CN109255803B - Displacement calculation method of moving target based on displacement heuristic - Google Patents

Displacement calculation method of moving target based on displacement heuristic Download PDF

Info

Publication number
CN109255803B
CN109255803B CN201810972565.4A CN201810972565A CN109255803B CN 109255803 B CN109255803 B CN 109255803B CN 201810972565 A CN201810972565 A CN 201810972565A CN 109255803 B CN109255803 B CN 109255803B
Authority
CN
China
Prior art keywords
dimensional
displacement
points
image
new
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.)
Active
Application number
CN201810972565.4A
Other languages
Chinese (zh)
Other versions
CN109255803A (en
Inventor
张朝阳
郑宝峰
张文涛
杨露
严腾
张向清
潘强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Changan University
Original Assignee
Changan University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Changan University filed Critical Changan University
Priority to CN201810972565.4A priority Critical patent/CN109255803B/en
Publication of CN109255803A publication Critical patent/CN109255803A/en
Application granted granted Critical
Publication of CN109255803B publication Critical patent/CN109255803B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20092—Interactive image processing based on input by user
    • G06T2207/20104—Interactive definition of region of interest [ROI]

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a displacement calculation method of a moving target based on displacement heuristic, which comprises the steps of selecting effective points on the three-dimensional surface of the moving target, carrying out displacement heuristic in a three-dimensional space along the driving direction of the moving target, and converting new three-dimensional space points into a two-dimensional image through a mapping matrix to form new image points after the heuristic is finished; generating a new intermediate image through the new image point and the current frame image, and taking the gray difference between the intermediate image and the previous frame image as the matched evaluation standard; after a certain distance is explored, selecting the exploration distance with the optimal matching cost as the displacement of the moving target; the displacement calculation method provided by the invention has higher precision and stability, and is a feasible displacement calculation method.

Description

Displacement calculation method of moving target based on displacement heuristic
Technical Field
The invention relates to a displacement calculation method, in particular to a displacement calculation method of a moving target based on displacement exploration.
Background
The traditional moving target displacement calculation method is realized by target detection and tracking, and the moving target tracking is to find the interested moving target in a sequence of images in real time. The traditional moving target tracking method is mainly divided into two types: the method has the advantages that a moving target is directly detected from an image sequence without relying on prior knowledge, and then the moving target is tracked; depending on prior knowledge, firstly establishing a proper model for a moving target, and then finding a target with the best matching degree in a sequence image in real time to realize tracking.
The existing displacement calculation method of the moving target based on target tracking mainly realizes the tracking of the moving target through a prediction model, and then calculates to obtain the displacement distance of the moving target, firstly, the moving target needs to be positioned and tracked, and when the moving target is positioned and tracked, characteristic points need to be selected on the moving target in an image, but the characteristic points in the image cannot realize the selection of the same characteristic points for each frame of image due to the influence of factors such as illumination, so the method has the main defect of low precision, and cannot meet the requirement of many practical applications on precision in real life.
Disclosure of Invention
The invention aims to provide a displacement calculation method of a moving target based on displacement exploration, which is used for solving the problems of low precision and the like of the displacement calculation method of the moving target in the prior art.
In order to realize the task, the invention adopts the following technical scheme:
a method for displacement computation of a moving object based on displacement heuristics, said method comprising:
step 1, acquiring n two-dimensional images containing a moving target and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image, wherein n is more than or equal to 2, normalizing the abscissa of each two-dimensional image to be between 0 and X, and normalizing the ordinate of each two-dimensional image to be between 0 and Y;
step 2, obtaining RGB data of a two-dimensional point according to the ith two-dimensional image and the (i + 1) th two-dimensional image, where i is (1,2, …, n-1), where the (i + 1) th two-dimensional image is a two-dimensional image of the ith two-dimensional image at a next time, and the method includes:
step 21, randomly selecting m two-dimensional points from the ith two-dimensional image to obtain RGB data of the m two-dimensional points and m three-dimensional points corresponding to the m two-dimensional points, wherein m is more than or equal to 1;
step 22, moving the m three-dimensional points by a step length delta d along the advancing direction of the moving target to obtain m new three-dimensional points, wherein the unit of the step length delta d is cm;
step 23, mapping the m new three-dimensional points into k new two-dimensional points, wherein k is less than or equal to m;
the abscissa of any one of the k new two-dimensional points is an integer from 0 to X, and the ordinate of any one of the k new two-dimensional points is an integer from 0 to Y;
step 24, selecting RGB data of two-dimensional points with the same coordinates as the k new two-dimensional points from the (i + 1) th two-dimensional image as RGB data of the k new two-dimensional points;
step 3, obtaining the tentative matching cost of the ith two-dimensional image by adopting a formula I:
Figure BDA0001776580380000021
wherein (b)j′,gj′,rj') RGB data for the j-th new two-dimensional point obtained by step 24, j ═ 1,2, …, k, (b)j,gj,rj) Reversely pushing back the RGB data of the two-dimensional point in the ith two-dimensional image corresponding to the jth new two-dimensional point in the step 21 by the step 24;
step 4, if i is not more than n-1, i is i +1, and the step 2 is returned; otherwise, selecting the first two-dimensional image with the minimum tentative matching cost from n-1 two-dimensional images, where l is (1,2, …, n-1), and the displacement s of the moving object is: where s is l × Δ d, the unit of the displacement s is cm.
Further, the step 1 comprises: a three-dimensional interested area is set in a space, and n two-dimensional images containing a moving target and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image are acquired in the interested area by utilizing an RGB-D camera.
Furthermore, when the RGB-D camera is used for acquiring n two-dimensional images containing the moving target and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image, the n two-dimensional images containing the moving target and the three-dimensional points corresponding to each two-dimensional point in each two-dimensional image are acquired at a fixed time interval t.
Compared with the prior art, the invention has the following technical characteristics:
1. the displacement calculation method of the moving target not only considers the integrity information of the two-dimensional image, but also utilizes the three-dimensional space information, thereby greatly improving the image matching precision and further improving the displacement calculation precision;
2. the displacement calculation method of the moving target provided by the invention simplifies the calculation steps and improves the calculation efficiency.
Drawings
FIG. 1 is a flow chart of a displacement calculation method provided by the present invention;
FIG. 2 is a schematic illustration of a three-dimensional region of interest provided in an embodiment of the present invention;
FIG. 3 is an intermediate image provided in one embodiment of the present invention;
FIG. 4 is a statistical graph of the heuristic matching cost provided in one embodiment of the present invention;
FIG. 5 is a graph of displacement statistics provided in one embodiment of the present invention.
Detailed Description
The following are specific examples provided by the inventors to further explain the technical solutions of the present invention.
Example one
The invention discloses a displacement calculation method of a moving object based on displacement exploration, which comprises the following steps of:
step 1, acquiring n two-dimensional images containing a moving target and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image, wherein n is more than or equal to 2, normalizing the abscissa of the n two-dimensional images to be between 0 and X, and normalizing the ordinate of the n two-dimensional images to be between 0 and Y;
in order to eliminate the interference of noise points and reduce the calculation amount, as a preferred embodiment, a three-dimensional region of interest is set in space, and n consecutive two-dimensional images containing a moving object and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image are acquired in the region of interest.
In order to improve the accuracy of the algorithm, n two-dimensional images containing a moving target and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image are collected at a fixed time interval t,
Figure BDA0001776580380000041
the time interval t is given in units of s and v is the average velocity of the moving object in units of cm/s.
In this embodiment, a RGB-D camera is used to acquire 20 consecutive two-dimensional images containing a moving object and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image in a three-dimensional region of interest as shown in fig. 2. Wherein, 20 continuous two-dimensional images containing the moving target mean that the moving target passes through the three-dimensional region of interest, and the two-dimensional images containing the moving target are collected at fixed time intervals from the time when the moving target enters the three-dimensional region of interest until the moving target leaves the three-dimensional region of interest, so as to obtain a series of two-dimensional images containing the moving target.
And 2, obtaining RGB data of two-dimensional points according to the ith two-dimensional image and the (i + 1) th two-dimensional image, wherein i is (1,2, …, n-1), and the (i + 1) th two-dimensional image is a two-dimensional image of the ith two-dimensional image at the next moment.
And the difference between the (i + 1) th two-dimensional image and the ith two-dimensional image is a fixed time interval t.
The step 2 comprises the following steps:
step 21, randomly selecting m two-dimensional points from the ith two-dimensional image to obtain RGB data of the m two-dimensional points and m three-dimensional points corresponding to the m two-dimensional points, wherein m is more than or equal to 1;
preferably, m two-dimensional points with obvious characteristics are selected from the selected two-dimensional image to increase the precision of the later exploration process, two-dimensional image points of the moving object and corresponding points on the three-dimensional surface are selected, and 2D-3D point pairs can be used for selecting the points
Figure BDA0001776580380000051
Where m represents the number of pairs of 2D-3D points.
Step 22, moving the m three-dimensional points by a step length delta d along the advancing direction of the moving target to obtain m new three-dimensional points, wherein the unit of the step length delta d is cm;
the advancing direction of the moving object may be a positive or negative direction along the X, Y or Z axis, and in the present embodiment, m three-dimensional images to be obtained from the i-th two-dimensional imageAll the points move along the X-axis direction by a step length delta d to obtain m new three-dimensional points P'i(X'i,Y'i,Z'i) Wherein x isi′＝xi+Δd。
Step 23, mapping the m new three-dimensional points into k new two-dimensional points, wherein k is less than or equal to m;
the abscissa of any one of the k new two-dimensional points is an integer from 0 to X, and the ordinate of any one of the k new two-dimensional points is an integer from 0 to Y;
in this step, m new three-dimensional points are mapped into k new two-dimensional points by using a mapping matrix method, error points may exist when the mapping matrix maps the m new three-dimensional points to the two-dimensional points, and the horizontal and vertical coordinates of the error points exceed the size range of the (i + 1) th image, so that the points are not counted into the calculation range, the new two-dimensional points meeting the requirements that the horizontal coordinates belong to [0, X ] and the vertical coordinates belong to [0, Y ] are selected, so that k new two-dimensional points are selected, k is less than or equal to m, and the horizontal and vertical coordinates of the k new two-dimensional points are within the horizontal and vertical coordinates range of the i-th two-dimensional image.
Step 24, selecting RGB data on two-dimensional points with the same coordinates as the m new two-dimensional points from the (i + 1) th two-dimensional image as RGB data of the m new two-dimensional points;
in the step, an intermediate image is reconstructed and obtained according to the RGB data of the m new two-dimensional points; two adjacent two-dimensional images are respectively Ii,Ii+1The two-dimensional point selected in the ith two-dimensional image is pj(uj,vj) j 1, 2.. m, a new two-dimensional point p can be calculated after one trial of steps 21-23j′(uj′,vj') | j ═ 1,2,. m; then, in a two-dimensional image Ii+1Midpoint pj′(uj′,vjThe RGB data at 1, 2.. m yields an intermediate image Imiddle。
In the present embodiment, the generated intermediate image is as shown in fig. 3.
Step 3, obtaining the tentative matching cost of the ith two-dimensional image by adopting a formula I:
Figure BDA0001776580380000061
wherein (b)j′,gj′,rj') RGB data for the j-th new two-dimensional point obtained by step 24, j ═ 1,2, …, k, (b)j,gj,rj) Reversely pushing back the RGB data of the two-dimensional point in the ith two-dimensional image corresponding to the jth new two-dimensional point in the step 21 by the step 24;
since the k new two-dimensional points in step 24 are obtained from step 21 through steps 22 and 23, the jth new two-dimensional point obtained in step 24 inevitably corresponds to one two-dimensional point in the ith image in step 21, that is, in this step, the RGB data of the original two-dimensional point and the RGB data of the new two-dimensional point obtained through steps 22 to 24 are calculated to obtain a tentative matching cost.
Step 4, if i is not more than n-1, i is i +1, and the step 2 is returned; otherwise, selecting the first two-dimensional image with the minimum tentative matching cost from n-1 two-dimensional images, where l is (1,2, …, n-1), and the displacement s of the moving object is: where s is l × Δ d, the unit of the displacement s is cm.
When the tentative matching cost is the minimum, the similarity between the ith image and the (i + 1) th image is the maximum, because the method combines the image information and the three-dimensional space information, the image information comprises the whole information of the local feature information of the image, the local error can be eliminated as far as possible, the feature point matching only utilizes the local feature of the image to easily cause a plurality of wrong matching points, the method can avoid the error caused by the local feature as far as possible, and the error is eliminated by adopting a statistical method, so the method has higher precision.
The invention provides a displacement tentative high-precision displacement calculation method, which comprises the steps of firstly selecting effective points on a three-dimensional surface of a moving target; secondly, displacement probing is carried out in a three-dimensional space along the driving direction of the moving target, and after probing is finished, a new three-dimensional space point is converted to a two-dimensional image through a mapping matrix to form a new image point; then, generating a new intermediate image through the new image point and the current frame image, and taking the gray difference between the intermediate image and the previous frame image as the matched evaluation standard; finally, after a certain distance is explored, the exploration distance with the optimal matching cost is selected as the displacement of the moving target, the method not only considers the integrity information of the two-dimensional image, but also utilizes the three-dimensional space information, the image matching precision is greatly improved, and further the displacement calculation precision is improved.
Example two
In the present embodiment, the displacement of the moving object is calculated, taking Δ d as 1cm as an example.
Step 1, collecting 21 continuous two-dimensional images containing moving targets and three-dimensional point coordinates corresponding to each two-dimensional point in each two-dimensional image at fixed time intervals of 0.033s in a three-dimensional region of interest shown in figure 2 by using an RGB-D camera;
step 2, obtaining an intermediate image between the 1 st image and the 2 nd image, comprising:
step 21, randomly selecting 1000 two-dimensional points and three-dimensional points corresponding to the 1000 two-dimensional points in the 1 st two-dimensional image;
step 22, moving the 1000 three-dimensional points along the X-axis direction by a step length Δ d of 1cm to obtain 1000 new three-dimensional points;
step 23, mapping 1000 new three-dimensional points into 1000 new two-dimensional points by adopting a mapping matrix method;
step 24, selecting RGB data on two-dimensional points with the same coordinates as the 1000 new two-dimensional points from the 2 nd two-dimensional image as RGB data of the 1000 new two-dimensional points, and establishing an intermediate image according to the RGB data of the 1000 new two-dimensional points;
step 3, obtaining a tentative matching cost of the 1 st two-dimensional image by adopting a formula I1＝45；
Step 4, repeating the step 2 to the step 3 to obtain cost2＝43.8，cost3＝43.1，……，cost16＝19.8，cost17＝21.2，cost18＝25，cost19＝31，cost20As shown in fig. 4, of the trial matching costs obtained from the 20 two-dimensional images, the trial matching cost of the 16 th two-dimensional image is the smallest, so that s is 16 × 1 is 16cm, that is, the displacement of the moving object is 16 cm.
EXAMPLE III
In order to verify the accuracy of the displacement calculation method based on displacement heuristic, the present embodiment uses a test vehicle to perform an experiment on the displacement heuristic method in an experimental scenario as shown in fig. 2. In the experimental process, the vehicle runs at a constant speed. To better analyze the displacement heuristic computation results, the single frame displacement computation results and the total displacement statistics are counted, as shown in fig. 5. And drawing a comparison statistical graph of the single-frame displacement value and the total displacement statistical value by taking the serial number of the acquired data as an x axis and the acquired single-frame displacement value/total displacement statistical value as a y axis (unit/cm). The displacement value of a single frame corresponding to each frame of image is basically kept unchanged, and the total displacement statistical curve is a straight line, so that the vehicle is shown to be in constant speed running and consistent with the actual running state of the vehicle.
In the embodiment, the vehicle runs at a constant speed in the experimental process, and the motion displacement of the vehicle under a single frame image should be similar or identical. In order to effectively verify the calculation accuracy of the displacement heuristic method, the present embodiment takes the average value and the variance of all displacement values calculated by the displacement heuristic method as the judgment criteria, and the specific calculation formula is shown in formula II.
Figure BDA0001776580380000091
Wherein the content of the first and second substances,
Figure BDA0001776580380000092
is the average of all the displacement values of the statistic, σ2Is the variance of all displacement values of the statistic, σ is the standard deviation of all displacement values of the statistic, and k is the number of frames of the image sequence. The mean and standard deviation of all the displacement values that can be obtained from the displacement heuristic method are:
Figure BDA0001776580380000101
σ＝0.361813
the method for calculating the displacement of the moving object based on the displacement heuristic has higher precision and stability, and is a feasible displacement calculation method.

Claims (3)

1. A method for calculating the displacement of a moving object based on displacement heuristics, the method comprising:
step 1, acquiring n two-dimensional images containing a moving target and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image, wherein n is more than or equal to 2, normalizing the abscissa of each two-dimensional image to be between 0 and X, and normalizing the ordinate of each two-dimensional image to be between 0 and Y;
step 2, obtaining RGB data of a two-dimensional point according to the ith two-dimensional image and the (i + 1) th two-dimensional image, where i is (1,2, …, n-1), where the (i + 1) th two-dimensional image is a two-dimensional image of the ith two-dimensional image at a next time, and the method includes:
step 21, randomly selecting m two-dimensional points from the ith two-dimensional image to obtain RGB data of the m two-dimensional points and m three-dimensional points corresponding to the m two-dimensional points, wherein m is more than or equal to 1;
step 22, moving the m three-dimensional points by a step length delta d along the advancing direction of the moving target to obtain m new three-dimensional points, wherein the unit of the step length delta d is cm;
step 23, mapping the m new three-dimensional points into k new two-dimensional points, wherein k is less than or equal to m;
the abscissa of any one of the k new two-dimensional points is an arbitrary numerical value between 0 and X, and the ordinate of any one of the k new two-dimensional points is an arbitrary numerical value between 0 and Y;
step 24, selecting RGB data of two-dimensional points with the same coordinates as the k new two-dimensional points from the (i + 1) th two-dimensional image as RGB data of the k new two-dimensional points;
step 3, obtaining the tentative matching cost of the ith two-dimensional image by adopting a formula I:
Figure FDA0003308866550000021
wherein (b)j′,gj′,rj') RGB data for the j-th new two-dimensional point obtained by step 24, j ═ 1,2, …, k, (b)j,gj,rj) Reversely pushing back the RGB data of the two-dimensional point in the ith two-dimensional image corresponding to the jth new two-dimensional point in the step 21 by the step 24;
step 4, if i is not more than n-1, i is i +1, and the step 2 is returned; otherwise, selecting the first two-dimensional image with the minimum tentative matching cost from n-1 two-dimensional images, where l is (1,2, …, n-1), and the displacement s of the moving object is: where s is l × Δ d, the unit of the displacement s is cm.
2. The method for calculating the displacement of a moving object based on displacement heuristic, as claimed in claim 1, characterized in that said step 1 comprises: a three-dimensional interested area is set in a space, and n two-dimensional images containing a moving target and three-dimensional points corresponding to each two-dimensional point in each two-dimensional image are acquired in the interested area by utilizing an RGB-D camera.
3. The method of claim 2, wherein the n two-dimensional images containing the moving object and the three-dimensional point corresponding to each two-dimensional point in each two-dimensional image are acquired at a fixed time interval t while acquiring n two-dimensional images containing the moving object and the three-dimensional point corresponding to each two-dimensional point in each two-dimensional image with the RGB-D camera.
CN201810972565.4A 2018-08-24 2018-08-24 Displacement calculation method of moving target based on displacement heuristic Active CN109255803B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810972565.4A CN109255803B (en) 2018-08-24 2018-08-24 Displacement calculation method of moving target based on displacement heuristic

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810972565.4A CN109255803B (en) 2018-08-24 2018-08-24 Displacement calculation method of moving target based on displacement heuristic

Publications (2)

Publication Number Publication Date
CN109255803A CN109255803A (en) 2019-01-22
CN109255803B true CN109255803B (en) 2022-04-12

Family

ID=65049508

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810972565.4A Active CN109255803B (en) 2018-08-24 2018-08-24 Displacement calculation method of moving target based on displacement heuristic

Country Status (1)

Country Link
CN (1) CN109255803B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101923717A (en) * 2009-06-10 2010-12-22 新奥特（北京）视频技术有限公司 Method for accurately tracking characteristic points of quick movement target
CN103425764A (en) * 2013-07-30 2013-12-04 广东工业大学 Vehicle matching method based on videos
CN103971383A (en) * 2014-05-21 2014-08-06 国家电网公司 Method for calculating velocity of movement of targets in video images
CN105526872A (en) * 2016-01-11 2016-04-27 浣石 Three-dimensional motion measurement system based on digital image measurement
CN106033614A (en) * 2015-03-20 2016-10-19 南京理工大学 Moving object detection method of mobile camera under high parallax
CN106846362A (en) * 2016-12-26 2017-06-13 歌尔科技有限公司 A kind of target detection tracking method and device
CN108550143A (en) * 2018-04-03 2018-09-18 长安大学 A kind of measurement method of the vehicle length, width and height size based on RGB-D cameras

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5463584B2 (en) * 2009-09-08 2014-04-09 株式会社パスコ Displacement measuring method, displacement measuring apparatus, and displacement measuring program

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101923717A (en) * 2009-06-10 2010-12-22 新奥特（北京）视频技术有限公司 Method for accurately tracking characteristic points of quick movement target
CN103425764A (en) * 2013-07-30 2013-12-04 广东工业大学 Vehicle matching method based on videos
CN103971383A (en) * 2014-05-21 2014-08-06 国家电网公司 Method for calculating velocity of movement of targets in video images
CN106033614A (en) * 2015-03-20 2016-10-19 南京理工大学 Moving object detection method of mobile camera under high parallax
CN105526872A (en) * 2016-01-11 2016-04-27 浣石 Three-dimensional motion measurement system based on digital image measurement
CN106846362A (en) * 2016-12-26 2017-06-13 歌尔科技有限公司 A kind of target detection tracking method and device
CN108550143A (en) * 2018-04-03 2018-09-18 长安大学 A kind of measurement method of the vehicle length, width and height size based on RGB-D cameras

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
视频流中的单向快速运动多目标可靠跟踪技术;王荣 等;《电视技术》;20130930;第447-450页 *

Also Published As

Publication number Publication date
CN109255803A (en) 2019-01-22

Similar Documents

Publication Publication Date Title
CN107038717A (en) A kind of method that 3D point cloud registration error is automatically analyzed based on three-dimensional grid
CN109903331B (en) Convolutional neural network target detection method based on RGB-D camera
CN104715251B (en) A kind of well-marked target detection method based on histogram linear fit
CN108627798B (en) WLAN indoor positioning algorithm based on linear discriminant analysis and gradient lifting tree
CN105809672A (en) Super pixels and structure constraint based image's multiple targets synchronous segmentation method
CN110443849B (en) Target positioning method for double-current convolution neural network regression learning based on depth image
CN104517289A (en) Indoor scene positioning method based on hybrid camera
CN110349186B (en) Large-displacement motion optical flow calculation method based on depth matching
CN110260795B (en) Absolute displacement detection method based on incremental absolute grating ruler
CN108960190A (en) SAR video object detection method based on FCN Image Sequence Model
CN111768447A (en) Monocular camera object pose estimation method and system based on template matching
CN106295532A (en) A kind of human motion recognition method in video image
CN105022101B (en) Severe Convective Cloud Cluster method for tracing
CN111210458B (en) Moving target tracking-before-detection method based on pre-detection confidence
CN113313047A (en) Lane line detection method and system based on lane structure prior
CN110598711B (en) Target segmentation method combined with classification task
CN109255803B (en) Displacement calculation method of moving target based on displacement heuristic
CN111598158B (en) Warehouse inventory state sensing method based on AI analysis technology
CN113327269A (en) Unmarked cervical vertebra movement detection method
CN111709998B (en) ELM space registration model method for TOF camera depth data measurement error correction
CN115457780B (en) Vehicle flow and velocity automatic measuring and calculating method and system based on priori knowledge set
CN104392469B (en) A kind of method for tracking target based on soft characteristic theory
Zhang et al. Depth Monocular Estimation with Attention-based Encoder-Decoder Network from Single Image
CN115035164A (en) Moving target identification method and device
Wu et al. Ground-distance segmentation of 3D LiDAR point cloud toward autonomous driving

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
GR01 Patent grant
GR01 Patent grant