Abstract

The invention belongs to the technical field of target real-time decoding and positioning, and provides a coding target real-time decoding and positioning method of a gap measurement system, which comprises the following steps of: S1, layout of coding targets of gap detection points; s2, binarizing a coding target image shot at a large inclination angle, and solving the central positions of a coding region and a positioning region as initial points of compression transformation; s3, carrying out compressed projection transformation on the coding target point by adopting a single-direction wide edge detection and edge projection combined method; s4, decoding the coding target; according to the method, edge projection reference is provided by adopting single-direction wide edge detection, the influence of large-dip-angle shooting on compressed projection transformation is reduced, compressed target point projection transformation and edge pixel optimization are realized by adopting edge projection, the edge of the positioning area is combined with the edge of the coding target, and the positioning accuracy is improved. The edge error extraction rate is reduced, the sub-pixel positioning precision is improved, meanwhile, the method has the characteristics of rapidness and real-time performance, and is suitable for long-gap multi-point detection.

Description

Technical Field

[0001] The invention belongs to the technical field of real-time decoding and positioning of targets, and in particular relates to a real-time decoding and positioning method for a coded target of a gap measurement system. Background Art

[0002] The coding target decoding and positioning method of the gap measurement system is based on the technical requirements of real-time dynamic detection of the gap status. It uses binocular measurement equipment to arrange coding targets on both sides of the gap to be detected, and finally solves the spatial distance of the gap by real-time decoding and positioning of the coding targets. In the actual engineering application environment, the installation space is limited, which usually brings about the problem of large-angle shooting of the measurement equipment and the area of ​​the coding target is limited to a range, which requires the design of the coding target to be streamlined. At the same time, the decoding and positioning algorithm of the coding target is required to be accurate and have a certain degree of real-time performance. Solving the problem of fast and accurate decoding and positioning of large-angle shooting and streamlined targets is the key to ensuring fast, real-time and accurate measurement of the gap measurement system.

[0003] There are mainly the following technical difficulties in gap measurement in actual engineering:

[0004] 1. Impact of actual working environment: In actual engineering application environment, the installation space of the measurement system is limited, which will lead to large-angle shooting. Especially when detecting long gaps at multiple locations, the measurement equipment is usually installed at a certain position, which will also lead to large-angle shooting. In order to ensure the accuracy of gap measurement, the decoding and positioning of the coded target must have a high accuracy rate when shooting at large angles;

[0005] 2. The size of the coding target is limited: In the actual engineering application environment, the layout space of the coding target will also be limited, which requires the use of smaller coding targets as much as possible to achieve accurate decoding and positioning;

[0006] 3. Real-time requirements: In actual engineering applications, the measurement of gaps is usually required to have a certain degree of real-time performance, and the decoding of the rich coded information of the coded target will bring a huge amount of recognition algorithm calculations, which requires simplifying the coded target and using a simple algorithm to achieve accurate decoding and positioning;

[0007] 4. Traditional algorithms have low decoding accuracy for targets photographed at large angles. Usually, locators or coded target layers are added during target design to improve decoding accuracy. However, complex target design will increase the target size and bring about complex algorithms. In actual projects, it cannot meet real-time measurement requirements. At the same time, the limited space cannot realize target arrangement, resulting in the inability to measure.

[0008] The above technical difficulties limit the use of existing mature decoding and positioning methods of coded targets. Especially in actual engineering applications, when the gap measurement system is used in an airborne environment, the target installation layout space and the measurement equipment installation space are limited, resulting in the size of the target being limited within the range and the measurement equipment being photographed at a large angle. This requires the use of simple coded targets to quickly, in real time, and accurately achieve gap measurement when photographing at a large angle. This requirement cannot be achieved by commonly used mature algorithms. First, the commonly used mature algorithms have low decoding accuracy for coded targets photographed at a large angle, and errors will occur in gap measurement; second, in order to improve the decoding accuracy of coded targets, it is usually achieved by adding locators or coded target layers when designing targets, which will increase the target size and bring about the recognition of complex algorithms, resulting in a serious decrease in the extraction speed, which cannot meet the real-time requirements. Especially for long gap multi-position detection, as the number of coded targets increases, the algorithm calculation amount will increase exponentially, and can only be measured after calculation. Therefore, it is necessary to propose a real-time decoding and positioning method for coded targets of gap measurement system. Summary of the invention

[0009] In view of the problems existing in the prior art, the present invention provides a real-time decoding and positioning method for a coded target of a gap measurement system, which adopts single-direction wide edge detection to provide an edge projection benchmark, thereby reducing the influence of compressed projection transformation on large-angle shooting. At the same time, the edge in a single direction selects the vertical direction of the gap and avoids the gap edge. Edge projection is used to realize compressed target point projection transformation. When shooting at large angles, there is no need to add additional locators or target coding layers to the target design. Edge pixel optimization is adopted to combine the edge of the positioning area with the edge of the coded target, and mutual constraints are used to reduce the edge error extraction rate. While improving the sub-pixel positioning accuracy, it has the characteristics of fastness and real-time, and is particularly practical for engineering multi-point detection of long gaps.

[0010] The real-time decoding and positioning method of a coded target of a gap measurement system of the present invention comprises the following steps:

[0011] S1. Layout of coding targets at gap detection points: m coding targets are arranged on both sides of the gap to be measured, and correspond to the targets on both sides of the gap. The measurement system adopts a binocular vision system, which is set above the gap to be measured, and takes images of 2m coding targets 2 in real time at a large angle;

[0012] S2. Extract the initial point of the compressed projection transformation: Binarize the target image of step S1, segment the positioning area and the coding area of ​​2m coded targets, and preliminarily extract the pixel coordinates of the center position of the positioning area: p i (i=1,2,......2m), pixel coordinates of the center position of the encoding area: i is the ith coding target, n is the number of coding regions of the ith coding target, and n ≥ 8 is used as the initial point of the compressed projection transformation;

[0013] S3, extracting the target point of compression projection transformation: using single-direction wide edge detection combined with edge projection method, the center point of the coding area and the positioning area is subjected to compression projection transformation;

[0014] S4, encoding target decoding, specifically including:

[0015] S41, passing through the center target point p′ of the positioning area of ​​the i-th coded target i (i=1,2,......2m) draw a straight line parallel to the X axis

[0016] S42, sequentially obtain the center target point of the positioning area of ​​the i-th coding target and each p′ i (i=1,2,......2m) encoding area center target point The straight line

[0017] S43, seek and Angle If the y coordinate of the compression transformation target point at the center of the coding area of ​​the i-th coding target is greater than the straight line L i x The y coordinate of

[0018] S44, If the angle area is coded as 1, then find the coding value of the coding target;

[0019] S5, sub-pixel positioning of encoding target edge pixel optimization, specifically including:

[0020] S51, thinning the wide edge region of the i-th coding target edge, find the center line

[0021] S52, respectively in straight line Take k points for the upper average and Where k is an even number and k ≥ 10;

[0022] S53, and and and For corresponding points, find the coordinates of the center point of each pair of corresponding points.

[0023] S54: Determine the pixel coordinate p of the center position of the preliminary positioning area in step S2 i Is it in If it is not within the range, search for p again. i and point until it is within this range; if it is within the range, the centroid method is used to calculate the sub-pixel positioning center to achieve sub-pixel positioning of the coded target.

[0024] Preferably, the step S3 includes the following sub-steps:

[0025] S31, perform single-direction wide edge detection on the binary image described in step S2, and detect two wide edges of each coded target in a direction perpendicular to the gap. Where i represents the i-th encoding target;

[0026] S32, calculate the center position p of all positioning areas i (i=1,2,......2m) and the center position of all coding areas For each straight line If the distance is less than a certain threshold value T, it is determined that the center of the positioning area, the center of the coding area and the straight line belong to the same coding target;

[0027] S33, using the scanning method to calculate the two wide edges of each target area The four endpoints and Find the points separately and Point The distance between point and Point Distance

[0028] S34, sequentially locate the center position p of the positioning area for 2m targets i (i=1,2,......2m) and the center position of the coding area Calculate the horizontal projection point coordinates respectively And the projection point coordinates in the vertical direction Where i is the i-th encoding target, q is the q-th projection point of the i-th encoding target;

[0029] S35, sequentially locate the center position p of the positioning area for 2m targets i (i=1,2,......2m) and the center position of the coding area Calculate the vertical projection point coordinates respectively

[0030]

[0031] S36, successively locate the center position p of the positioning area of ​​2m targets i (i=1,2,......2m) and the center position of the coding area Calculate the corresponding compressed projection transformation target point p′ i (i=1,2,......2m) and The calculation formula is:

[0032]

[0033] In the formula, α i is the scaling factor of the i-th target point, is the x-coordinate value of the upper left point of the i-th target point, is the y coordinate value of the upper right point of the i-th target point, is the horizontal projection point of the qth center position of the ith target point With upper left point The distance The projection point in the vertical direction of the qth center position of the i-th encoded target point With upper right point The distance is the number of targets in the ith and The minimum value of .

[0034] Preferably, the width of the single-direction wide edge detection in step S31 is greater than 3 pixels.

[0035] Preferably, the same coded target determined in step S32 includes a positioning area center position p i , a set of coding region center positions and two straight lines

[0036]

[0037] Preferably, the horizontal projection coordinate of the qth point of the i-th encoded target in step S34 is The solution is:

[0038] S341, Solve the problem of passing through the qth point and parallel to the straight line Equation of the line

[0039] S342, Solve the passing point and Point Equation of the line

[0040] S343, Solving Straight Lines With straight line The intersection of is the horizontal projection coordinate of the qth point of the i-th coding target

[0041] Preferably, the vertical projection point coordinates of step S35 are The solution is:

[0042] S351, Solve the problem of passing through the qth point and parallel to the straight line Equation of the line

[0043] S352, Solving Straight Lines With straight line The intersection point is the vertical projection point coordinate of the qth point of the i-th coding target.

[0044] Preferably, the scaling factor α of the i-th encoding target in step S36 is i The value ranges from 0.8 to 1.

[0045] Compared with the prior art, the present invention has the following beneficial technical effects:

[0046] 1. The real-time decoding and positioning method of the coded target of the gap measurement system of the present invention adopts single-direction wide edge detection to provide an edge projection reference, which reduces the influence of the compressed projection transformation on the large-angle shooting. At the same time, the edge of the single direction selects the vertical direction of the gap to avoid the gap edge. The algorithm is not disturbed by the gap. With the wide edge detection, the area is easy to mark, which reduces the algorithm calculation amount and further improves the decoding accuracy.

[0047] 2. The real-time decoding and positioning method of the coded target of the gap measurement system of the present invention adopts edge projection to realize compressed projection transformation for the positioning center position and the coding center position point. When shooting at a large angle, there is no need to add additional locators or target coding layers to the target design. The target design is simple and the target size is miniaturized, so that the target layout is not easily restricted by space. The algorithm transforms the positioning center position and the coding center position, and no longer uses the affine transformation of the whole image. In particular, when detecting multiple positions of long gaps, the algorithm is simplified, and the operation speed is improved while meeting the requirements of accurate decoding at large angles, which has more engineering use value.

[0048] 3. The real-time decoding and positioning method of the coded target of the gap measurement system of the present invention adopts edge pixel optimization for sub-pixel positioning, so that the edge of the positioning area is combined with the edge of the coded target, constraining each other, reducing the edge error extraction rate, and further improving the sub-pixel positioning accuracy.

[0049] 4. The real-time decoding and positioning method of the coded target of the gap measurement system of the present invention can realize real-time measurement of the gap of automatic docking, the gap of the cabin door, the gap of the elevator door, etc., has good engineering use value, small amount of calculation, and good robustness of decoding sub-pixel positioning. It is suitable for various engineering applications such as large inclination measurement and long gap measurement with limited installation layout. While the basic principle remains unchanged, the coded target is changed to annular or circular, etc., and can be applied to more systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of the flow of the real-time decoding and positioning method of the coded target of the gap measurement system of the present invention;

[0051] Figure 2 It is a schematic diagram of the arrangement of the coded targets in the real-time decoding and positioning method of the coded targets in the gap measurement system of the present invention;

[0052] Figure 3 A schematic diagram of a single-direction wide edge in a real-time decoding and positioning method for a coded target of a gap measurement system of the present invention;

[0053] Figure 4 The wide edge, four endpoints and and Schematic diagram;

[0054] Figure 5 is the i-th encoding target according to the embodiment of the present invention Schematic diagram;

[0055] Figure 6 This is a schematic diagram of the angle between the center of the coding area and the X-axis after compression transformation in an embodiment of the present invention;

[0056] Figure 7 Schematic diagram of sub-pixel positioning for edge optimization in an embodiment of the present invention.

[0057] Main reference numerals:

[0058] Gap to be measured 1, coded target 2, measurement system 3. DETAILED DESCRIPTION

[0059] In order to fully describe the technical content, structural features, objectives and effects of the present invention, the following will be described in detail with reference to the accompanying drawings.

[0060] like Figure 1 As shown, the posture measurement method of the automatic alignment system of the embodiment of the present invention includes the following steps:

[0061] S1. m coded targets 2 are arranged on both sides of the gap 1 to be measured, and the targets on both sides of the gap correspond to each other one by one. The measurement system 3 adopts a binocular vision system, which is installed above the gap 1 to be measured, and takes images of 2m coded targets 2 in real time at a large angle.

[0062] In this application example, the coded targets on both sides of the gap are arranged as follows Figure 2 As shown. The coding target is designed with a circle as the basic element. It consists of a central positioning circle and a peripheral coding circle (8 bits). The counterclockwise direction is specified as the positive direction. The coding circle obtains an 8-bit binary code, which is cyclically shifted 7 times. The smallest value is selected as the coding value. The outer size of the mark of this scheme is 20mm*20mm, the diameter of the central positioning circle is 6mm, and the diameter of the coding circle is 3mm. In the example of the present invention, the camera is shot at a 60° inclination angle.

[0063] S2. Binarize the captured target image, segment the positioning area and coding area of ​​all 2m coded targets, and preliminarily extract the pixel coordinates p of the center position of the positioning area. i (i=1,2,......2m) (2m is the number of coded targets) and the pixel coordinates of the center position of the coded area (i is the i-th coding target, n is the number of coding regions of the i-th coding target, and n≥8) is used as the initial point of the compressed projection transformation.

[0064] In this application example, 5 coded targets are arranged on both sides of the gap. The maximum inter-class variance method is used to binarize the collected image, segment all 10 coded targets, and extract 10 sets of center position coordinates of the positioning area of ​​10 coded targets, i.e., the center circle of the coded targets, and 37 sets of center position coordinates of the coding area of ​​10 coded targets, i.e., the small circles around the center circle of the coded targets.

[0065] S3, extracting the compressed projection transformation target point based on single-direction wide edge detection combined with edge projection method, including the following sub-steps:

[0066] S31, perform single-direction wide edge detection on the binary image of step S2, and detect two wide edges of each coded target in a direction perpendicular to the gap. Where i represents the i-th encoding target.

[0067] S32, calculate the center position p of all positioning areas i (i=1,2,......2m) and the center position of all coding areas For each straight line If the distance is less than a certain threshold T, it is determined that the center of the positioning area, the center of the coding area and the straight line belong to the same coding target.

[0068] S33, scanning two wide edge areas And extract the two wide edges of each target area The four endpoints and Find the points separately and Point The distance between point and Point Distance

[0069] In this application example, for the 10 targets on both sides of the gap, the Sobel edge detection method with an edge detection width of 3 pixels is used to extract the target edges perpendicular to the gap direction. A total of 20 wide edges are extracted, such as Figure 3 As shown. The threshold T is selected based on the length of a single edge. In this example, T=40 is selected to complete the same target determination. Taking the 8th coded target as an example, its wide edge, four endpoints and and like Figure 4 shown.

[0070] S34, sequentially locate the center position p of the positioning area for 2m targets i (i=1,2,......2m) and the center position of the coding area Calculate the horizontal projection point coordinates respectively Where i represents the i-th encoding target, and q represents the q-th projection point of the i-th encoding target.

[0071] S341, Solve the problem of passing through the qth point and parallel to the straight line Equation of the line

[0072] S342, Solve the passing point and Point Equation of the line

[0073] S343, Solving Straight Lines With straight line The intersection of is the horizontal projection coordinate of the qth point of the i-th coding target

[0074] S35, sequentially locate the center position p of the positioning area for 2m targets i (i=1,2,......2m) and coding area

[0075] Central location Calculate the vertical projection point coordinates respectively The solution is as follows:

[0076] S351, Solve the problem of passing through the qth point and being parallel to the straight line Equation of the line

[0077] S352, Solving Straight Lines With straight line The intersection point is the vertical projection point coordinate of the qth point of the i-th coding target.

[0078] S353, successively locate the center position p of the positioning area of ​​2m targets i (i=1,2,......2m) and the center position of the coding area Calculate the corresponding compressed projection transformation target point p′ i (i=1,2,......2m) and The calculation formula is:

[0079]

[0080] In the formula, α i is the scaling factor of the ith target point, usually ranging from 0.8 to 1. is the x-coordinate value of the upper left point of the i-th target point, is the y coordinate value of the upper right point of the i-th target point, The horizontal projection point of the qth center position of the i-th coded target point With upper left point The distance The projection point of the qth center position of the i-th target point in the vertical direction With upper right point The distance is the number of targets in the ith and The minimum value of .

[0081] In this application example, the 8th coded target is taken as an example. Assuming that the leftmost coded circle is numbered 1, then like Figure 5 shown.

[0082] S4, encoding target decoding, specifically including:

[0083] S41, passing through the center target point p′ of the positioning area of ​​the i-th coded target i (i=1,2,......2m) draw a straight line parallel to the X axis

[0084] S42, sequentially obtain the center target point of the positioning area of ​​the i-th coding target and each p′ i(i=1,2,......2m) encoding area center target point The straight line

[0085] S43, seek and Angle If the y coordinate of the compression transformation target point at the center of the coding area of ​​the i-th coding target is greater than the straight line The y coordinate of

[0086] S44, If the angle area is coded as 1, then find the coding value of the coding target.

[0087] S5, sub-pixel positioning of encoding target edge pixel optimization, specifically including:

[0088] S51, thinning the wide edge region of the i-th coding target edge, find the center line

[0089] S52, respectively in straight line Take k points for the upper average and Where k is an even number and k ≥ 10.

[0090] S53, and and and For corresponding points, find the coordinates of the center point of each pair of corresponding points.

[0091] S54: Determine the pixel coordinate p of the center position of the preliminary positioning area in step S2 i Is it in If it is not within the range, search for p again. i and point until it is within this range; if it is within the range, the centroid method is used to calculate the sub-pixel positioning center to achieve sub-pixel positioning of the coded target.

[0092] In the present invention, the eighth coding target is taken as an example. The eighth coding target has four coding regions. According to the description of steps S41 to S44, after the center position of the positioning region of the coding target and the center position of the coding region are compressed and transformed, the straight line between the center target point of the positioning region and the center target point of each coding region is and Angle like Figure 6 shown.

[0093] The schematic diagram of obtaining the sub-pixel positioning center in the embodiment of the present invention is as follows: Figure 7 As shown. Take 10 points on the center line of the wide edge on average, that is, k = 10, and select a range of 2 pixels in both horizontal and vertical directions, that is, the center positions of 10 groups of positioning circles are fused with the center positions of the positioning areas initially extracted in step S2, totaling 11 groups of positions. If the horizontal and vertical pixel coordinate position differences between each two groups are less than 2 pixels, it is considered to be within the range, and the sub-pixel positioning center is obtained by the centroid method to achieve sub-pixel positioning of the coded target.

[0094] The embodiments described above are only descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A real-time decoding and positioning method for a coded target of a gap measurement system, characterized in that: It includes the following steps: S1. Layout of coding targets at gap detection points: m coding targets are arranged on both sides of the gap to be measured, and correspond to the targets on both sides of the gap. The measurement system adopts a binocular vision system, which is set above the gap to be measured, and takes 2m coding target images in real time; S2. Extract the initial point of the compressed projection transformation: Binarize the target image of step S1, segment the positioning area and the coding area of ​​2m coded targets, and preliminarily extract the pixel coordinates of the center position of the positioning area: p i (i=1,2,......2m), pixel coordinates of the center position of the encoding area: i is the ith coding target, n is the number of coding regions of the ith coding target, and n ≥ 8 is used as the initial point of the compressed projection transformation; S3, extracting the target point of compression projection transformation: using single-direction wide edge detection combined with edge projection method, the center point of the coding area and the positioning area is subjected to compression projection transformation; S4, encoding target decoding, specifically including: S41, passing through the center target point p′ of the positioning area of ​​the i-th coded target i (i=1,2,......2m) draw a straight line parallel to the X axis S42, sequentially obtain the center target point of the positioning area of ​​the i-th coding target and each p′ i (i=1,2,......2m) encoding area center target point The straight line S43, seek and Angle If the y coordinate of the compression transformation target point at the center of the coding area of ​​the i-th coding target is greater than the straight line The y coordinate of S44, If the angle area is coded as 1, then find the coding value of the coding target; S5, sub-pixel positioning of encoding target edge pixel optimization, specifically including: S51, refine the wide edge region of the i-th coding target edge, find the center line S52, respectively in straight line Take k points for the upper average and Where k is an even number and k ≥ 10; S53, and and and For corresponding points, find the coordinates of the center point of each pair of corresponding points. S54: Determine the pixel coordinate p of the center position of the preliminary positioning area in step S2 i Is it in If it is not within the range, search for p again. i and point until it is within this range; if it is within the range, the centroid method is used to calculate the sub-pixel positioning center to achieve sub-pixel positioning of the coded target.

2. The real-time decoding and positioning method for coding targets of a gap measurement system according to claim 1, characterized in that: Step S3 includes the following sub-steps: S31, perform single-direction wide edge detection on the binary image described in step S2, and detect two wide edges of each coded target in a direction perpendicular to the gap. Where i represents the i-th encoding target; S32, calculate the center position p of all positioning areas i (i=1,2,......2m) and the center position of all coding areas For each straight line If the distance is less than a certain threshold value T, it is determined that the center of the positioning area, the center of the coding area and the straight line belong to the same coding target; S33, using the scanning method to calculate the two wide edges of each target area The four endpoints and Find the points separately and Point The distance between point and Point Distance S34, sequentially locate the center position p of the positioning area for 2m targets i (i=1,2,......2m) and the center position of the coding area Calculate the horizontal projection point coordinates respectively And the projection point coordinates in the vertical direction Where i is the i-th encoding target, q is the q-th projection point of the i-th encoding target; S35, sequentially locate the center position p of the positioning area for 2m targets i (i=1,2,......2m) and the center position of the coding area Calculate the vertical projection point coordinates respectively S36, successively locate the center position p of the positioning area of ​​2m targets i (i=1,2,......2m) and the center position of the coding area Calculate the corresponding compressed projection transformation target point p′ i (i=1,2,......2m) and The calculation formula is: In the formula, α i is the scaling factor of the i-th target point, is the x-coordinate value of the upper left point of the i-th target point, is the y coordinate value of the upper right point of the i-th target point, is the horizontal projection point of the qth center position of the ith target point With upper left point The distance The projection point in the vertical direction of the qth center position of the i-th encoded target point With upper right point The distance is the number of targets in the ith and The minimum value of .

3. The real-time decoding and positioning method for coding targets of a gap measurement system according to claim 2, characterized in that: The width of the single-direction wide edge detected in step S31 is greater than 3 pixels.

4. The real-time decoding and positioning method for coded targets of a gap measurement system according to claim 2, characterized in that: The same coded target determined in step S32 includes a positioning area center position p i , a set of coding region center positions and two straight lines 5. The real-time decoding and positioning method for coded targets of a gap measurement system according to claim 2, characterized in that: The horizontal projection coordinates of the qth point of the i-th encoded target in step S34 are The solution is: S341, Solve the problem of passing through the qth point and parallel to the straight line Equation of the line S342, Solve the passing point and Point The equation of the line S343, Solving Straight Lines With straight line The intersection of is the horizontal projection coordinate of the qth point of the i-th coding target 6. The real-time decoding and positioning method for coded targets of a gap measurement system according to claim 2, characterized in that: The vertical projection point coordinates of step S35 The solution is: S351, Solve the problem of passing through the qth point and parallel to the straight line Equation of the line S352, Solving Straight Lines With straight line The intersection point is the vertical projection point coordinate of the qth point of the i-th coding target.

7. The real-time decoding and positioning method for coded targets of a gap measurement system according to claim 2, characterized in that: The scaling factor α of the i-th encoded target in step S36 i The value ranges from 0.8 to 1.

Images