CN111337939A - Method and device for estimating outer frame of rectangular object - Google Patents
Method and device for estimating outer frame of rectangular object Download PDFInfo
- Publication number
- CN111337939A CN111337939A CN201910174238.9A CN201910174238A CN111337939A CN 111337939 A CN111337939 A CN 111337939A CN 201910174238 A CN201910174238 A CN 201910174238A CN 111337939 A CN111337939 A CN 111337939A
- Authority
- CN
- China
- Prior art keywords
- straight line
- point
- triangle
- rectangular object
- angle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 114
- 235000004522 Pentaglottis sempervirens Nutrition 0.000 claims abstract description 4
- 238000004364 calculation method Methods 0.000 claims description 37
- 238000012937 correction Methods 0.000 claims description 25
- 238000007781 pre-processing Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 8
- 238000012986 modification Methods 0.000 claims description 6
- 230000004048 modification Effects 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 238000000513 principal component analysis Methods 0.000 claims description 5
- 230000000007 visual effect Effects 0.000 claims description 2
- 230000008569 process Effects 0.000 description 13
- 238000010586 diagram Methods 0.000 description 11
- 238000012360 testing method Methods 0.000 description 7
- 230000000694 effects Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000013499 data model Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/02—Systems using the reflection of electromagnetic waves other than radio waves
- G01S17/06—Systems determining position data of a target
- G01S17/42—Simultaneous measurement of distance and other co-ordinates
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/497—Means for monitoring or calibrating
Abstract
A method for estimating the outer frame of a rectangular object comprises the following steps: based on the three-dimensional point cloud of the rectangular object, under a bird's eye view, obtaining a rectangular convex hull, and connecting points positioned at the edge in the convex hull to obtain a first frame; respectively connecting each point in the convex hull with the origin of the coordinate to obtain a plurality of line segments, wherein the first endpoint of each line segment is the point in the convex hull, and the second endpoint is the origin of the coordinate system; if the line segment in the line segments has an intersection point with the first frame except the first endpoint, removing the first endpoint of the line segment in the convex hull, and connecting points at the edge in the rest points in the convex hull to obtain a second frame; calculating the area of the second frame; connecting two points farthest away in the second frame to serve as the hypotenuse of the first right-angle triangle; calculating to obtain a first right-angle triangle based on the area and the hypotenuse; based on the first right triangle, an outer frame of the rectangular object is obtained. Compared with the prior art, the method provided by the embodiment of the invention has stronger robustness.
Description
Technical Field
The invention relates to the technical field of unmanned driving, in particular to a method and a device for estimating an outer frame of a rectangular object.
Background
The existing method for estimating the outer frame of the vehicle mainly adopts the ergodic idea, initializes a rectangle and searches the whole solution space of a rotating angle, and finds the optimal solution by calculating different evaluation criteria, such as minimum area, maximum correlation or minimum variance; or the minimum circumscribed rectangle is calculated, and the feature points are selected by a motion method and rotated to the correct position. Although the prior art can estimate the frame of the vehicle, the prior art generally has the disadvantages of poor implementation and robustness, and is mainly embodied in the following two aspects:
(1) in the prior art, the idea of circular traversal is mainly adopted to traverse point clouds, the time complexity is greatly influenced by the size of the point clouds, if the number of points in the point clouds is too many, the operation speed of the algorithm is very low, and the time for obtaining a rectangular frame of a vehicle is very long, so that the prior art has the defect of real-time performance.
(2) In the prior art, the solution space is traversed mainly by adopting a circular traversal idea, that is, in the prior art, a rectangle with the minimum outer order or a rectangle with the minimum distance needs to be obtained by calculation, and optimization is performed by using the minimum cost function. If the point set is too sparse or the distribution is not uniform, the characteristic function is easy to fall into a local extremum in the optimization process, and a correct result cannot be obtained, so that the prior art has the defect of poor robustness.
Disclosure of Invention
The invention aims to provide a method and a device for estimating an outer frame of a rectangular object. On one hand, the points in the convex hull of the plane of the vehicle are processed by a virtual scanning line method, and the wild points belonging to the parts such as the roof, the engine hood, the trunk and the like are removed, so that the processing amount of the points can be reduced, the calculation speed is improved, and the calculation accuracy is also improved in the calculation process; on the other hand, the laser radar has universality for the fact that the point cloud scanned by the front vehicle is L-shaped, the rectangular right-angled triangle is obtained by calculation by utilizing the characteristic of the laser radar, the rectangular frame can be quickly obtained, and compared with the prior art that the rectangular model is adopted to process the L-shaped point cloud, the robustness is higher.
The first aspect of the present invention provides a method for estimating an outer frame of a rectangular object, including: based on the three-dimensional point cloud of the rectangular object, under a bird's eye view, obtaining a rectangular convex hull, and connecting points positioned at the edge in the convex hull to obtain a first frame; respectively connecting each point in the convex hull with the origin of coordinates to obtain a plurality of line segments, wherein a first endpoint of each line segment is a point located in the convex hull, and a second endpoint is the origin of coordinates; if the line segments in the line segments have intersection points with the first frame except the first end points, removing the first end points of the line segments in the convex hull, and connecting points at the edges of the remaining points in the convex hull to obtain a second frame; calculating the area of the second frame; connecting two points farthest away in the second frame to serve as the hypotenuse of the first right-angle triangle; calculating to obtain a first right-angle triangle based on the area and the hypotenuse; based on the first right triangle, an outer frame of the rectangular object is obtained.
Further, the step of acquiring a three-dimensional point cloud of the rectangular object comprises: acquiring a depth map of the laser radar; fitting the ground based on preset height information, and removing the ground from the depth map to obtain a first point set; clustering the first point set by using a region growing method to obtain a second point set; and obtaining a three-dimensional point cloud of the rectangular object based on the second point set.
Further, the step of obtaining a three-dimensional point cloud of the vehicle based on the second point set comprises: screening the second point set to obtain a point set of the rectangular object on the depth map; and converting the point set of the rectangular object on the depth map into a three-dimensional point cloud of the rectangular object under the space rectangular coordinate.
Further, the step of calculating a first right triangle based on area and hypotenuse includes: calculating to obtain a plurality of right triangles and coordinates of right angle points thereof based on the areas and the hypotenuses; and taking the right-angle triangle with the highest coincidence degree with the second frame in the plurality of right-angle triangles as the first right-angle triangle.
Further, based on the first right triangle, the step of obtaining the outer frame of the rectangular object includes: and rotating the first right-angle triangle by taking the middle point of the oblique edge as a central symmetry point to obtain the outer frame of the rectangular object.
Further, the method also comprises the step of correcting the outer frame of the rectangular object: making a vertical line from the right-angle point of the first right-angle triangle to the hypotenuse of the first right-angle triangle, and dividing the point cloud in the second frame into two parts; performing RANSAC straight line fitting on the two point clouds respectively based on two right-angle sides of the first right-angle triangle to respectively obtain a first straight line and a second straight line; and correcting the first straight line or the second straight line.
Further, the RANSAC straight line fitting step of the two point clouds based on the two right-angle sides of the first right-angle triangle respectively comprises the following steps: performing principal component analysis on the inner point of the first straight line and the inner point of the second straight line obtained by RANSAC to obtain a first characteristic value and a second characteristic value, wherein the first characteristic value is the maximum characteristic value of the inner point of the first straight line, and the second characteristic value is the maximum characteristic value of the inner point of the second straight line; if the first characteristic value is larger than the second characteristic value, determining that the first straight line is the long right-angle side of the first right-angle triangle, and the second straight line is the short right-angle side of the first right-angle triangle; and if the second characteristic value is larger than the first characteristic value, determining that the first straight line is the short right-angle side of the first right-angle triangle, and the second straight line is the long right-angle side of the first right-angle triangle.
Further, the step of correcting the first straight line or the second straight line includes: respectively comparing the first straight line and the second straight line with the three-dimensional point cloud of the rectangular object to respectively obtain a first confidence value and a second confidence value; wherein the confidence value is in direct proportion to the coincidence rate of the three-dimensional point cloud of the straight line and the rectangular object; the first line or the second line is corrected based on a difference between the first confidence value and the second confidence value.
Further, if the first confidence value is larger than the second confidence value, correcting the second line segment; and if the first confidence value is smaller than the second confidence value, correcting the first line segment.
Further, the step of correcting the second straight line includes: calculating the slope of a second straight line based on the slope of the first straight line; performing RANSAC straight line fitting on the inner point of the second straight line based on the slope of the second straight line to obtain an equation of the second straight line; obtaining a second right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses; and obtaining the outer frame of the corrected rectangular object based on the second right triangle.
Further, the step of correcting the first straight line includes: calculating the slope of the first straight line based on the slope of the second straight line; performing RANSAC straight line fitting on an inner point of the first straight line based on the slope of the first straight line to obtain an equation of the first straight line; determining a second right-angled triangle corrected for the first right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses; and obtaining the outer frame of the corrected rectangular object based on the second right triangle.
According to a second aspect of the present invention, there is provided an apparatus for estimating an outline of a rectangular object, comprising: the data processing module is used for obtaining a rectangular convex hull under the aerial view based on the three-dimensional point cloud of the rectangular object and obtaining a first frame by connecting points positioned at the edge in the convex hull; the calculation module is used for respectively connecting each point in the convex hull with the origin of coordinates to obtain a plurality of line segments; the first endpoint of each line segment is a point in the convex hull, and the second endpoint is the origin of the coordinate system; if the line segment in the line segments has an intersection point with the first frame besides the first endpoint, removing the first endpoint of the line segment in the convex hull, and connecting points at the edge in the rest points in the convex hull to obtain a second frame; calculating the area of the second frame; connecting two adjacent points which are farthest away in the second frame to serve as the hypotenuse of the first right-angle triangle; calculating to obtain a first right-angle triangle based on the area and the hypotenuse; based on the first right triangle, an outer frame of the rectangular object is obtained.
Further, the system also comprises a laser radar and a data preprocessing module; the laser radar scans the visible range to obtain a depth map; the data preprocessing unit is used for fitting the ground based on preset height information and removing the ground from the depth map to obtain a first point set; clustering the first point set by using a region growing method to obtain a second point set; and obtaining a three-dimensional point cloud of the rectangular object based on the second point set.
Further, the data preprocessing module, based on the second point set, obtains the three-dimensional point cloud of the rectangular object, including: screening the second point set to obtain a point set of a rectangular object on the depth map; and converting the point set of the rectangular object on the depth map into a three-dimensional point cloud of the rectangular object under the space rectangular coordinate.
Further, the calculating module, wherein the step of calculating the first right triangle based on the area and the hypotenuse includes: calculating to obtain a plurality of right triangles and coordinates of right angle points thereof based on the areas and the hypotenuses; and taking the right-angle triangle with the highest coincidence degree with the second frame in the plurality of right-angle triangles as the first right-angle triangle.
Further, the step of obtaining the outer frame of the rectangular object by the calculation module based on the first right triangle includes: and rotating the first right-angle triangle by taking the middle point of the oblique edge as a central symmetry point to obtain the outer frame of the rectangular object.
Further, still include: the correction module is used for making a perpendicular line from the right-angle point of the first right-angle triangle to the hypotenuse of the first right-angle triangle and dividing the three-dimensional point cloud of the vehicle into two parts; performing RANSAC straight line fitting on the two point clouds respectively based on two right-angle sides of the first right-angle triangle to respectively obtain a first straight line and a second straight line; and correcting the first straight line or the second straight line.
Further, the step of performing RANSAC straight line fitting on the two point clouds respectively by the correction module based on the two right-angle sides of the first right-angle triangle comprises the following steps of: performing principal component analysis on the inner point of the first straight line and the inner point of the second straight line obtained by RANSAC to obtain a first characteristic value and a second characteristic value, wherein the first characteristic value is the maximum characteristic value of the inner point of the first straight line, and the second characteristic value is the maximum characteristic value of the inner point of the second straight line; if the first characteristic value is larger than the second characteristic value, determining that the first straight line is the long right-angle side of the first right-angle triangle, and the second straight line is the short right-angle side of the first right-angle triangle; and if the second characteristic value is larger than the first characteristic value, determining that the first straight line is the short right-angle side of the first right-angle triangle, and the second straight line is the long right-angle side of the first right-angle triangle.
Further, the step of correcting the first straight line or the second straight line by the correction module includes: respectively comparing the first straight line and the second straight line with the three-dimensional point cloud of the rectangular object to respectively obtain a first confidence value and a second confidence value; wherein the confidence value is in direct proportion to the coincidence rate of the three-dimensional point cloud of the straight line and the rectangular object; the first line or the second line is corrected based on a difference between the first confidence value and the second confidence value.
Further, the correction module determines that the second segment is corrected if the first confidence value is greater than the second confidence value; and if the first confidence value is smaller than the second confidence value, correcting the first line segment.
Further, the step of the correction module correcting the second straight line includes: calculating the slope of a second straight line based on the slope of the first straight line; performing RANSAC straight line fitting on the inner point of the second straight line based on the slope of the second straight line to obtain an equation of the second straight line; obtaining a second right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses; and obtaining the corrected outer frame of the rectangular object based on the second right triangle.
Further, the step of the correction module correcting the first straight line comprises: calculating the slope of the first straight line based on the slope of the second straight line; performing RANSAC straight line fitting on an inner point of the first straight line based on the slope of the first straight line to obtain an equation of the first straight line; determining a second right-angled triangle corrected for the first right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses; and obtaining the outer frame of the corrected rectangular object based on the second right triangle.
According to a third aspect of the present invention, there is provided a computer storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the above-described method for estimating an outline of a rectangular object.
According to a fourth aspect of the present invention, there is provided an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program by the steps of the method for estimating the outline of the rectangular object.
The technical scheme of the invention has the following beneficial technical effects:
(1) laser radars are widely used in the field of autonomous driving due to their high accuracy and long effective detection distance. Compared with a millimeter wave radar, the laser radar can realize accurate modeling of surrounding environment, pedestrians and vehicles. In the aspect of vehicle target tracking, an accurate judgment on the position and the size of a vehicle is required. According to the method, the point cloud of the vehicle is obtained through the laser radar, the vehicle model is subjected to rapid frame estimation by using a geometric method, and an accurate vehicle rectangular outer frame is obtained through further correction, so that the vehicle can be tracked rapidly and stably.
(2) According to the rectangular frame estimation method and device provided by the embodiment of the invention, on one hand, the points in the convex hull of the plane of the vehicle are processed by a virtual scanning line method, and the wild points belonging to the components such as the vehicle roof, the engine hood, the trunk and the like are removed, so that the processing amount of the points can be reduced, the calculation speed is increased, and the calculation accuracy is also improved in the calculation process; on the other hand, the laser radar has universality for the fact that the point cloud scanned by the front vehicle is L-shaped, the rectangular right-angled triangle is obtained by calculation by utilizing the characteristic of the laser radar, the rectangular frame can be quickly obtained, and compared with the prior art that the rectangular model is adopted to process the L-shaped point cloud, the robustness is higher.
(3) According to the rectangular frame calculating method and device provided by the embodiment of the invention, RANSAC is adopted to further correct the initially obtained frame of the rectangular object, and compared with the prior art, the frame calculating method and device are easier to converge and have fewer convergence steps during iterative calculation, so that compared with the prior art, the correction process is more stable, the calculation speed is higher, and the accuracy is higher.
Drawings
FIG. 1 is a schematic flow chart of a method for estimating the outer frame of a rectangular object according to a first embodiment of the present invention;
FIG. 2 is a lidar depth map of a first embodiment of the invention;
FIG. 3 is a three-dimensional point cloud scanned over a rectangular object by a lidar in accordance with a first embodiment of the invention;
FIG. 4(a) is a first schematic diagram of a first embodiment of the present invention;
FIG. 4(b) is a schematic diagram of a second frame of the first embodiment of the present invention;
FIG. 4(c) is a schematic diagram of determining a first right-angled triangle slash according to the first embodiment of the invention;
FIG. 4(d) is a schematic diagram of determining a first right triangle point according to the first embodiment of the present invention;
FIG. 4(e) is a schematic view of the outer rim of a rectangular object according to the first embodiment of the present invention;
FIG. 5(a) is a schematic diagram of the modification of the outer frame of the estimated rectangular object according to the first embodiment of the present invention;
FIG. 5(b) is a schematic diagram of two edges of the estimated rectangle according to the first embodiment of the present invention;
FIG. 5(c) is a schematic diagram of the first embodiment of the present invention for correcting two sides of the estimated rectangular object;
FIG. 5(d) is a schematic view of the modified rectangular object outline according to the first embodiment of the present invention;
FIG. 6 is a graph showing the time consumption for calculation according to the estimation method of the first embodiment of the present invention;
FIG. 7 is a diagram of the relative error results of the estimation method provided by the second embodiment of the present invention to the first embodiment;
FIG. 8 is a diagram of the result of the absolute error of the estimation method provided by the second embodiment of the present invention with respect to the first embodiment;
fig. 9 is a schematic structural diagram of an estimation apparatus for an outer frame of a rectangular object according to a third embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings in conjunction with the following detailed description. It should be understood that the description is intended to be exemplary only, and is not intended to limit the scope of the present invention. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present invention.
Fig. 1 is a schematic flow chart of a method for estimating an outer frame of a rectangular object according to a first embodiment of the present invention, where the method includes steps S101 to S107.
In a preferred embodiment, before step S101, the laser point cloud of the rectangular object scanned by the laser radar is preprocessed. The rectangular object in the present application is exemplified by a vehicle, but is not limited thereto.
Optionally, the step of preprocessing the lidar of the rectangular object scanned by the lidar includes:
firstly, a depth map of a laser radar visible range is obtained, and the depth map of the laser radar visible range can be seen in fig. 2.
And secondly, fitting the ground based on preset height information and removing the ground from the depth map to obtain a first point set.
Specifically, in this step, the method for removing the ground may be: firstly, an original point set in the depth map is intercepted according to a higher preset height to obtain a point set A. And then carrying out plane fitting on the point set A to obtain a ground plane equation. Secondly, the distance between each point in the depth map and the ground plane equation is respectively calculated, the points with the distance smaller than the threshold value are deleted (the deleted points are the points belonging to the ground), and the original point set in the depth map is the points of all targets except the ground. The ground is removed mainly to reduce the data processing amount of subsequent estimation, so that the clustering process in the third step is simpler.
And thirdly, clustering the first point set by using a region growing method to obtain a second point set.
In particular, a first set of points presents a plurality of obstacles, including: a leading vehicle, a road sign, etc. The first set of points after the removal of the ground, which may include many piles of point clouds, needs to be clustered to connect point clouds belonging to an object.
The second point set refers to a point set of a rectangular object on the depth map. That is, in the third step, clustering is performed under the image coordinate system of the depth map.
And fourthly, obtaining the three-dimensional point cloud of the rectangular object based on the second point set.
Specifically, screening the second point set to find a point set belonging to a rectangular object; and after finding, converting the second point set on the depth map into a three-dimensional point cloud of the rectangular object under the space rectangular coordinate. The three-dimensional point cloud of the rectangular object can be seen in fig. 3.
The method for estimating the outline of the rectangular object will be described in detail below.
Step S101, based on the three-dimensional point cloud of the rectangular object, under the aerial view, a rectangular convex hull is obtained, and a first frame is obtained by connecting points at the edge in the convex hull. The frame enclosed by the dotted line shown in fig. 4(a) is the first frame.
And S102, respectively connecting each point in the convex hull with a coordinate origin to obtain a plurality of line segments, wherein a first endpoint of each line segment is a point located in the convex hull, a second endpoint is the coordinate origin, and the coordinate origin is a point represented by the laser radar.
Step S103, traversing a plurality of line segments, if the line segment in the plurality of line segments has an intersection point with the first frame besides the first end point, removing the first end point of the line segment in the convex hull, and connecting points at the edge in the rest points in the convex hull to obtain a second frame. In the example shown in fig. 4(b), the frame surrounded by the solid line is the second frame.
In step S103, if a line connecting a certain point in the convex hull and the laser radar and the first frame have an intersection, it is assumed that the certain point is a point blocked by another object and that the certain point is a point which is not useful for estimating the frame of the rectangular object, and therefore the certain point may be a outlier (for example, the certain point may be a point belonging to the roof of the vehicle), and the certain point is removed.
It should be noted that, in this step, the virtual connection method is used to simplify the point set on the convex hull, and is not an actual connection. The points in the convex hull of the plane of the vehicle are processed by a virtual scanning line method, and the wild points belonging to the parts such as the roof, the engine hood, the trunk and the like are removed, so that the processing amount of the points can be reduced, the calculation speed is increased, and the calculation accuracy is also improved.
Step S104, calculating the area of the second frame.
In step S105, the two points farthest apart in the second frame are connected to be the hypotenuse of the first right triangle, which is shown by the oblique line in fig. 4 (c).
And step S106, calculating to obtain a first right-angle triangle based on the area and the hypotenuse.
Specifically, a plurality of right triangles and coordinates of right angle points thereof are calculated based on the area and the hypotenuse; the right-angled triangle with the highest coincidence degree with the second frame in the plurality of right-angled triangles is taken as the first right-angled triangle, and see fig. 4 (d).
In step S107, based on the first right triangle, an outer frame of the rectangular object is obtained, see fig. 4 (e).
Optionally, the middle point of the oblique side is taken as a central symmetry point, and the outer frame of the rectangular object is obtained. Or the outer frame of the rectangular object can be obtained by respectively making parallel lines of two right-angle sides of a right-angle triangle along two end points of the hypotenuse.
In a preferred embodiment, the method further comprises a step S108 of modifying the outer frame of the rectangular object.
Optional step S108 includes the following sub-steps:
step S108-1, a perpendicular line is drawn from the right-angle point of the first right-angle triangle to the hypotenuse thereof, and the point cloud in the second frame is divided into two parts, see fig. 5 (a).
Step S108-2, RANSAC (random sample consensus) straight line fitting is respectively carried out on the two point clouds based on the two right-angle sides of the first right-angle triangle, so as to respectively obtain a first straight line and a second straight line.
The RANSAC straight line simulation step of two part point clouds based on two right-angle sides of the first right-angle triangle respectively comprises the following steps:
and respectively carrying out principal component analysis on the inner point of the first straight line and the inner point of the second straight line obtained by RANSAC to obtain a first characteristic value and a second characteristic value. The first characteristic value is the maximum characteristic value of the inner point of the first straight line, and the second characteristic value is the maximum characteristic value of the inner point of the second straight line;
if the first characteristic value is larger than the second characteristic value, determining that the first straight line is a long right-angle side of the first right-angle triangle; if the second eigenvalue is greater than the first eigenvalue, the second straight line is determined to be the long right-angle side of the first right-angle triangle, see fig. 5 (b).
In step S108-2, the "eigenvalue" is proportional to the length.
And S108-3, correcting the first straight line or the second straight line.
Specifically, the first straight line and the second straight line are respectively compared with the three-dimensional point cloud of the rectangular object, and a first confidence value and a second confidence value are respectively obtained, see fig. 5 (c). Wherein, the confidence value is in direct proportion to the coincidence rate of the three-dimensional point cloud of the straight line and the rectangular object.
The first line or the second line is corrected based on a difference between the first confidence value and the second confidence value.
Preferably, the second segment is modified if the first confidence value is greater than the second confidence value.
Specifically, the correction of the second segment may be: calculating the slope of a second straight line based on the slope of the first straight line; performing RANSAC straight line fitting on the inner point of the second straight line based on the slope of the second straight line to obtain an equation of the second straight line; obtaining a second right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses; the outer frame of the modified rectangular object is obtained based on the second right triangle, see fig. 5 (d).
Specifically, since the first straight line, the second straight line, and the hypotenuse form a right triangle, the slope of the second straight line and the slope of the first straight line have an inverse relationship with each other. The slope of the second line is equal to 1 divided by the slope of the first line.
Preferably, the first segment is corrected if the first confidence value is less than the second confidence value.
Specifically, the step of correcting the first straight line includes: calculating the slope of the first straight line based on the slope of the second straight line; performing RANSAC straight line fitting on an inner point of the first straight line based on the slope of the first straight line to obtain an equation of the first straight line; determining a second right-angled triangle corrected for the first right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses; and obtaining the outer frame of the corrected rectangular object based on the second right triangle.
It should be noted that RANSAC straight line fitting is respectively performed on two point clouds based on two right-angle sides of the first right-angle triangle, which straight line is the long right-angle side of the first straight line and the second straight line can be determined, and the short right-angle side can be directly corrected. The confidence of the first straight line and the confidence of the second straight line are calculated, and actually, the confidence is also used for determining which straight line of the first straight line and the second straight line is the long right-angle side, and generally, the higher the confidence is, the higher the length of the straight line is, and the short right-angle side is corrected.
It should be noted that, the invention corrects the initially obtained rectangle by the random sampling method RANSAC, which can effectively reduce the interference of the field to the solved equation. Because only the points on the surface perpendicular to the ground on the vehicle are needed when the second frame is calculated, but the points on the roof, the bonnet and the trunk lid still participate in the calculation at this time, if a virtual connection method is not adopted for random sampling, the influence of the wild point on the solution cannot be avoided at all.
The technical scheme of the invention has the following beneficial technical effects:
(1) laser radars are widely used in the field of autonomous driving due to their high accuracy and long effective detection distance. Compared with a millimeter wave radar, the laser radar can realize accurate modeling of surrounding environment, pedestrians and vehicles. In the aspect of vehicle target tracking, an accurate judgment on the position and the size of a vehicle is required. According to the method, the point cloud of the vehicle is obtained through the laser radar, the vehicle model is subjected to rapid frame estimation by using a geometric method, and an accurate vehicle rectangular outer frame is obtained through further correction, so that the vehicle can be tracked rapidly and stably.
(2) According to the rectangular frame estimation method and device provided by the embodiment of the invention, on one hand, the points in the convex hull of the plane of the vehicle are processed by a virtual scanning line method, and the wild points belonging to the parts such as the vehicle roof, the engine hood, the trunk and the like are removed, so that the processing amount of the points can be reduced, the calculation speed is increased, and the calculation accuracy is also improved in the calculation process; on the other hand, the laser radar has universality for the point cloud scanned by the front vehicle in an L shape, the rectangular right-angled triangle is obtained by utilizing the characteristic of the laser radar, the rectangular frame can be quickly obtained, and compared with the prior art that the rectangular model is adopted to process the point cloud in the L shape, the robustness is stronger.
(3) According to the rectangular frame calculating method and device provided by the embodiment of the invention, RANSAC is adopted to further correct the initially obtained frame of the rectangular object, and compared with the prior art, the frame calculating method and device are easier to converge and have fewer convergence steps during iterative calculation, so that compared with the prior art, the correction process is more stable, the calculation speed is higher, and the accuracy is higher.
The accuracy of the method for estimating the outer frame of the rectangular object according to the first embodiment of the present invention will be described in detail with reference to fig. 6 to 8.
The second embodiment of the present invention uses the estimation method provided in the first embodiment (without correction), uses the estimation method provided in the first embodiment with correction, and uses the algorithm in the "PCL" library to test and calculate a plurality of sets of data. The test data is from the actual road test of a 40-line laser radar of the standing grain, 28 groups of data models are randomly selected for sampling, the actual vehicle running direction is manually marked as a true value, and the operation consumption time, the relative error and the absolute error are subjected to statistical analysis.
Fig. 6 is a result graph of the calculation time consumption of the estimation method according to the second embodiment of the present invention with respect to the first embodiment.
As shown in fig. 6-8, the first method represents the test data (shown as a circular broken line) using the estimation method (without correction) provided by the first embodiment, the second method represents the test data (shown as a triangular broken line) using the estimation method (with correction) provided by the first embodiment, and the third method represents the test data (shown as a square broken line) using the algorithm in the "PCL" library.
As can be seen from comparison of fig. 6, the estimation method (modified) provided by the first embodiment does not require multiple iterations, and therefore, compared with the estimation method of the third embodiment, the time consumed by calculation is greatly reduced.
As can be seen from the comparison of FIG. 7, the relative error of each set of data in the second method is around "0", and the fluctuation is small under the test of different data models, so that the second method has stronger stability and robustness. The relative error of each group of data of the first method fluctuates around 0, the error fluctuation is obvious among different models, and the stability is lower compared with that of the first method. The relative error of each group of data of the third method fluctuates greatly around 0, and the error fluctuation is large among different models, so that the relative error estimated by adopting the third method is large, and more instability factors exist. In summary, the second method is the best estimation method for estimating the rectangular object, the first method, and the third method.
As can be seen from comparison in fig. 8, the absolute error of the second method for estimating the rectangular object is relatively small, and each set of data is relatively stable, the first method and the third method for estimating the rectangular object generally have relatively significant errors, but the first method has more cases of significant errors during estimation, and the second method is more prone to extreme errors during estimation.
The following table 1 shows more intuitively the values of the time consumption, the relative error and the absolute error of the calculation of the three methods.
As can be seen from table 1 above, the outer frame of the vehicle estimated by the first method has a considerably low variance of response time (4.887deg) at 0.0190 ms, and is therefore hardly affected by the size of the dot set, and the average error obtained by the second method is 0.476 degrees, and therefore, the average error obtained by the second method is relatively small. The average value and the variance of the time consumed by adopting the third method are large, which shows that the calculation of the method is time-consuming, and therefore, the consumed time has a large relation with the data quantity; moreover, the variance of the angle of the third method is large, which means that the calculation result of the third method is not very stable. Therefore, the result obtained by the second method is the result of the optimal solution as a whole, and the effect is the best.
Fig. 9 is a schematic structural diagram of an estimation apparatus for an outer frame of a rectangular object according to a third embodiment of the present invention.
As shown in fig. 9, the estimation apparatus 100 for the outer frame of the rectangular object includes a data processing module 30 and a calculation module 40.
And the data processing module 30 is used for obtaining a rectangular convex hull under the aerial view based on the three-dimensional point cloud of the rectangular object, and connecting points at the edge in the convex hull to obtain a first frame.
The calculation module is used for respectively connecting each point in the convex hull with the origin of coordinates to obtain a plurality of line segments, wherein a first endpoint of each line segment is a point located in the convex hull, a second endpoint is the origin of coordinates, and the origin of coordinates is a point represented by the laser radar 20; traversing a plurality of line segments, if the line segments in the plurality of line segments have intersection points with the first frame besides the first end points, removing the first end points of the line segments in the convex hull, and connecting points at the edges in the rest points in the convex hull to obtain a second frame; calculating the area of the second frame; and connecting two points which are farthest away in the second frame to serve as the hypotenuse of the first right-angle triangle.
In a preferred embodiment, the apparatus further comprises a lidar 20 and a data pre-processing module 10.
And the laser radar 20 scans the visual range to acquire a depth map.
The data preprocessing module 10 fits the ground based on preset height information, and removes the ground from the depth map to obtain a first point set; clustering the first point set by using a region growing method to obtain a second point set; and obtaining a three-dimensional point cloud of the rectangular object based on the second point set.
In one embodiment, the data preprocessing module 10, obtaining the three-dimensional point cloud of the rectangular object based on the second point set, includes: screening the second point set to obtain a point set of a rectangular object on the depth map; and converting the point set of the rectangular object on the depth map into a three-dimensional point cloud of the rectangular object under the space rectangular coordinate.
In one embodiment, the step of calculating the first right triangle by the calculation module 40 based on the area and the hypotenuse comprises: calculating to obtain a plurality of right triangles and coordinates of right angle points thereof based on the areas and the hypotenuses; and taking the right-angle triangle with the highest coincidence degree with the second frame in the plurality of right-angle triangles as the first right-angle triangle.
In one embodiment, the step of obtaining the outer frame of the rectangular object by the calculation module 40 based on the first right triangle includes: and rotating the first right-angle triangle by taking the middle point of the oblique edge as a central symmetry point to obtain the outer frame of the rectangular object.
In a preferred embodiment, the above apparatus further comprises: and a correction module.
The correction module makes a perpendicular line from the right-angle point of the first right-angle triangle to the hypotenuse thereof, and divides the three-dimensional point cloud of the vehicle into two parts; performing RANSAC straight line fitting on the two point clouds respectively based on two right-angle sides of the first right-angle triangle to obtain a first characteristic value and a second characteristic value respectively, wherein the first characteristic value is the maximum characteristic value of an inner point of the first straight line, and the second characteristic value is the maximum characteristic value of an inner point of the second straight line; if the first characteristic value is larger than the second characteristic value, determining that the first straight line is a long right-angle side of the first right-angle triangle; and if the second characteristic value is larger than the first characteristic value, determining that the second straight line is the long right-angle side of the first right-angle triangle.
In one embodiment, the step of correcting the first straight line or the second straight line by the correction module comprises: respectively comparing the first straight line and the second straight line with the three-dimensional point cloud of the rectangular object to respectively obtain a first confidence value and a second confidence value; wherein the confidence value is in direct proportion to the coincidence rate of the three-dimensional point cloud of the straight line and the rectangular object; the first line or the second line is corrected based on a difference between the first confidence value and the second confidence value.
In one embodiment, the correction module 50 determines that the second segment is corrected if the first confidence value is greater than the second confidence value.
Specifically, the step of the correction module 50 correcting the second straight line includes: calculating the slope of a second straight line based on the slope of the first straight line; performing RANSAC straight line fitting on the inner point of the second straight line based on the slope of the second straight line to obtain an equation of the second straight line; obtaining a second right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses; and obtaining the outer frame of the corrected rectangular object based on the second right triangle.
Specifically, since the first straight line, the second straight line, and the hypotenuse form a right triangle, the slope of the second straight line and the slope of the first straight line have an inverse relationship with each other. The slope of the second line is equal to 1 divided by the slope of the first line.
In one embodiment, the modification module 50 determines to modify the first segment if the first confidence value is less than the second confidence value.
Specifically, the step of correcting the first straight line includes: calculating the slope of the first straight line based on the slope of the second straight line; performing RANSAC straight line fitting on an inner point of the first straight line based on the slope of the first straight line to obtain an equation of the first straight line; determining a second right-angled triangle corrected for the first right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses; and obtaining the outer frame of the corrected rectangular object based on the second right triangle.
The technical scheme of the invention has the following beneficial technical effects:
(1) laser radars are widely used in the field of autonomous driving due to their high accuracy and long effective detection distance. Compared with a millimeter wave radar, the laser radar can realize accurate modeling of surrounding environment, pedestrians and vehicles. In the aspect of vehicle target tracking, an accurate judgment on the position and the size of a vehicle is required. According to the method, the point cloud of the vehicle is obtained through the laser radar, the vehicle model is subjected to rapid frame estimation by using a geometric method, and an accurate vehicle rectangular outer frame is obtained through further correction, so that the vehicle can be tracked rapidly and stably.
(2) According to the rectangular frame estimation method and device provided by the embodiment of the invention, on one hand, the points in the convex hull of the plane of the vehicle are processed by a virtual scanning line method, and the wild points belonging to the parts such as the vehicle roof, the engine hood, the trunk and the like are removed, so that the processing amount of the points can be reduced, the calculation speed is increased, and the calculation accuracy is also improved in the calculation process; on the other hand, the laser radar has universality for the point cloud scanned by the front vehicle in an L shape, the rectangular right-angled triangle is obtained by utilizing the characteristic of the laser radar, the rectangular frame can be quickly obtained, and compared with the prior art that the rectangular model is adopted to process the point cloud in the L shape, the robustness is stronger.
(3) According to the rectangular frame calculating method and device provided by the embodiment of the invention, RANSAC is adopted to further correct the initially obtained frame of the rectangular object, and compared with the prior art, the frame calculating method and device are easier to converge and have fewer convergence steps during iterative calculation, so that compared with the prior art, the correction process is more stable, the calculation speed is higher, and the accuracy is higher.
A fourth embodiment of the present invention provides a computer storage medium having a computer program stored thereon, where the computer program, when executed by a processor, implements the steps of the method for estimating the outline of a rectangular object provided in the first embodiment.
A fifth embodiment of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for estimating the outline of the rectangular object according to the first embodiment when executing the computer program.
It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explaining the principles of the invention and are not to be construed as limiting the invention. Therefore, any modification, equivalent replacement, improvement and the like made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. Further, it is intended that the appended claims cover all such variations and modifications as fall within the scope and boundaries of the appended claims or the equivalents of such scope and boundaries.
Claims (24)
1. A method for estimating the outer frame of a rectangular object is characterized by comprising the following steps:
based on the three-dimensional point cloud of the rectangular object, under a bird's eye view, obtaining a convex hull of the rectangle, and connecting points at the edge in the convex hull to obtain a first frame;
respectively connecting each point in the convex hull with a coordinate origin to obtain a plurality of line segments, wherein a first endpoint of each line segment is a point located in the convex hull, and a second endpoint is the coordinate origin;
if the line segment in the line segments has an intersection point with the first frame except the first endpoint, removing the first endpoint of the line segment in the convex hull, and connecting points at the edge in the rest points in the convex hull to obtain a second frame;
calculating the area of the second frame;
connecting two points farthest away in the second frame to serve as the hypotenuse of the first right-angled triangle;
calculating to obtain the first right-angle triangle based on the area and the hypotenuse;
and obtaining the outer frame of the rectangular object based on the first right-angle triangle.
2. The method of claim 1, wherein the step of obtaining a three-dimensional point cloud of the rectangular object comprises:
acquiring a depth map of the laser radar;
fitting the ground based on preset height information, and removing the ground from the depth map to obtain a first point set;
clustering the first point set by using a region growing method to obtain a second point set;
obtaining a three-dimensional point cloud of the rectangular object based on the second point set.
3. The method of claim 2, wherein obtaining the three-dimensional point cloud of the rectangular object based on the second set of points comprises: screening the second point set to obtain a point set of the rectangular object on the depth map;
and converting the point set of the rectangular object on the depth map into a three-dimensional point cloud of the rectangular object under a space rectangular coordinate.
4. The method of claim 1, wherein the step of calculating the first right triangle based on the area and the hypotenuse comprises:
calculating to obtain a plurality of right triangles and coordinates of right angle points thereof based on the areas and the hypotenuses;
and taking the right-angle triangle with the highest coincidence degree with the second frame in the plurality of right-angle triangles as the first right-angle triangle.
5. The method of claim 1, wherein the step of obtaining an outer frame of the rectangular object based on the first right triangle comprises:
and rotating the first right-angle triangle by taking the middle point of the oblique edge as a central symmetry point to obtain the outer frame of the rectangular object.
6. The method of claim 1, further comprising the step of modifying the outer border of the rectangular object by:
making a perpendicular line from the right-angle point of the first right-angle triangle to the hypotenuse of the first right-angle triangle, and dividing the point cloud in the second frame into two parts;
performing RANSAC straight line fitting on the two point clouds respectively based on the two right-angle sides of the first right-angle triangle to respectively obtain a first straight line and a second straight line;
and correcting the first straight line or the second straight line.
7. The method of claim 6, wherein performing RANSAC line fitting on the two point clouds based on the two edges of the first right triangle respectively comprises:
performing principal component analysis on the inner point of the first straight line and the inner point of the second straight line obtained by RANSAC respectively to obtain a first characteristic value and a second characteristic value, wherein the characteristic values are in direct proportion to the length; the first characteristic value is the maximum characteristic value of the inner point of the first straight line, and the second characteristic value is the maximum characteristic value of the inner point of the second straight line;
if the first characteristic value is greater than the second characteristic value, determining that the first straight line is a long right-angle side of the first right-angle triangle, and the second straight line is a short right-angle side of the first right-angle triangle;
and if the second characteristic value is greater than the first characteristic value, determining that the first straight line is the short right-angle side of the first right-angle triangle, and the second straight line is the long right-angle side of the first right-angle triangle.
8. The method of claim 6, wherein the step of modifying the first line or the second line comprises:
comparing the first straight line and the second straight line with the three-dimensional point cloud of the rectangular object respectively to obtain a first confidence value and a second confidence value respectively; wherein the confidence value is in direct proportion to the coincidence rate of the three-dimensional point cloud of the straight line and the rectangular object;
and correcting the first straight line or the second straight line based on the difference value of the first confidence value and the second confidence value.
9. The method of claim 8, wherein the second segment is modified if the first confidence value is greater than the second confidence value;
and if the first confidence value is smaller than the second confidence value, correcting the first line segment.
10. The method according to claim 8 or 9, wherein the step of correcting the second straight line comprises:
calculating the slope of the second straight line based on the slope of the first straight line;
performing RANSAC straight line fitting on the inner point of the second straight line based on the slope of the second straight line to obtain an equation of the second straight line;
obtaining a second right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses;
and obtaining the corrected outer frame of the rectangular object based on the second right triangle.
11. The method of claim 8 or 9, wherein the step of modifying the first line comprises:
calculating the slope of the first straight line based on the slope of the second straight line;
performing RANSAC straight line fitting on an inner point of the first straight line based on the slope of the first straight line to obtain an equation of the first straight line;
determining a second right triangle corrected for the first right triangle according to the equation of the first straight line, the equation of the second straight line and three intersection points of the hypotenuses;
and obtaining the corrected outer frame of the rectangular object based on the second right triangle.
12. An apparatus for estimating an outer frame of a rectangular object, comprising:
the data processing module (30) is used for obtaining a rectangular convex hull under a bird's eye view based on the three-dimensional point cloud of the rectangular object, and connecting points at the edge in the convex hull to obtain a first frame;
the calculation module (40) is used for respectively connecting each point in the convex hull with the origin of coordinates to obtain a plurality of line segments; the first endpoint of each line segment is a point in the convex hull, and the second endpoint is the origin of the coordinate system; if the line segment in the line segments has an intersection point with the first frame except the first endpoint, removing the first endpoint of the line segment in the convex hull, and connecting points at the edge in the rest points in the convex hull to obtain a second frame;
calculating the area of the second frame;
connecting two points which are farthest away in the second frame and are adjacent to each other to form a connecting line, wherein the connecting line is used as the hypotenuse of the first right-angled triangle;
calculating to obtain a first right-angle triangle based on the area and the hypotenuse;
and obtaining the outer frame of the rectangular object based on the first right-angle triangle.
13. The apparatus of claim 12, further comprising a lidar (20) and a data pre-processing module (10);
the laser radar (20) scans the visual range to acquire a depth map;
the data preprocessing module (10) fits the ground based on preset height information, and removes the ground from the depth map to obtain a first point set;
clustering the first point set by using a region growing method to obtain a second point set;
obtaining a three-dimensional point cloud of the rectangular object based on the second point set.
14. The apparatus according to claim 13, wherein the data preprocessing module (10), the step of obtaining a three-dimensional point cloud of the rectangular object based on the second set of points comprises:
screening the second point set to obtain a point set of the rectangular object on the depth map;
and converting the point set of the rectangular object on the depth map into a three-dimensional point cloud of the rectangular object under the space rectangular coordinate.
15. The apparatus of claim 12, wherein the calculating module (40), wherein the step of calculating the first right triangle based on the area and the hypotenuse comprises:
calculating to obtain a plurality of right triangles and coordinates of right angle points thereof based on the areas and the hypotenuses;
and taking the right-angle triangle with the highest coincidence degree with the second frame in the plurality of right-angle triangles as the first right-angle triangle.
16. The apparatus according to claim 12, wherein the computing module (40), based on the first right triangle, the step of obtaining the outer frame of the rectangular object comprises:
and rotating the first right-angle triangle by taking the middle point of the oblique edge as a central symmetry point to obtain the outer frame of the rectangular object.
17. The apparatus of claim 12, further comprising:
the correction module is used for making a perpendicular line from the right-angle point of the first right-angle triangle to the hypotenuse of the first right-angle triangle and dividing the three-dimensional point cloud of the vehicle into two parts;
performing RANSAC straight line fitting on the two point clouds respectively based on the two right-angle sides of the first right-angle triangle to respectively obtain a first straight line and a second straight line;
and correcting the first straight line or the second straight line.
18. The apparatus of claim 17, wherein the correction module performs RANSAC line fitting on the two point clouds based on the two edges of the first right triangle respectively comprises:
performing principal component analysis on an inner point of a first straight line and an inner point of a second straight line obtained by RANSAC respectively to obtain a first characteristic value and a second characteristic value, wherein the characteristic values are in direct proportion to the length, the first characteristic value is the maximum characteristic value of the inner point of the first straight line, and the second characteristic value is the maximum characteristic value of the inner point of the second straight line;
if the first characteristic value is larger than the second characteristic value, determining that a first straight line is a long right-angle side of the first right-angle triangle, and the second straight line is a short right-angle side of the first right-angle triangle;
and if the second characteristic value is greater than the first characteristic value, determining that the first straight line is the short right-angle side of the first right-angle triangle, and the second straight line is the long right-angle side of the first right-angle triangle.
19. The apparatus of claim 17, wherein the step of modifying the first line or the second line by a modification module comprises:
comparing the first straight line and the second straight line with the three-dimensional point cloud of the rectangular object respectively to obtain a first confidence value and a second confidence value respectively; wherein the confidence value is in direct proportion to the coincidence rate of the three-dimensional point cloud of the straight line and the rectangular object;
and correcting the first straight line or the second straight line based on the difference value of the first confidence value and the second confidence value.
20. The apparatus of claim 19, wherein the correction module determines: if the first confidence value is greater than the second confidence value, correcting the second line segment;
and if the first confidence value is smaller than the second confidence value, correcting the first line segment.
21. The apparatus of claim 19 or 20, wherein the step of modifying the second line by the modifying module comprises:
calculating the slope of the second straight line based on the slope of the first straight line;
performing RANSAC straight line fitting on the inner point of the second straight line based on the slope of the second straight line to obtain an equation of the second straight line;
obtaining a second right-angled triangle according to the equation of the first straight line, the equation of the second straight line and the three intersection points of the hypotenuses;
and obtaining the corrected outer frame of the rectangular object based on the second right triangle.
22. The apparatus of claim 19 or 20, wherein the step of modifying the first line by a modification module comprises:
calculating the slope of the first straight line based on the slope of the second straight line;
performing RANSAC straight line fitting on an inner point of the first straight line based on the slope of the first straight line to obtain an equation of the first straight line;
determining a second right triangle corrected for the first right triangle according to the equation of the first straight line, the equation of the second straight line and three intersection points of the hypotenuses;
and obtaining the corrected outer frame of the rectangular object based on the second right triangle.
23. A computer storage medium, characterized in that the storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the method for estimating the outline of a rectangular object according to any one of claims 1 to 11.
24. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method for estimating the outline of a rectangular object according to any one of claims 1 to 11.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811566476 | 2018-12-19 | ||
| CN2018115664766 | 2018-12-19 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111337939A true CN111337939A (en) | 2020-06-26 |
Family
ID=71185771
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910174238.9A Pending CN111337939A (en) | 2018-12-19 | 2019-03-08 | Method and device for estimating outer frame of rectangular object |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111337939A (en) |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| NO993749D0 (en) * | 1998-08-04 | 1999-08-03 | Japan Radio Co Ltd | Three-dimensional radar apparatus and method for producing three-dimensional radar image |
| GB0307652D0 (en) * | 2003-04-02 | 2003-05-07 | Canon Europa Nv | Generating texture maps for use in 3D computer graphics |
| JP2004102402A (en) * | 2002-09-05 | 2004-04-02 | Hitachi Software Eng Co Ltd | Partition data creating method and device |
| CN101334811A (en) * | 2007-06-27 | 2008-12-31 | 中国科学院微电子研究所 | Method for cutting CIF format polygon into PG3600 format rectangle |
| CN101661741A (en) * | 2008-08-29 | 2010-03-03 | 富士通株式会社 | Method and device for traversing triangle in graphical raster scanning |
| CN102915560A (en) * | 2012-09-21 | 2013-02-06 | 中国石油大学(华东) | Threshold-irrelative point cloud filtering method and device for airborne laser radar |
| US20130246020A1 (en) * | 2012-03-15 | 2013-09-19 | GM Global Technology Operations LLC | BAYESIAN NETWORK TO TRACK OBJECTS USING SCAN POINTS USING MULTIPLE LiDAR SENSORS |
| CN105388465A (en) * | 2015-12-17 | 2016-03-09 | 西安电子科技大学 | Sea clutter simulation method based on sea wave spectrum model |
| CN105572687A (en) * | 2015-12-11 | 2016-05-11 | 中国测绘科学研究院 | Method for manufacturing building digital line map based on vehicle-mounted laser radar point cloud |
| CN106291506A (en) * | 2016-08-16 | 2017-01-04 | 长春理工大学 | Vehicle target recognition methods based on single line cloud data machine learning and device |
| JP2017078989A (en) * | 2015-10-21 | 2017-04-27 | 株式会社パスコ | Image processing device, image processing method, and program |
| CN107656287A (en) * | 2017-10-30 | 2018-02-02 | 中国科学院合肥物质科学研究院 | A kind of Boundary Extraction device and method of the crudefiber crop row based on laser radar |
| CN108828621A (en) * | 2018-04-20 | 2018-11-16 | 武汉理工大学 | Obstacle detection and road surface partitioning algorithm based on three-dimensional laser radar |
| CN108873943A (en) * | 2018-07-20 | 2018-11-23 | 南京奇蛙智能科技有限公司 | A kind of image processing method that unmanned plane Centimeter Level is precisely landed |
-
2019
- 2019-03-08 CN CN201910174238.9A patent/CN111337939A/en active Pending
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| NO993749D0 (en) * | 1998-08-04 | 1999-08-03 | Japan Radio Co Ltd | Three-dimensional radar apparatus and method for producing three-dimensional radar image |
| JP2004102402A (en) * | 2002-09-05 | 2004-04-02 | Hitachi Software Eng Co Ltd | Partition data creating method and device |
| GB0307652D0 (en) * | 2003-04-02 | 2003-05-07 | Canon Europa Nv | Generating texture maps for use in 3D computer graphics |
| CN101334811A (en) * | 2007-06-27 | 2008-12-31 | 中国科学院微电子研究所 | Method for cutting CIF format polygon into PG3600 format rectangle |
| CN101661741A (en) * | 2008-08-29 | 2010-03-03 | 富士通株式会社 | Method and device for traversing triangle in graphical raster scanning |
| US20130246020A1 (en) * | 2012-03-15 | 2013-09-19 | GM Global Technology Operations LLC | BAYESIAN NETWORK TO TRACK OBJECTS USING SCAN POINTS USING MULTIPLE LiDAR SENSORS |
| CN102915560A (en) * | 2012-09-21 | 2013-02-06 | 中国石油大学(华东) | Threshold-irrelative point cloud filtering method and device for airborne laser radar |
| JP2017078989A (en) * | 2015-10-21 | 2017-04-27 | 株式会社パスコ | Image processing device, image processing method, and program |
| CN105572687A (en) * | 2015-12-11 | 2016-05-11 | 中国测绘科学研究院 | Method for manufacturing building digital line map based on vehicle-mounted laser radar point cloud |
| CN105388465A (en) * | 2015-12-17 | 2016-03-09 | 西安电子科技大学 | Sea clutter simulation method based on sea wave spectrum model |
| CN106291506A (en) * | 2016-08-16 | 2017-01-04 | 长春理工大学 | Vehicle target recognition methods based on single line cloud data machine learning and device |
| CN107656287A (en) * | 2017-10-30 | 2018-02-02 | 中国科学院合肥物质科学研究院 | A kind of Boundary Extraction device and method of the crudefiber crop row based on laser radar |
| CN108828621A (en) * | 2018-04-20 | 2018-11-16 | 武汉理工大学 | Obstacle detection and road surface partitioning algorithm based on three-dimensional laser radar |
| CN108873943A (en) * | 2018-07-20 | 2018-11-23 | 南京奇蛙智能科技有限公司 | A kind of image processing method that unmanned plane Centimeter Level is precisely landed |
Non-Patent Citations (1)
| Title |
|---|
| 王平 等: ""基于轮廓特征点凹性分析的遮挡车辆分割算法"", 《科学技术与工程》, 18 January 2018 (2018-01-18), pages 290 - 295 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110031824B (en) | Laser radar combined calibration method and device | |
| CN106780524B (en) | Automatic extraction method for three-dimensional point cloud road boundary | |
| US8199977B2 (en) | System and method for extraction of features from a 3-D point cloud | |
| Kang et al. | Automatic targetless camera–lidar calibration by aligning edge with gaussian mixture model | |
| CN109146948B (en) | Crop growth phenotype parameter quantification and yield correlation analysis method based on vision | |
| CN109752701A (en) | A kind of road edge detection method based on laser point cloud | |
| CN110673107B (en) | Road edge detection method and device based on multi-line laser radar | |
| CN109635816B (en) | Lane line generation method, apparatus, device, and storage medium | |
| WO2020168685A1 (en) | Three-dimensional scanning viewpoint planning method, device, and computer readable storage medium | |
| Cheng et al. | Building boundary extraction from high resolution imagery and lidar data | |
| WO2022179094A1 (en) | Vehicle-mounted lidar external parameter joint calibration method and system, medium and device | |
| CN113917487A (en) | Laser radar-based method for detecting road edge and drivable area of closed road | |
| CN113096181B (en) | Method and device for determining equipment pose, storage medium and electronic device | |
| Xinmei et al. | Passive measurement method of tree height and crown diameter using a smartphone | |
| CN112232248B (en) | Method and device for extracting plane features of multi-line LiDAR point cloud data | |
| CN116206094B (en) | Fan blade angle measuring method, device and system and electronic equipment | |
| CN111337939A (en) | Method and device for estimating outer frame of rectangular object | |
| CN113205563B (en) | Automatic driving sensor combined calibration target and calibration method | |
| CN112508970B (en) | Point cloud data segmentation method and device | |
| CN112017290B (en) | Conical surface fitting method and device, computer equipment and storage medium | |
| CN113627569A (en) | Data fusion method for radar video all-in-one machine used for traffic large scene | |
| CN112884026A (en) | Image recognition assisted power transmission line laser LiDAR point cloud classification method | |
| Douillard et al. | Fft-based terrain segmentation for underwater mapping | |
| Fan et al. | Noncontact measurement of indoor objects with 3D laser camera-based | |
| CN115147612B (en) | Processing method for estimating vehicle size in real time based on accumulated point cloud |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |