CN115690508A - Detection target obtaining method, medium and equipment based on Hilbert curve coding - Google Patents
Detection target obtaining method, medium and equipment based on Hilbert curve coding Download PDFInfo
- Publication number
- CN115690508A CN115690508A CN202211394270.6A CN202211394270A CN115690508A CN 115690508 A CN115690508 A CN 115690508A CN 202211394270 A CN202211394270 A CN 202211394270A CN 115690508 A CN115690508 A CN 115690508A
- Authority
- CN
- China
- Prior art keywords
- target
- point cloud
- coding
- code
- hilbert curve
- 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
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a detection target obtaining method, medium and equipment based on Hilbert curve coding. The method comprises the following steps: and performing Hilbert curve space coding processing on the background point cloud set to obtain a background point cloud coding set A. And performing Hilbert curve space coding processing on the cloud set of the point to be measured to obtain a point cloud coding set B to be measured. According to a preset precision value c 1 Determining a target spatial error value E m . Respectively determining a spatial error value E from A and B m First target coding subset a of m With a second target coding subset B m . B is to be m Each code in (1) and (A) m The codes in (1) are matched one by one. B is to be m All the codes which fail to be matched are put into the same set to obtain a target code set. Due to the present applicationThe data volume used in calculation is less, and the algorithm is simpler and more efficient, so that the calculation efficiency can be greatly improved, and the calculation result can be output more timely to improve the real-time property.
Description
Technical Field
The present invention relates to the field of target detection, and in particular, to a detection target acquisition method, medium, and device based on hilbert curve coding.
Background
With the development of technical points, the target detection technology is more mature and widely applied to various fields of life and production, such as the field of automatic driving. The existing method for acquiring the detection target comprises the following steps: the convolutional neural network is utilized to identify and detect the 3D information in the corresponding scene, and a corresponding detection target can be found after a large amount of convolutional operations are carried out in the convolutional neural network. Furthermore, the purpose of indiscriminate detection of objects at any position in any direction in the whole three-dimensional space is achieved. However, it is limited by the computational complexity of the algorithm itself, the computing power of the hardware device, and the complexity of the image information in the actual application scene. When the amount of 3D information data input into the convolutional neural network is extremely large, a large amount of time and resources are consumed for calculation, and a detection result cannot be output in time.
In the field of airport apron safety monitoring, the size of an aircraft is huge, the airport apron scene is complex, and numerous interferents such as lamp posts, gallery bridge fixed ends, gallery bridge moving ends and the like exist on the airport apron. Therefore, the 3D information data volume generated in the scene is extremely large, the existing detection method is difficult to output the detection result in time, and the real-time performance is low.
Disclosure of Invention
Aiming at the technical problems, the technical scheme adopted by the invention is as follows:
according to an aspect of the present invention, there is provided a detection target acquiring method based on hilbert curve coding, including the following steps:
the method comprises the steps of obtaining a background point cloud set of a target area, wherein the background point cloud set comprises a point cloud subset of a fixed target in a background and a position point cloud set of at least one movable target, and the position point cloud set comprises at least one position point cloud subset. The position point cloud subset is a point cloud set corresponding to the corresponding movable target at any position in a preset moving path.
Hilbert is performed on background point cloud setsSpecial curve space coding processing to obtain background point cloud coding set A = (A) 1 ,A 2 ,…,A i ,…,A n ) Wherein A is i And segmenting the target area for the ith-order Hilbert curve to obtain a background coding subset corresponding to the background point cloud. i =1,2, \8230;, n, n is the total number of background coding subsets.
And acquiring a cloud set of points to be measured of the target area, wherein the cloud set of points to be measured comprises a background point cloud set and a target point cloud set to be identified, and the target point cloud set to be identified comprises at least one target point cloud subset corresponding to the target to be identified.
Performing Hilbert curve space coding processing on the cloud set of the point to be measured to obtain a point cloud coding set B = (B) 1 ,B 2 ,…,B i ,…,B n ). Wherein, B i And (4) segmenting the target area for the ith-order Hilbert curve to obtain a real-time coding subset corresponding to the point cloud to be detected.
According to a preset precision value c 1 Determining a target spatial error value E m . Wherein E is m <c 1 And c is a 1 -E m ≤Y 1 。Y 1 Is a first threshold. E m And after the target region is segmented for the mth-order Hilbert curve, constructing the distance between the points on any two adjacent Hilbert curves. m is an element of [1, n ]]。
Respectively determining a spatial error value E from A and B m First target coding subset a of m And a second target encoding subset B m 。
B is to be m Each code in (1) and (A) m The codes in (1) are matched one by one. When B is present m Any one of (1) and A m When all codes in (1) are different, the code matching fails.
B is to be m All the codes which fail to be matched are put into the same set to obtain a target code set.
According to a second aspect of the present invention, there is provided a non-transitory computer-readable storage medium storing a computer program, which when executed by a processor, implements a hilbert curve coding-based detection target acquisition method as described above.
According to a third 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 implements the above-mentioned detection target acquisition method based on hilbert curve coding when executing the computer program.
The invention has at least the following beneficial effects:
the method comprises the steps of respectively carrying out spatial coding on background point clouds of a target area in an airport environment through a space coding method of a Hilbert curve to obtain a background point cloud coding set; and carrying out spatial encoding on the point cloud to be detected in the target area (namely the real-time environment point cloud of the target area) to obtain a point cloud encoding set to be detected. The space coding method of the Hilbert curve used in the method is used for coding the point cloud in the target area. When the hilbert curve is used to encode the space, each of the hilbert curves in each step has a corresponding cubic spatial region, and the cubic spatial region corresponding to the code in each step is gradually reduced as the order of the hilbert curve increases. Correspondingly, as the order of the hilbert curve increases, the number of cubic spatial regions existing in the target region increases, so that the target region can be divided more finely, and the accuracy of correspondence between the coding and the spatial position can be improved. Moreover, since each code corresponds to a cubic space region, one code can represent all point clouds in the cubic space region. Therefore, the data volume can be greatly reduced by encoding the point cloud by the method.
Meanwhile, by setting Y in this application 1 A first target code subset a with a spatial error value closest to a predetermined precision value may be determined m With a second target coding subset B m . Therefore, on the basis of ensuring the calculation accuracy, the order of the Hilbert curve can be further reduced, namely, the number of codes can be further reduced, and the data volume used in calculation is further reduced.
In addition, the point cloud coding set to be detected and the codes in the background point cloud coding set are matched one by one, and all the codes which fail to be matched are put into the same set, so that the target coding set is obtained. The target code set is the detection target to be searched, thereby completing the acquisition work of the detection target. Because the algorithm of the matching calculation of the codes is simpler, the calculation efficiency is higher, and the calculation result can be output more timely.
In conclusion, the data volume used in the calculation is small, and the algorithm is simpler and more efficient, so that the calculation efficiency can be greatly improved, and the calculation result can be output more timely to improve the real-time performance.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a flowchart of a detection target obtaining method based on hilbert curve coding according to an embodiment of the present invention.
Fig. 2 is a flowchart of multi-step splitting a target square plane based on a hilbert curve according to an embodiment of the present invention.
Reference numerals
1. A first order construction point; 2. a second order construction point; 3. three-order construction points; 4. the target square plane.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As an embodiment of the present invention, a detection target acquiring method based on hilbert curve coding is provided, as shown in fig. 1, the method includes the following steps:
step S100: acquiring a background point cloud set of a target area, wherein the background point cloud set comprises a point cloud subset of a fixed target in a background and a position point cloud set of at least one movable target, and the position point cloud set comprises at least one position point cloud subset; the position point cloud subset is a point cloud set corresponding to a corresponding movable target at any position in a preset moving path;
specifically, the target area may be a certain area in the apron, the background point cloud in the target area is concentrated, and the included fixed target may be a fixed building, a lamp post, a fixed end of a bridge, and the like. The movable targets included may be gallery bridge moving ends and some work vehicles.
When background point cloud collection is carried out, the method can be realized by the following steps
Step S101: determining a plurality of target time points; the target time points are a plurality of time points in the day, such as 30 minutes each from 0. The point cloud is collected by setting a plurality of target time points in one day, so that background point cloud data under different solar illumination conditions can be obtained.
Step S102: acquiring a background point cloud data frame of a target area at each target time point; the background point cloud data frame is a position point cloud set of a fixed target and a movable target in the background when the target time point is reached; preferably, the acquisition is by lidar. More preferably, the laser radar uses a laser radar with a visual range exceeding 450 meters, 180 lines and a 180-degree visual angle, the device can acquire point cloud data on the apron at a speed of 10 frames, and the data volume of each frame exceeds 100 ten thousand points, so that the acquired point cloud data can be ensured to be denser, further more detailed characteristics of an object to be identified can be acquired, and the later-stage calculation accuracy can be ensured.
Step S103: and generating a background point cloud set of the target area according to the plurality of background point cloud data frames.
Because the laser radar is more easily influenced and interfered by ambient light (mainly sunlight) in the process of collecting the background point cloud, the background point cloud data collected by the laser radar has errors. Therefore, the point cloud is collected by setting a plurality of target time points, and therefore background point cloud data under different solar illumination conditions can be obtained. Then, useless point cloud data is removed through the existing point cloud denoising method, and the actual part is completed, so that a more accurate background point cloud set can be obtained.
Step S200: performing Hilbert curve spatial coding processing on the background point cloud set to obtain a background point cloud coding set A = (A) 1 ,A 2 ,…,A i ,…,A n ) Wherein A is i Segmenting a target region for an ith-order Hilbert curve to obtain a background coding subset corresponding to background point cloud; i =1,2, \8230, n is the total number of background coding subsets;
the hilbert curve is a curve that can completely fill a square space. Specifically, the generation method of the hilbert curve of each step is the prior art, and is not described herein again. The generation of the multi-scale Hilbert curve is also recursive, for example, the construction process of the Hilbert curve of the order n is to generate the Hilbert curve of the order n-1 first, and then form the Hilbert curve of the order n after connecting 4 Hilbert curves of the order n-1 end to end.
The generation process of the first, second and third order Hilbert curves is illustrated by dividing the Hilbert curve into a target square plane 4, as shown in fig. 2. Specifically, the target square plane 4 is divided into four small squares through a horizontal center line and a vertical center line, then the center points of the four small squares are used as first-order construction points 1, and the 4 first-order construction points 1 are connected to form a first-order Hilbert curve. And then repeating the operation on squares corresponding to 4 first-order construction points 1 in the first-order Hilbert curves to generate second-order construction points 2 corresponding to 4 small squares respectively, simultaneously generating a first-order Hilbert curve corresponding to each small square, and connecting the 4 first-order Hilbert curves end to generate a second-order Hilbert curve. After the operations are repeated, a third-order construction point 3 corresponding to the third-order Hilbert curve can be generated, and a corresponding third-order Hilbert curve is generated.
Similarly, according to the above steps, a multi-order Hilbert curve in a direction perpendicular to the target square plane 4 can be generated, and a cube space can be divided by combining the Hilbert curves in the two directions perpendicular to each other.
Therefore, each order of Hilbert curve corresponds to a square space, and the square space is gradually reduced as the order of the Hilbert curve is increased. When encoding is performed, each construction point corresponds to one code, that is, each code corresponds to one square space.
And when the background point cloud exists in the corresponding square space, adding the code corresponding to the square space into the background point cloud code set, thereby establishing the background point cloud code set corresponding to each step of Hilbert curve. Similarly, the point cloud coding set to be measured corresponding to each step of the Hilbert curve can be obtained by the same method.
In addition, the hubert curve spatial coding method can be replaced by an octree spatial coding method. The gathering performance of the method of the space coding of the Burbert curve is better, so the precision of the coding is higher.
Step S300: acquiring a cloud set of points to be measured of a target area, wherein the cloud set of the points to be measured comprises a background point cloud set and a target point cloud set to be identified, and the target point cloud set to be identified comprises at least one target point cloud subset corresponding to the target to be identified;
step S400: performing Hilbert curve space coding processing on the cloud set of the point to be measured to obtain a point cloud coding set B = (B) 1 ,B 2 ,…,B i ,…,B n ) (ii) a Wherein, B i Dividing a target area for an ith-order Hilbert curve to obtain a real-time coding subset corresponding to the point cloud to be detected;
step S500: according to a preset precision value c 1 Determining a target spatial error value E m (ii) a Wherein E is m <c 1 And c is and c 1 -E m ≤Y 1 ；Y 1 Is a first threshold value; e m After the target area is segmented for the mth-order Hilbert curve, the distance between construction points on any two adjacent Hilbert curves; m is an element of [1, n ]]。
The method comprises the steps of respectively carrying out spatial coding on background point clouds of a target area in an airport environment through a space coding method of a Hilbert curve to obtain a background point cloud coding set; and carrying out spatial encoding on the point cloud to be detected in the target area (namely the real-time environment point cloud of the target area) to obtain a point cloud encoding set to be detected. The space coding method of the Hilbert curve used in the method is used for coding the point cloud in the target area. When the hilbert curve is used to encode the space, each of the hilbert curves in each step has a corresponding cubic spatial region, and the cubic spatial region corresponding to the code in each step is gradually reduced as the order of the hilbert curve increases. Correspondingly, as the order of the hilbert curve increases, the number of cube spatial regions existing in the target region increases, so that the target region can be divided more finely, and the accuracy of correspondence between the coding and the spatial position can be improved. Moreover, since each code corresponds to a cubic space region, one code can represent all point clouds in the cubic space region. Therefore, the data volume can be greatly reduced by encoding the point cloud by the method.
Meanwhile, by setting Y in this application 1 The first target code subset A with the spatial error value closest to the predetermined precision value can be determined m With a second target coding subset B m . Therefore, on the basis of ensuring the calculation accuracy, the order of the Hilbert curve can be further reduced, namely, the number of codes can be further reduced, and the data volume used in calculation is further reduced.
Step S600: respectively determining a spatial error value E from A and B m First target coding subset a of m With a second target coding subset B m ；
Step S700: b is to be m Each code in (1) and (A) m The codes in (1) are matched one by one; when B is present m Any one of (1) and A m When all codes in the code list are different, the code matching fails;
step S800: b is to be m All the codes which fail to be matched are put into the same set to obtain a target code set.
In addition, the point cloud coding set to be detected and the codes in the background point cloud coding set are matched one by one, and all the codes which fail to be matched are put into the same set, so that the target coding set is obtained. The target code set is the detection target to be searched, thereby completing the acquisition work of the detection target. Because the algorithm of the matching calculation of the codes is simpler, the calculation efficiency is higher, and the calculation result can be output more timely.
In conclusion, the data volume used in the calculation is small, and the algorithm is simpler and more efficient, so that the calculation efficiency can be greatly improved, and the calculation result can be output more timely to improve the real-time performance.
As another embodiment of the present invention, the method is applied to a first system, the first system includes a first buffer space and a second buffer space, and the first buffer space and the second buffer space are independent from each other;
the first cache space is used for conducting Hilbert curve space coding processing on the background point cloud set and storing a background point cloud coding set A;
the second cache space is used for conducting Hilbert curve space coding processing on the point cloud set to be measured obtained in real time and storing the point cloud coding set B to be measured at the corresponding moment.
In this embodiment, when the method is applied to the first system, two independent cache spaces need to be opened up in the memory. And the two cache spaces respectively carry out Hilbert curve space coding processing on the background point cloud sets and the cloud sets of the points to be measured obtained in real time so as to obtain and store corresponding coding sets. And then the later matching processing of the codes is convenient. Because the background point cloud set corresponding to one target area is usually fixed and unchangeable, the background point cloud set can be stored in the first cache space after being coded once, so that the background point cloud set can be used for later code matching. When the codes are matched at the later stage, the cloud sets of the background points do not need to be re-coded and calculated, so a large amount of computing resources can be distributed to the process of performing Hilbert curve space coding processing on the cloud sets of the points to be measured, and the efficiency of performing Hilbert curve space coding processing on the cloud sets of the points to be measured can be further improved; the calculation efficiency is further improved, and the calculation result can be output more timely, so that the real-time performance is improved.
As another embodiment of the present invention, step S500: according to a preset precision value c 1 Determining a target spatial error value E m The method comprises the following steps:
step S501: obtaining a spatial error value d corresponding to each-order Hilbert curve after the target area is divided 1 ,d 2 ,…,d i ,…,d n Wherein d is i A corresponding spatial error value is obtained after the target area is divided for the ith-order Hilbert curve;
step S502: whenever d is i ≤c 1 When d is determined i Is an initial target spatial error value;
step S503: from all initial target spatial error values d c1 ,d c2 ,…,d cv ,…,d cu In (1), determining a target spatial error value E m ，E m The following conditions are satisfied:
E m ＝Max(d c1 ,d c2 ,…,d cv ,…,d cu )；
wherein Max () is a maximum function; d cv Is the v-th initial target spatial error value; u is the total number of initial target spatial error values; v =1,2, \8230;, u, u ≦ n.
Preferably, d i The following conditions are satisfied:
wherein, X i1 、Y i1 And Z i1 The coordinate value of the space coordinate of the corresponding target construction point after the target area is divided for the ith-order Hilbert curve; x i2 、Y i2 And Z i2 The coordinate value of the space coordinate of any point adjacent to the target structure point on the ith-order Hilbert curve; the target construction point is any construction point on the ith-order Hilbert curve.
And taking the distance value between any two adjacent construction points on each step of Hilbert curve as a corresponding spatial error value. And selecting the order of the Hilbert curve corresponding to the largest selected spatial error value from the plurality of initial target spatial error values as the target order. And respectively determining a first target coding subset A corresponding to the target order from A and B m With a second target coding subset B m . Therefore, on the premise of ensuring that the precision requirement is met, the selected A is further reduced m And B m The number of the codes is reduced, and the data volume used in calculation is further reduced, so that the calculation efficiency can be greatly improved, and the calculation result can be output more timely to improve the real-time performance.
As another embodiment of the present invention, step S501: obtaining a spatial error value corresponding to each-order Hilbert curve after the target area is divided, wherein the spatial error value comprises the following steps:
step S511: establishing a mapping relation table of the Hilbert segmentation order of the target area and the corresponding spatial error value;
step S521: and determining a space error value corresponding to each-order Hilbert curve after the target area is divided according to the mapping relation table and the Hilbert dividing order.
In this embodiment, the mapping relationship table is used to determine the spatial error value corresponding to each hubert curve after the target region is divided, and the mapping relationship table may be prepared in advance and stored in the database.
As another embodiment of the present invention, after step S800, the method further comprises:
step S810: obtaining the space coordinate e corresponding to each code in the target code set 1 ,e 2 ,…,e b ,…,e q Wherein e is b The spatial coordinates corresponding to the b-th code in the target code set; b =1,2, \8230;, q, q is the total number of codes in the target code set;
step S820: to e 1 ,e 2 ,…,e b ,…,e q Performing point cloud segmentation processing to generate a plurality of fractal cloud clusters; each fractal cloud cluster consists of a plurality of points corresponding to the space coordinates;
step S830: determining a target fractal cloud cluster from the plurality of fractal cloud clusters; the number of corresponding spatial coordinates in the target fractal cloud cluster is greater than a second threshold.
And generating a point cloud picture corresponding to the target coding set in the same coordinate system according to the space coordinate corresponding to each code. At least one fractal cloud cluster with dense aggregation exists in the point cloud image, and each fractal cloud cluster can be segmented by the existing point cloud segmentation processing method. Because objects of different volume sizes will correspond to different numbers of fractal clouds. Therefore, when the difference between the point cloud number in a fractal cloud cluster and the volume of the corresponding target object is large, the fact that the point cloud number is the noise to be removed can be determined, therefore, the noise cloud cluster in the fractal cloud clusters can be removed by setting a second threshold, and the remaining point cloud is the target fractal cloud cluster.
Taking an airplane as an example, since vehicles which do not belong to the background point cloud, such as temporary rescue vehicles, may appear on the apron, the fractal cloud clusters corresponding to the rescue vehicles may exist in the determined fractal cloud clusters. However, as the volume difference between the rescue vehicle and the airplane is large, the number of point clouds contained in the fractal cloud clusters respectively corresponding to the rescue vehicle also has a large difference, and therefore the fractal cloud clusters corresponding to the rescue vehicle can be removed by setting a second threshold. And then can more quick noise removal to more accurate target fractal cloud group that obtains.
As an embodiment of the present application, there is provided a target attribute determining method based on hilbert coding, the method including the steps of:
step A100: generating a k-order Hilbert curve in the target region to obtain a structure point information set F = (F) of the target region 1 ,F 2 ,…,F j ,…,F k )，F j ＝(F j1 ,F j2 ,…,F ja ,…,F je ). Wherein, F j To construct a set of spatial coordinates of the construction points of the jth order Hilbert curve. k is the total order of the hilbert curve generated in the target region. F ja Is F j The spatial coordinates of the a-th construction point in (a). e is F j The total number of spatial coordinates of the construction points. f =4 × e. F is F j+1 The total number of spatial coordinates of the construction points.
Step A200: configuring a code for each construction point in F to generate a code set G = (G) for the target region 1 ,G 2 ,…,G j ,…,G k )，G j ＝(G j1 ,G j2 ,…,G ja ,…,G je ). Wherein G is j Is F j A corresponding code set. G ja Is F ja And (4) correspondingly coding.
Step A300: according to F and G, generating a mapping relation set H = (H) of the space coordinates and codes of the target area 1 ,H 2 ,…,H j ,…,H k ). Wherein H j And the first mapping relation table is used for mapping the space coordinates of the construction points corresponding to the Hilbert curve of the jth order to the codes.
Specifically, after each step of Hilbert curve is generated in the target area, all construction points on the step of Hilbert curve are coded, meanwhile, all construction point spatial position information on the step of Hilbert curve is obtained, and the same construction point spatial position information and the corresponding relation of the codes are put into the corresponding first mapping relation table. Repeating the above operations until the corresponding relation between the spatial position information and the codes of all the construction points on the Hilbert curve of the order is established, and at this time, generating a first mapping relation table corresponding to the Hilbert curve of the order. The spatial position information of the construction point may be spatial position information composed of longitude, latitude, and altitude of the construction point.
Step A400: and acquiring a target code set I of the target area, wherein the I comprises target codes corresponding to point clouds of all targets to be detected in the target area. The target code is obtained by coding the point cloud of the target to be detected in the target area by using a Hilbert curve. I is epsilon to F.
The target code included in I is also obtained by encoding the point cloud of the target to be measured in the target region by using the above encoding method through the hilbert curve, and thus I belongs to F.
In this step, the target code set of the target area is acquired according to the method from step S100 to step S800.
Step A500: and determining the space coordinate corresponding to each target code in the I according to the first mapping relation table in the H. To generate a target set of spatial coordinates J for the target region.
Step A600: and determining the attribute of the target to be detected according to the J.
Specifically, the attribute of the object to be measured corresponding to the coordinate group may be determined based on the shape of the coordinate groups formed by the spatial coordinates in J. The attributes may be determined based on attributes of the test object that may be present in the particular usage scenario. The description will be given taking an example of detecting an object on the apron as a usage scene, and the attribute of the target to be detected in the scene may be a large passenger aircraft, a small passenger aircraft, a vehicle, or the like.
The target coding set is obtained by coding point clouds of a target to be detected in a target area through a Hilbert curve. Since, when the hilbert curve is used to encode the space, each encoding in each order of the hilbert curve has a corresponding cubic space region, and as the order of the hilbert curve increases, the cubic space region corresponding to the encoding in each order gradually decreases. Correspondingly, as the order of the hilbert curve increases, the number of cubic spatial regions existing in the target region increases, so that the target region can be divided more finely, and the accuracy of correspondence between the coding and the spatial position can be improved. Moreover, since each code corresponds to a cubic space region, one code can represent all point clouds in the cubic space region. Therefore, the data volume can be greatly reduced by encoding the point cloud by the method. Therefore, the calculation efficiency can be greatly improved, and the calculation result can be output more timely so as to improve the real-time performance.
Meanwhile, in the prior art, the position coordinate corresponding to a certain code is determined by an inverse operation mode, but in the process of coding the space by the hilbert curve, each cube space region corresponds to one code, so that the position coordinate obtained by the inverse operation is the coordinate of any position in the cube space region, and is different from the actual position coordinate of the point cloud, so that a certain error exists. Compared with the prior art, the method and the device establish a mapping relation set of the space coordinates and the codes of the target area. Therefore, the space coordinate corresponding to each code can be found directly according to the mapping relation set. Since the spatial coordinates are actual coordinates of the point cloud, the spatial coordinates obtained by encoding and mapping the set of relationships are actual position coordinates of the point cloud, and are not coordinates of other positions in the cube spatial region. Therefore, the accuracy of the position coordinate corresponding to each code is improved, meanwhile, the calculation process of inverse operation is not needed, the speed of obtaining the space coordinate corresponding to the code through the mapping relation set is higher, the efficiency is higher, and the calculation result can be output more timely.
As another possible embodiment of the present application, after step a500, the method further includes:
step A510: and determining a point corresponding to the position in the target coordinate system according to each space coordinate in the J to generate a point cloud picture corresponding to the J.
Step A520: and performing point cloud segmentation processing on the point cloud picture to generate a plurality of fractal cloud clusters. Each fractal cloud cluster is formed by a plurality of points corresponding to the space coordinates.
Step A530: and determining a target fractal cloud cluster from the plurality of fractal clouds. The number of corresponding spatial coordinates in the target fractal cloud cluster is greater than a second threshold.
And generating a point cloud picture corresponding to the target space coordinate set J in the same coordinate system according to the space coordinate corresponding to each target code. At least one fractal cloud cluster with dense aggregation exists in the point cloud image, and each fractal cloud cluster can be segmented by the existing point cloud segmentation processing method. Because objects of different volume sizes will correspond to different numbers of fractal clouds. Therefore, when the difference between the point cloud number in a fractal cloud cluster and the volume of the corresponding target object is large, it can be determined that the point cloud number is the noise to be removed, so that the noise cloud cluster in the fractal cloud clusters can be rapidly removed by setting a second threshold, and the retained point cloud cluster is the target fractal cloud cluster.
Taking an airplane as an example, since vehicles which do not belong to background point clouds, such as temporary rescue vehicles, may appear on the apron, fractal clouds corresponding to the rescue vehicles may exist in the determined fractal clouds. However, the volumes of the rescue vehicle and the aircraft are greatly different, so that the number of point clouds contained in the fractal cloud clouds respectively corresponding to the rescue vehicle is also greatly different, and therefore the fractal cloud clouds corresponding to the rescue vehicle can be removed by setting a second threshold. And then can more quick noise removal to more accurate target fractal cloud group that obtains.
As another possible embodiment of the present application, step a600: determining the attribute of the target to be measured according to J, wherein the determining comprises the following steps:
step A601: and establishing a second mapping relation table between the total number of the space coordinates in the fractal cloud cluster and the attribute labels.
Step A602: and determining an attribute tag corresponding to each target fractal cloud cluster according to the second mapping relation table.
Step A603: and determining the attribute of the target to be detected corresponding to the target fractal cloud cluster according to the attribute tag corresponding to the target fractal cloud cluster.
Because the total number of the spatial coordinates in the fractal cloud cluster is generally in direct proportion to the volume size between the targets to be detected, the attribute of the targets to be detected can be judged by taking the total number of the spatial coordinates in the fractal cloud cluster as a characteristic value. Because the volume difference between the airplane and other objects on the apron is large, the total number of the spatial coordinates in the corresponding fractal clouds is also large, and the difference is obvious. Therefore, when the airplane is the target object to be measured, the range of the total number of the coordinates corresponding to the airplane can be determined more easily and accurately. And further, the accuracy of determining the attribute of the target to be detected corresponding to the target fractal cloud cluster can be improved.
According to one possible embodiment of the present invention, a method for clustering targets based on hilbert coding is provided, which includes the following steps:
step B100: acquiring a point cloud coding set M = (M) corresponding to an interested target in a target area g 1 ,M g 2 ,…,M g h ,…,M g z ). Wherein M is g h The target point cloud of the h-th laser point constituting the target of interest is encoded. Each target point cloud code in M is a point cloud code obtained by coding an interested target in a target area by a g-order Hilbert curve. h =1,2, \ 8230;, z, z is the total number of point cloud encodings in M. The object of interest comprises a plurality of sub-objects.
In this step, the point cloud code set corresponding to the target of interest in the target area is obtained, which may be obtained according to the method from step S100 to step S800.
Step B200: obtaining a tagging code corresponding to each target point cloud code in M to obtain a tagging code set N = (N) corresponding to M g-1 1 ,N g-1 2 ,…,N g-1 h ,…,N g-1 z ). Wherein N is g-1 h Is M g h And (5) corresponding tagging codes. Each labeling code is a g-1 order Hilbert code of the corresponding target point cloud code.
Step B300: according to N and M, generating a plurality of tagging subsets O corresponding to the interested target 1 ,O 2 ,…,O m ,…,O y ，O m ＝(O m 1 ,O m 2 ). Wherein, O m And the mth tagging subset corresponding to the target of interest. y is a label attached corresponding to the interested targetThe total number of sets. m =1,2, \ 8230, y, y ≦ z. O is m 1 Is O m And (4) corresponding tagging codes. O is m 2 Is O m And (4) corresponding point cloud coding arrays. O is m 2 ＝(M gm 1 ,M gm 2 ,…,M gm p ,…,M gm f(m) )。M gm p Is O m 2 The p-th target point cloud of (1) is encoded. p =1,2, \ 8230;, f (m), f (m) is O m 2 The total number of target point cloud encodings. f (m) is less than or equal to z. O is m 2 The tagging codes corresponding to each target point cloud code are all O m 1 。
And generating a plurality of tagging subsets corresponding to the interested target through the N and the M. And generating a plurality of space aggregation subsets corresponding to the interested target according to the tagging codes corresponding to each tagging subset. Therefore, when subsequent clustering processing is performed, processing is performed on the basis of the tagging codes. Meanwhile, the tagging codes are g-1 order Hilbert codes of corresponding target point cloud codes, so that the data size during clustering processing can be further reduced, and the calculation efficiency is further improved.
Step B400: generating a plurality of space aggregation subsets Q corresponding to the interested target according to the tagging codes corresponding to each tagging subset 1 ,Q 2 ,…,Q q ,…,Q x . Wherein Q is q A corresponding qth spatially clustered subset for the object of interest. x is the total number of spatially aggregated subsets corresponding to the object of interest. q =1,2, \ 8230, x, x ≦ y. Q q ＝(Q q 1 ,Q q 2 ,…,Q q r ,…,Q q f(q) ) And Q q 1 <Q q 2 <…<Q q r <…<Q q f(q) 。Q q r Is Q q The middle-r mark-giving code. r =1,2, \8230;, f (q), f (q) ≦ y. f (Q) is Q q The total number of marked codes in the code. Q q f(q) The following conditions are satisfied: q q f(q) -Q q 1 R is less than or equal to R. Wherein R is a preset distance radius. Preferably, the first and second liquid crystal materials are,R＝5。
the index code in this embodiment is decimal and is Q q 1 =50,r =5 as an example, the assigned codes included in the corresponding spatial aggregation subset have a value in the range of 50-55.
Specifically, step B400 includes the following steps:
step B401: and performing second matching processing on the tagging codes corresponding to the rest tagging subsets in the multiple pairs to generate multiple spatial aggregation subsets.
The second matching process includes:
step B402: and selecting the minimum tagging code in the current residual tagging subsets as a target tagging code T.
Step B403: and establishing a corresponding space aggregation subset, and storing the target tagging codes into the space aggregation subset.
Step B404: and matching the target tagging codes with the tagging codes corresponding to each of the remaining tagging subsets.
Step B405: and storing the tagging codes into the spatial aggregation subset whenever any remaining tagging codes and the target tagging code meet a first condition.
The first condition is: i T t R is less than or equal to-T |. Wherein, T t And coding any given mark in the current remaining mark subset.
Step B406: and deleting the tagging codes added into the spatial aggregation subset from the tagging codes corresponding to the current residual tagging subsets so as to update the tagging codes corresponding to the residual tagging subsets.
Repeating steps B402-B406 multiple times can rapidly cluster multiple tagged subsets. In this embodiment, due to the size of the difference between any two hilbert codes, the distance of the points corresponding to the two codes in space can be reflected. Therefore, whether g-1 order Hilbert codes corresponding to the tagging subsets are close in space can be quickly judged by utilizing the numerical proximity of the codes. I.e. will satisfy Q q f(q) -Q q 1 And placing a plurality of marking codes less than or equal to R into the same spatial aggregation subset. Because, in front of every second aircraft on the apronThe distance of the target point cloud is larger than the width of a fracture zone in the target point cloud, so that the point cloud corresponding to different airplanes can be segmented by controlling the size of the R, and the point cloud corresponding to the same airplane can be prevented from being separated from a fracture position and clustered into different targets respectively, so that the accuracy of the finally obtained clustering result can be improved.
Step B500: obtaining a spatial aggregation coding set S corresponding to each spatial aggregation subset 1 ,S 2 ,…,S q ,…,S x . Wherein S is q Is Q q A corresponding spatially aggregated encoded set. S q ＝(S q 1 ,S q 2 ,…,S q r ,…,S q f(q) )。S q r Is Q q r Corresponding spatial aggregation coding. Each spatial aggregation code is a g-2 order hilbert code of the corresponding target point cloud code.
Step B600: and performing data cleaning processing on each space aggregation coding set to obtain a target point cloud coding set corresponding to each sub-target.
Step B700: and determining the attribute of the sub-target corresponding to each target point cloud coding set according to the number of the target point cloud codes in each target point cloud coding set.
The data cleaning process includes:
step B601: and acquiring the total number of codes of each spatial aggregation code in the spatial aggregation code set.
Step B601 may be implemented by:
step B6011: acquiring a first space aggregation code in a current space aggregation code set as a target space aggregation code;
step B6012: recording the total number of codes corresponding to the target space aggregation codes as 0;
step B6013: matching the target space aggregation code with each unmarked space aggregation code in the current space aggregation code set;
step B6014: after matching each same space aggregation code, adding 1 to the total number of codes corresponding to the target space aggregation code, and adding a mark to the corresponding space aggregation code;
step B6015: step B6011-step B6014 are repeatedly performed to obtain the total number of codes of each spatially aggregated code in the spatially aggregated code set.
Step B602: and removing the spatial aggregation codes corresponding to the total number of codes from the spatial aggregation code set when the total number of codes is smaller than the cleaning threshold value. So as to obtain an initial target point cloud encoding set corresponding to the spatial aggregation encoding set.
Step B603: and adding the target point cloud codes in the tagging subsets corresponding to each space aggregation code in the initial target point cloud code set into the corresponding target point cloud code sets to generate the target point cloud code sets of the corresponding sub-tags.
In this step, pass Q q f(q) -Q q 1 In the spatial aggregation subset screened under the condition of being less than or equal to R, besides the target point cloud data corresponding to the sub-targets, some noise point cloud data also exist. The noise point cloud is usually distributed at a position far away from the target point cloud, i.e. Q q f(q) -Q q 1 Point cloud of = R. Therefore, the spatial clustering code corresponding to the noise point cloud data is usually different from the spatial clustering code corresponding to the target point cloud data. Meanwhile, the distribution of the target point cloud has higher aggregation property relative to the noise point cloud. There will be multiple tagged codes for the target point cloud corresponding to the same spatial aggregation code. Because the distribution of the noise point clouds is relatively discrete, and the space distance between any two noise point clouds is relatively large, the corresponding space aggregation codes between different noise point clouds are basically different. Therefore, noise point cloud data in the spatial aggregation subset can be quickly removed through data cleaning processing. And further, the number of the target point cloud codes corresponding to each sub-target is more accurate, and the accuracy of the clustering result finally determined by the number of the target point cloud codes is further improved.
The method and the device use Hilbert codes to code the interested target, and obtain a clustering result by processing g-1 order and g-2 order Hilbert codes of target point cloud codes. Since one hilbert code can represent all point clouds in a corresponding cubic spatial region. Therefore, the data volume can be greatly reduced by encoding the point cloud by the method. Therefore, the calculation efficiency can be greatly improved, and the calculation result can be output more timely so as to improve the real-time performance.
Meanwhile, whether the point clouds corresponding to the target point cloud codes in the plurality of tagging subsets are close in space is determined by judging whether the g-1 order Hilbert codes corresponding to the tagging subsets are close or not by utilizing the numerical proximity of the Hilbert codes. I.e. will satisfy Q q f(q) -Q q 1 And placing a plurality of marking codes less than or equal to R into the same spatial aggregation subset. Therefore, the point clouds corresponding to different airplanes can be segmented by controlling the size of the R, and the point clouds corresponding to the same airplane can be prevented from being separated from the fracture positions and clustered into different targets respectively, so that the accuracy of the finally obtained clustering result can be improved.
As a possible embodiment of the present invention, step B300: generating a plurality of tagging subsets corresponding to the object of interest according to N and M, including:
step B301: the remaining tagged codes in N are subjected to a first matching process a plurality of times to generate a plurality of tagged subsets.
The first matching process includes:
step B302: and selecting the minimum tagging code in the current N as the target tagging code.
Step B303: and establishing a corresponding tagging subset, taking the target tagging codes as tagging codes of the tagging subset, and storing the target point cloud codes corresponding to the target tagging codes in the M into a point cloud code array of the tagging subset.
Step B304: and matching the target tagging code with each remaining tagging code in the N.
Step B305: and when any residual marking code is successfully matched with the target marking code, storing the target point cloud code corresponding to the marking code in the M into the point cloud code array of the marking subset.
Step B306: and deleting all the tagging codes which are the same as the target tagging codes in the current N so as to update the remaining tagging codes in the N.
Repeating steps B302-B306 multiple times can store the same tagging codes in N in the same tagging subset. Since the labeling code is a g-1 order Hilbert code of the corresponding target point cloud code. If a plurality of target point cloud codes correspond to the same g-1 order Hilbert code, it can be known that the plurality of target point cloud codes are gathered in the same spatial region, that is, the point clouds corresponding to the plurality of target point cloud codes belong to the same sub-target point cloud, so that the embodiment of the market can quickly complete the preliminary clustering of the point clouds of the plurality of sub-targets in the target region.
As a possible embodiment of the present invention, step B500: obtaining a spatial aggregation coding set corresponding to each spatial aggregation subset, including:
step B501: a third matching process is performed on each spatially aggregated subset to generate a corresponding spatially aggregated encoded set.
The third matching process includes:
step B502: and acquiring a binary code corresponding to each tagged code in the spatial aggregation subset.
Step B503: and taking the first w-2 bit code of the binary code corresponding to each tagging code as the parent code of each corresponding tagging code. w is the total number of bits per binary code.
The hilbert curve is a curve that can completely fill a square space. Specifically, the generation method of the hilbert curve of each step is the prior art, and is not described herein again. The generation of the multi-scale Hilbert curve is also recursive, for example, the construction process of the g-order Hilbert curve is to generate the g-1-order Hilbert curve first and then form the g-order Hilbert curve after connecting 4 g-1-order Hilbert curves end to end.
Corresponding multi-level hilbert codes are also formed recursively. If a g-2 order hilbert code is 000010, then its corresponding 4 g-1 order hilbert codes are 00001000, 00001001, 00001010, and 00001011. Similarly, the hubert codes of 4 g orders corresponding to the hubert code 00001000 of 1 g-1 order are 0000100000, 0000100001, 0000100010, and 0000100011. Wherein g-2 order hilbert code 000010 is the parent code of the hilbert codes of its corresponding 4 g-1 orders.
Step B504: the first parent code in the spatially aggregated subset is matched to each of the remaining parent codes.
Step B505: and when any one of the rest parent-level codes fails to be matched with the first parent-level code, deleting the tagging codes corresponding to the rest parent-level codes from the space aggregation subset.
Step B506: and after matching is finished, adding all the remaining marked codes in the space aggregation subset into the corresponding space aggregation code set.
Since the determining manner of the parent-level encoding in this embodiment is simpler and faster, the third matching processing may be performed on each spatial aggregation subset more quickly to generate a corresponding spatial aggregation encoding set.
As a possible embodiment of the present invention, step B700: determining the attributes of sub-targets corresponding to each target point cloud coding set according to the number of the target point cloud codes in each target point cloud coding set, wherein the attributes comprise:
step B701: and taking the target point cloud coding set with the target point cloud coding number larger than the second threshold value as a target coding set to be identified.
Step B702: and determining the attribute of the sub-target corresponding to each target coding set to be identified according to the mapping relation between the point cloud coding number and the sub-target attribute.
Because objects with different volume sizes correspond to different numbers of point clouds, namely, different numbers of target point cloud codes. Therefore, when the difference between the point cloud number in a fractal cloud cluster and the volume of the corresponding target object is large, it can be determined that the point cloud number is the noise to be removed, so that the noise point clouds in the fractal cloud clusters can be rapidly removed by setting a second threshold value, and the retained point clouds are the target fractal cloud clusters.
Taking an airplane as an example, since vehicles which do not belong to background point clouds, such as temporary rescue vehicles, may appear on the apron, fractal clouds corresponding to the rescue vehicles may exist in the determined fractal clouds. However, as the volume difference between the rescue vehicle and the airplane is large, the number of point clouds contained in the fractal cloud clusters respectively corresponding to the rescue vehicle also has a large difference, and therefore the fractal cloud clusters corresponding to the rescue vehicle can be removed by setting a second threshold. And then can more quick noise removal to more accurate target fractal cloud group that obtains.
Embodiments of the present invention also provide a non-transitory computer-readable storage medium, which may be disposed in an electronic device to store at least one instruction or at least one program for implementing a method of the method embodiments, where the at least one instruction or the at least one program is loaded into and executed by a processor to implement the method provided by the above embodiments.
Embodiments of the present invention also provide an electronic device comprising a processor and the aforementioned non-transitory computer-readable storage medium.
Embodiments of the present invention further provide a computer program product comprising program code means for causing an electronic device to carry out the steps of the method according to various exemplary embodiments of the invention described above when the program product is run on the electronic device.
Although some specific embodiments of the present invention have been described in detail by way of illustration, it should be understood by those skilled in the art that the above illustration is only for the purpose of illustration and is not intended to limit the scope of the invention. It will also be appreciated by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.
Claims (10)
1. A detection target acquisition method based on Hilbert curve coding is characterized by comprising the following steps of:
acquiring a background point cloud set of a target area, wherein the background point cloud set comprises a point cloud subset of a fixed target in a background and a position point cloud set of at least one movable target, and the position point cloud set comprises at least one position point cloud subset; the position point cloud subset is a point cloud set corresponding to a corresponding movable target at any position in a preset moving path;
performing Hilbert curve spatial coding processing on the background point cloud set to obtain a background point cloud coding set A = (A) 1 ,A 2 ,…,A i ,…,A n ) Wherein A is i Segmenting the target region for the ith-order Hilbert curve to obtain a background coding subset corresponding to the background point cloud; i =1,2, \8230, n, n is the total number of background coding subsets;
acquiring a point cloud set to be measured of a target area, wherein the point cloud set to be measured comprises the background point cloud set and a point cloud set to be identified, and the point cloud set to be identified comprises at least one target point cloud subset corresponding to a target to be identified;
performing Hilbert curve space coding processing on the cloud set of the point to be measured to obtain a point cloud coding set B = (B) 1 ,B 2 ,…,B i ,…,B n ) (ii) a Wherein, B i Partitioning the target area for the ith-order Hilbert curve to obtain a real-time coding subset corresponding to the point cloud to be detected;
according to a preset precision value c 1 Determining a target spatial error value E m (ii) a Wherein E is m <c 1 And c is and c 1 -E m ≤Y 1 ；Y 1 Is a first threshold value; e m After the target region is segmented for the mth-order Hilbert curve, the distance between construction points on any two adjacent Hilbert curves is calculated; m is an element of [1, n ]]；
Respectively determining a spatial error value E from A and B m First target coding subset a of m And a second target encoding subset B m ；
B is to be m Each code in (1) and (A) m The codes in (1) are matched one by one; when B is present m Any one of (1) and A m If all codes in the code list are different, the code matching fails;
b is to be m In which all matches failThe codes are put into the same set to obtain a target code set.
2. The method of claim 1, applied to a first system, wherein the first system includes a first buffer space and a second buffer space, and the first buffer space and the second buffer space are independent from each other;
the first cache space is used for conducting Hilbert curve space coding processing on the background point cloud set and storing the background point cloud coding set A;
and the second cache space is used for performing Hilbert curve space coding processing on the point cloud set to be measured obtained in real time and storing the point cloud coding set B to be measured.
3. Method according to claim 1, characterized in that it is based on a preset precision value c 1 Determining a target spatial error value E m The method comprises the following steps:
obtaining a spatial error value d corresponding to each-order Hilbert curve after the target area is divided 1 ,d 2 ,…,d i ,…,d n Wherein d is i A corresponding spatial error value after the target region is divided for the ith order Hilbert curve;
whenever d is i ≤c 1 When d is determined i Is an initial target spatial error value;
from all of said initial target spatial error values d c1 ,d c2 ,…,d cv ,…,d cu In (1), determining a target spatial error value E m ，E m The following conditions are satisfied:
E m ＝Max(d c1 ,d c2 ,…,d cv ,…,d cu )；
wherein Max () is a maximum function; d cv Is the v-th initial target spatial error value; u is the total number of initial target spatial error values; v =1,2, \8230;, u, u ≦ n.
4. The method of claim 3The method of (a) is characterized in that d i The following conditions are satisfied:
wherein, X i1 、Y i1 And Z i1 The coordinate value of the space coordinate of the corresponding target structure point after the target area is divided is the ith-order Hilbert curve; x i2 、Y i2 And Z i2 The coordinate value of the space coordinate of any point adjacent to the target construction point on the ith-order Hilbert curve; the target construction point is any construction point on the ith-order Hilbert curve.
5. The method of claim 3, wherein obtaining a spatial error value corresponding to each Hilbert curve after segmenting the target region comprises:
establishing a mapping relation table of the Hilbert segmentation order of the target area and the corresponding spatial error value;
and determining a spatial error value corresponding to each Hilbert curve after the target area is divided according to the mapping relation table and the Hilbert division order.
6. The method of claim 1, wherein after obtaining the target code set, the method further comprises:
obtaining a space coordinate e corresponding to each code in the target code set 1 ,e 2 ,…,e b ,…,e q Wherein e is b The spatial coordinates corresponding to the b-th code in the target code set; b =1,2, \ 8230;, q, q being the total number of codes in the target code set;
to e for 1 ,e 2 ,…,e b ,…,e q Performing point cloud segmentation to generate a plurality of fractal cloud clusters; each fractal cloud cluster consists of a plurality of points corresponding to the space coordinates;
determining a target fractal cloud cluster from a plurality of fractal cloud clusters; the number of corresponding spatial coordinates in the target fractal cloud cluster is greater than a second threshold.
7. The method of claim 1, wherein obtaining a background cloud of points for the target region comprises:
determining a plurality of target time points;
acquiring a background point cloud data frame of the target area at each target time point; the background point cloud data frame is a position point cloud set of a fixed target and a movable target in the background when the target time point is reached;
and generating a background point cloud set of the target area according to the plurality of background point cloud data frames.
8. The method of claim 7, wherein the frame of background point cloud data for the target area at each target time point is obtained by lidar.
9. A non-transitory computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the hilbert curve coding-based detection target obtaining method according to any one of claims 1 to 8.
10. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the hilbert curve coding-based detection target acquisition method according to any one of claims 1 to 8 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211394270.6A CN115690508A (en) | 2022-11-08 | 2022-11-08 | Detection target obtaining method, medium and equipment based on Hilbert curve coding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211394270.6A CN115690508A (en) | 2022-11-08 | 2022-11-08 | Detection target obtaining method, medium and equipment based on Hilbert curve coding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115690508A true CN115690508A (en) | 2023-02-03 |
Family
ID=85049758
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211394270.6A Pending CN115690508A (en) | 2022-11-08 | 2022-11-08 | Detection target obtaining method, medium and equipment based on Hilbert curve coding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115690508A (en) |
-
2022
- 2022-11-08 CN CN202211394270.6A patent/CN115690508A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yu et al. | A real-time detection approach for bridge cracks based on YOLOv4-FPM | |
| CN111091105B (en) | Remote sensing image target detection method based on new frame regression loss function | |
| CN108594816B (en) | Method and system for realizing positioning and composition by improving ORB-SLAM algorithm | |
| CN114120102A (en) | Boundary-optimized remote sensing image semantic segmentation method, device, equipment and medium | |
| CN113487730B (en) | Urban three-dimensional automatic modeling method based on laser radar point cloud data | |
| CN111462210B (en) | Monocular line feature map construction method based on epipolar constraint | |
| CN110659664B (en) | SSD-based high-precision small object identification method | |
| CN113126115B (en) | Semantic SLAM method and device based on point cloud, electronic equipment and storage medium | |
| CN110838112A (en) | Insulator defect detection method based on Hough transform and YOLOv3 network | |
| CN108428248B (en) | Vehicle window positioning method, system, equipment and storage medium | |
| CN114638954B (en) | Training method of point cloud segmentation model, point cloud data segmentation method and related device | |
| WO2021185121A1 (en) | Model generation method and apparatus, object detection method and apparatus, device, and storage medium | |
| CN113109816B (en) | Echo block tracking method, device and storage medium of radar echo image | |
| CN113724388B (en) | High-precision map generation method, device, equipment and storage medium | |
| CN112819873B (en) | High-generalization cross-domain road scene semantic segmentation method and system | |
| CN115861619A (en) | Airborne LiDAR (light detection and ranging) urban point cloud semantic segmentation method and system of recursive residual double-attention kernel point convolution network | |
| Ding et al. | Efficient vanishing point detection method in complex urban road environments | |
| CN113627440A (en) | Large-scale point cloud semantic segmentation method based on lightweight neural network | |
| CN115098717A (en) | Three-dimensional model retrieval method and device, electronic equipment and storage medium | |
| CN107529647B (en) | Cloud picture cloud amount calculation method based on multilayer unsupervised sparse learning network | |
| CN113704276A (en) | Map updating method and device, electronic equipment and computer readable storage medium | |
| CN115688004B (en) | Target attribute determining method, medium and device based on Hilbert coding | |
| Zhou | 3D urban modeling from city-scale aerial LiDAR data | |
| CN115687953B (en) | Target clustering method, medium and device based on Hilbert coding | |
| CN116630975A (en) | Semantic scene completion method based on feature representation decomposition and bird's eye view fusion |
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 |