CN108335325A - A kind of cube method for fast measuring based on depth camera data - Google Patents
A kind of cube method for fast measuring based on depth camera data Download PDFInfo
- Publication number
- CN108335325A CN108335325A CN201810089186.0A CN201810089186A CN108335325A CN 108335325 A CN108335325 A CN 108335325A CN 201810089186 A CN201810089186 A CN 201810089186A CN 108335325 A CN108335325 A CN 108335325A
- Authority
- CN
- China
- Prior art keywords
- cube
- coordinate
- cloud
- pixel
- top surface
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of cube method for fast measuring based on depth camera data, this method include：Depth map background removal, the identification of cube top surface and alignment, cuboid sides alignment, cube size parameter measurement.The algorithm of above-mentioned each step is optimized on operand, can be exported in 3D depth cameras and complete to measure in real time on video sequence.
Description
Technical field
The present invention relates to computer image processing technology fields, and in particular to the processing of the depth image data of depth camera and
Measuring technique.
Background technology
Solid identifies and measurement is the important foundation of industrial automation and robot control, several by Machine Vision Recognition
The geomery of what body provides important input data for Industry Control.Traditional algorithm is based on RGB camera data or laser
Distance-measuring equipment, wherein RGB camera since distance measure can not be directly obtained, need to calculate from binocular by largely post-processing or
Person obtains the Space geometric parameter of subject with structure light image, and operation is complicated, and efficiency is low.And the side based on laser measurement
Method precision is high but since laser measurement principle limits, and the point measured space object is sparse, and measurement accuracy suffers from one or two
The influence interfered in measurement point, it is difficult to improve precision.
The depth information that solid measuring technique based on 3D depth cameras is exported using depth camera number, can directly obtain
The three-dimensional coordinate information for obtaining space object, avoids the complicated calculations of traditional RGB camera, in addition depth camera is due to its resolution ratio
Range information that is high, can once obtaining each pixel within the scope of whole visual field, it is faster at high speed than laser optical method efficiency, therefore
Depth camera data is to improve the important hardware means of 3D solid measurement performances.
Although depth camera greatly reduces the computational complexity of geometry bulk measurement, due to the data of depth camera itself
Characteristic, in order to realize target measurement, it is necessary first to solve the solid identified in a large amount of background pixel points, then need to count
The angles and positions of testee are calculated, subsequent measurement can be implemented by only completing this two step, these steps are all linked with one another,
It is that the geometry bulk measurement based on depth map is essential.
Invention content
The present invention is directed to above-mentioned each step based on depth map geometry bulk measurement, the depth map based on depth camera output
Data quickly survey spatial cuboids to reach using the analysis calculating that carries out of corresponding cloud of each pixel in depth map
The method of amount, specifically includes following steps：
S1：The corresponding pixel of cube in depth map will be inputted according to existing background depth map to cut out；
S2：The corresponding pixel in cube top surface is identified from depth map and calculates its angle parameter；
S3：Calculate the rotation angle of 4 sides of cube and reference axis alignment needs；
S4：Cube size parameter measurement.
Preferably, the step S1 is specially：To the input depth map, calculate the corresponding depth of wherein each pixel and
The depth value difference of known background depth map corresponding position pixel, before the pixel that given thresholding is more than for difference is considered as
Scene element, i.e., cube respective pixel to be measured.Specific algorithm is：For the depth map pixel of location of pixels (u, v), if：|
DInput(u,v)-DBackground(u,v)|>θ then thinks that it is cubical pixel, otherwise wherein thinks that it is background pixel, is deleted.
Wherein u and v is pixel coordinate both horizontally and vertically of the pixel in depth map, D respectivelyInput(u, v) is that input depth map exists
The corresponding depth of pixel of (u, v) location of pixels, DBackground(u, v) is pixel of the known background depth map in (u, v) location of pixels
Corresponding depth, θ are thresholdings given by man, it can be provided specific range, such as 0.01 Μ.
Preferably, the step S2 is specially：Has background data to the basis cube in depth map is corresponding
In the cube depth map to be measured that the step of pixel is cut out exports, existing pixel include cube top surface pixel,
Cuboid sides pixel and noise pixel, cube top surface pixel refer to the cubical face of depth camera camera lens face.
In order to correctly identify the cube top surface, using with constrained plane fitting algorithm realize, algorithm flow chart by
Fig. 2 is provided.Steps are as follows：
S21：Calculate the coordinate (x that each of corresponding cloud of cube pixel is put in input depth mapn,yn), i.e.,：
xn=(un-cu)*DInput(un,vn)/f
yn=(vn-cv)*DInput(un,vn)/f
Wherein unAnd vnIt is that the corresponding nth pixel point of cube volumetric pixel is horizontal and vertical in inputting depth map respectively
The coordinate in direction, DInput(un,vn) it is to input depth map in (un,vn) location of pixels depth value, f be determined by depth camera hardware
Fixed internal reference.
S22：Partial dot is randomly selected from corresponding cloud of cube volumetric pixel by a certain percentage first, such as 1 institute of frame of Fig. 2
Show；
S23：3D plane fittings are carried out to the partial dot randomly selected, as shown in the frame 2 of Fig. 2, then meter is fitted
The included angle of the planar process vector sum Z axis arrived, if φ is more than threshold phithWhen think fitting failure and return to step S22, otherwise
Plane corresponding to fitting result further calculates its distance and is less than given thresholding dthCube volumetric pixel quantity K, such as scheme
Shown in 2 frame 3, when pixel quantity K is more than certain thresholding KthWhen, then enter step S24.Otherwise step S22 is returned to.
S24：Plane fitting is recalculated according to these pixels as a result, as shown in the frame 4 of Fig. 2, what above-mentioned algorithm used
The thought of RANSAC, but step S23 pixels update Calculation Plane fitting is only only performed once, when can reduce operation in this way
Between, RANSAC algorithms are in addition compared, additionally add the included angle and threshold phi of planar process vector sum Z axis in step S23 hereth
Comparison, quickly exclude the wrong fit Plane of non-cubic top surface.The distance of plane corresponding with above-mentioned fitting result is less than
Point in the point cloud of thresholding, the coordinate of the cubical top surface exactly identified, its corresponding cloud are denoted as { xn,yn,
zn}N=1,2 ..., K, wherein K is the quantity of cube top surface pixel.In order to further calculate cuboid sides angle, need pair
Cubical top surface carries out Space Rotating so that the corresponding plane in postrotational cube top surface and Z axis are vertical.Rotation angle is just
It is the normal vector of fit Plane and the angle of Z axis.
Preferably, the step S3 is specially：In order to quickly calculate the corresponding angle in cubical side, the algorithm is not straight
It connects using cubical side point cloud, but uses cubical top surface point cloud data, pass through step S2, revolving cube top surface
So that it is vertical with Z axis, the cube top surface having rotated at this time is also parallel with XOY plane, and top surface point cloud is projected to XOY plane
The point cloud chart of 2D is obtained, shown in the label 6 of Fig. 3, top surface point cloud is as shown in the label 5 of Fig. 3.The point cloud chart of the 2D shows
Rectangle, as shown in the label 7 of Fig. 4.The side of the rectangle do not have with the reference axis of XOY plane usually it is parallel or vertical, in order to count
Cube size is calculated, needs the angle on the side and X-axis or Y-axis of the rectangle measured first, this passes through the corresponding algorithms of Fig. 4
Schematic diagram carries out.The label 8 of Fig. 4 is the rotation results for rotating the point cloud chart of 2D around the Z axis of coordinate system, is rotated every time
Afterwards, for each position x in X-axis, calculate 2D point cloud chart rotate about the z axis after the obtained X-coordinate value of point cloud be less than the number of x
Amount：RX(x, α), as shown in the label 9 of Fig. 4.Wherein α is rotation angle.RX(x, α) is serpentine curve, and value range is incremented by from 0
To K, wherein RX(x, α) value range is denoted as interval width W in the range of [γ * K, (1- γ) * K] the corresponding X-axis in sectionX(α),
As shown in label 10 in Fig. 4, γ is artificially specified thresholding, value such as 0.01.Corresponding Y-axis equally executes above-mentioned algorithm, i.e.,
The Y-coordinate value for calculating the point cloud that the point cloud chart of 2D obtains after rotation alpha angle about the z axis is less than the quantity of y：RY(y, α), then calculates
RY(y, α) value range is in [γ * K, (1- γ) * K] the corresponding interval width W in sectionY(α).According to WX(α) and WY(α) calculates vertical
Projection and coordinate axis aligned angle [alpha] of the cube top surface point cloud on XOY plane*, i.e.,：α*=argminα{min[WX(α),WY
(α)]}。
Preferably, the step S4 is specially：The angle [alpha] that length and width is obtained by step S3*It is calculated, i.e.,：WX
(α*) and WY(α*) respectively as cubical length and width.Cubical height is the top surface plane corresponding points identified
The average value of the z coordinate of cloud, i.e.,：
Description of the drawings
Fig. 1 is algorithm overview flow chart
Fig. 2 is the identification of cube top surface and alignment algorithm flow chart
Fig. 3 is cube top projection schematic diagram
Fig. 4 is cuboid sides alignment algorithm schematic diagram
Claims (5)
1. a kind of depth map data based on depth camera output, utilizes the progress of corresponding cloud of each pixel in depth map
Analysis calculates to reach the method quickly measured spatial cuboids, which is characterized in that includes the following steps：
S1：The corresponding pixel of cube in depth map will be inputted according to existing background depth map to cut out；
S2：The corresponding pixel in cube top surface is identified from input depth map and calculates its angle parameter；
S3：Calculate the rotation angle of 4 sides of cube and reference axis alignment needs；
S4：Cube size parameter measurement.
2. the cube method for fast measuring according to claim 1 based on depth camera data, which is characterized in that described
Step S1 is specially：Compare the depth value difference of each corresponding position pixel in the input depth map and the background depth map
It is different, the pixel that difference is more than thresholding is filtered out, as the corresponding pixel of cube.
3. the cube method for fast measuring according to claim 1 based on depth camera data, which is characterized in that described
Step S2 is specially：Corresponding cloud of the corresponding pixel of the cube is calculated, is then randomly selected from described cloud repeatedly
Partial dot obtains fit Plane with their corresponding parametric equation of coordinate fitting geometrical plane, calculates the fit Plane
The angle of the Z axis of normal vector and coordinate system and the point quantity for counting the point cloud near plane, the angle for choosing the Z axis are less than door
Limit and described quantity are more than the fit Plane of thresholding as cubical top surface.Using the angle of the Z axis to described vertical
The coordinate of cube corresponding cloud of corresponding pixel carries out that postrotational coordinate is calculated, and the postrotational coordinate pair is made to answer
The corresponding cube top surface of point cloud it is parallel with the XOY plane of coordinate system.
4. the cube method for fast measuring according to claim 1 based on depth camera data, which is characterized in that described
Step S3 is specially：Based on described corresponding cloud of postrotational coordinate that step S2 is obtained, its Z around space coordinates is calculated
Postrotational cloud coordinate of axis, and calculate the x coordinate value and y-coordinate of the corresponding cube top surface point cloud coordinate of postrotational cloud
The value range of value, required for finding out so that the side of the corresponding rectangle of cube top surface point cloud is parallel with XY reference axis or vertical
The Z axis around coordinate system rotation angle.
5. the cube method for fast measuring according to claim 1 based on depth camera data, which is characterized in that described
Step S4 is specially：According to described so that the side of the corresponding rectangle of cube top surface point cloud is parallel with XY reference axis or vertical institute
The Z axis rotation angle around space coordinates needed calculates corresponding postrotational seat of cloud of cube top surface pixel
Mark, then by the value range of the value range and y-coordinate value of the x coordinate value of postrotational cloud coordinate, as a cube side
The length and width dimensions data in face export.The average value of the z coordinate value of the cube top surface point cloud, as cubical altitude information
Output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810089186.0A CN108335325A (en) | 2018-01-30 | 2018-01-30 | A kind of cube method for fast measuring based on depth camera data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810089186.0A CN108335325A (en) | 2018-01-30 | 2018-01-30 | A kind of cube method for fast measuring based on depth camera data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108335325A true CN108335325A (en) | 2018-07-27 |
Family
ID=62926263
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810089186.0A Pending CN108335325A (en) | 2018-01-30 | 2018-01-30 | A kind of cube method for fast measuring based on depth camera data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108335325A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110095062A (en) * | 2019-04-17 | 2019-08-06 | 北京华捷艾米科技有限公司 | A kind of object volume measurement method of parameters, device and equipment |
| CN110136193A (en) * | 2019-05-08 | 2019-08-16 | 广东嘉腾机器人自动化有限公司 | Cubold cabinet three-dimensional dimension measurement method and storage medium based on depth image |
| CN110223336A (en) * | 2019-05-27 | 2019-09-10 | 上海交通大学 | A kind of planar fit method based on TOF camera data |
| CN110335295A (en) * | 2019-06-06 | 2019-10-15 | 浙江大学 | A kind of plant point cloud acquisition registration and optimization method based on TOF camera |
| WO2021127947A1 (en) * | 2019-12-23 | 2021-07-01 | 华为技术有限公司 | Method and apparatus for measuring spatial dimension of object in image |
| CN113418467A (en) * | 2021-06-16 | 2021-09-21 | 厦门硅谷动能信息技术有限公司 | Method for detecting general and black luggage size based on ToF point cloud data |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150062301A1 (en) * | 2013-08-30 | 2015-03-05 | National Tsing Hua University | Non-contact 3d human feature data acquisition system and method |
| CN105389539A (en) * | 2015-10-15 | 2016-03-09 | 电子科技大学 | Three-dimensional gesture estimation method and three-dimensional gesture estimation system based on depth data |
| US20160102975A1 (en) * | 2014-10-10 | 2016-04-14 | Hand Held Products, Inc. | Methods for improving the accuracy of dimensioning-system measurements |
| CN106096512A (en) * | 2016-05-31 | 2016-11-09 | 上海美迪索科电子科技有限公司 | Utilize the detection device and method that vehicles or pedestrians are identified by depth camera |
| CN106651926A (en) * | 2016-12-28 | 2017-05-10 | 华东师范大学 | Regional registration-based depth point cloud three-dimensional reconstruction method |
| CN106780592A (en) * | 2016-06-30 | 2017-05-31 | 华南理工大学 | Kinect depth reconstruction algorithms based on camera motion and image light and shade |
| CN106803267A (en) * | 2017-01-10 | 2017-06-06 | 西安电子科技大学 | Indoor scene three-dimensional rebuilding method based on Kinect |
| CN106813568A (en) * | 2015-11-27 | 2017-06-09 | 阿里巴巴集团控股有限公司 | object measuring method and device |
| CN107292925A (en) * | 2017-06-06 | 2017-10-24 | 哈尔滨工业大学深圳研究生院 | Based on Kinect depth camera measuring methods |
| CN107292921A (en) * | 2017-06-19 | 2017-10-24 | 电子科技大学 | A kind of quick three-dimensional reconstructing method based on kinect cameras |
-
2018
- 2018-01-30 CN CN201810089186.0A patent/CN108335325A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150062301A1 (en) * | 2013-08-30 | 2015-03-05 | National Tsing Hua University | Non-contact 3d human feature data acquisition system and method |
| US20160102975A1 (en) * | 2014-10-10 | 2016-04-14 | Hand Held Products, Inc. | Methods for improving the accuracy of dimensioning-system measurements |
| CN105389539A (en) * | 2015-10-15 | 2016-03-09 | 电子科技大学 | Three-dimensional gesture estimation method and three-dimensional gesture estimation system based on depth data |
| CN106813568A (en) * | 2015-11-27 | 2017-06-09 | 阿里巴巴集团控股有限公司 | object measuring method and device |
| CN106096512A (en) * | 2016-05-31 | 2016-11-09 | 上海美迪索科电子科技有限公司 | Utilize the detection device and method that vehicles or pedestrians are identified by depth camera |
| CN106780592A (en) * | 2016-06-30 | 2017-05-31 | 华南理工大学 | Kinect depth reconstruction algorithms based on camera motion and image light and shade |
| CN106651926A (en) * | 2016-12-28 | 2017-05-10 | 华东师范大学 | Regional registration-based depth point cloud three-dimensional reconstruction method |
| CN106803267A (en) * | 2017-01-10 | 2017-06-06 | 西安电子科技大学 | Indoor scene three-dimensional rebuilding method based on Kinect |
| CN107292925A (en) * | 2017-06-06 | 2017-10-24 | 哈尔滨工业大学深圳研究生院 | Based on Kinect depth camera measuring methods |
| CN107292921A (en) * | 2017-06-19 | 2017-10-24 | 电子科技大学 | A kind of quick three-dimensional reconstructing method based on kinect cameras |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110095062A (en) * | 2019-04-17 | 2019-08-06 | 北京华捷艾米科技有限公司 | A kind of object volume measurement method of parameters, device and equipment |
| CN110136193A (en) * | 2019-05-08 | 2019-08-16 | 广东嘉腾机器人自动化有限公司 | Cubold cabinet three-dimensional dimension measurement method and storage medium based on depth image |
| CN110136193B (en) * | 2019-05-08 | 2021-06-11 | 广东嘉腾机器人自动化有限公司 | Rectangular box three-dimensional size measuring method based on depth image and storage medium |
| CN110223336A (en) * | 2019-05-27 | 2019-09-10 | 上海交通大学 | A kind of planar fit method based on TOF camera data |
| CN110223336B (en) * | 2019-05-27 | 2023-10-17 | 上海交通大学 | Plane fitting method based on TOF camera data |
| CN110335295A (en) * | 2019-06-06 | 2019-10-15 | 浙江大学 | A kind of plant point cloud acquisition registration and optimization method based on TOF camera |
| WO2021127947A1 (en) * | 2019-12-23 | 2021-07-01 | 华为技术有限公司 | Method and apparatus for measuring spatial dimension of object in image |
| CN113302654A (en) * | 2019-12-23 | 2021-08-24 | 华为技术有限公司 | Method and device for measuring spatial dimension of object in image |
| CN113418467A (en) * | 2021-06-16 | 2021-09-21 | 厦门硅谷动能信息技术有限公司 | Method for detecting general and black luggage size based on ToF point cloud data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108335325A (en) | A kind of cube method for fast measuring based on depth camera data | |
| CN103678754B (en) | Information processor and information processing method | |
| Orghidan et al. | Camera calibration using two or three vanishing points | |
| CN109544628B (en) | Accurate reading identification system and method for pointer instrument | |
| EP3273412B1 (en) | Three-dimensional modelling method and device | |
| CN111192235B (en) | Image measurement method based on monocular vision model and perspective transformation | |
| CN109523595A (en) | A kind of architectural engineering straight line corner angle spacing vision measuring method | |
| CN104976950B (en) | Object space information measuring device and method and image capturing path calculating method | |
| CN113393524A (en) | Target pose estimation method combining deep learning and contour point cloud reconstruction | |
| US12033406B2 (en) | Method and device for identifying presence of three-dimensional objects using images | |
| US20090080767A1 (en) | Method for determining a depth map from images, device for determining a depth map | |
| CN109493378B (en) | Verticality detection method based on combination of monocular vision and binocular vision | |
| Real-Moreno et al. | Camera calibration method through multivariate quadratic regression for depth estimation on a stereo vision system | |
| CN112991372B (en) | 2D-3D camera external parameter calibration method based on polygon matching | |
| CN115409693A (en) | Two-dimensional positioning method based on pipeline foreign matters in three-dimensional image | |
| CN108428250A (en) | A kind of X angular-point detection methods applied to vision positioning and calibration | |
| Kang et al. | 3D urban reconstruction from wide area aerial surveillance video | |
| Cai et al. | A target tracking and location robot system based on omnistereo vision | |
| Anwer et al. | Calculating real world object dimensions from Kinect RGB-D image using dynamic resolution | |
| Zhu et al. | Target Measurement Method Based on Sparse Disparity for Live Power Lines Maintaining Robot | |
| Kim et al. | Pose initialization method of mixed reality system for inspection using convolutional neural network | |
| Du et al. | Optimization of stereo vision depth estimation using edge-based disparity map | |
| Sheng et al. | Image segmentation and object measurement based on stereo vision | |
| Li et al. | Accurate calibration of stereo cameras for machine vision | |
| CN110706334B (en) | Three-dimensional reconstruction method for industrial part based on three-dimensional vision |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180727 |