CN111260720A - Target height measuring system based on deep learning method - Google Patents
Target height measuring system based on deep learning method Download PDFInfo
- Publication number
- CN111260720A CN111260720A CN202010029985.6A CN202010029985A CN111260720A CN 111260720 A CN111260720 A CN 111260720A CN 202010029985 A CN202010029985 A CN 202010029985A CN 111260720 A CN111260720 A CN 111260720A
- Authority
- CN
- China
- Prior art keywords
- target
- module
- height
- deep learning
- detection frame
- 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
Classifications
-
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
- G01B11/03—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness by measuring coordinates of points
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
- G01B11/06—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness for measuring thickness ; e.g. of sheet material
- G01B11/0608—Height gauges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06T2207/20081—Training; Learning
Abstract
The invention provides a target height measuring system based on a depth learning method, which comprises an image acquisition module, a target detection module, a depth calculation module, a detection frame correction module, a height calculation module and an output display module, wherein the image acquisition module is used for acquiring left and right views of a front scene by using a binocular camera; the target detection module is used for acquiring a target detection frame in the left view; the depth calculation module is used for acquiring depth information of a front target distance camera; the detection frame correction module is used for further correcting the target detection frame on the basis of acquiring the target detection frame; the height calculation module is used for calculating the height of the target; and the output display module is used for displaying the result of the target height calculation in real time. Compared with the traditional contact type measuring method and the non-contact type measuring method using infrared rays and the like, the invention can realize more accurate height measurement on a target with a larger scale.
Description
Technical Field
The invention belongs to the field of computer vision and deep learning, and particularly relates to a target height measuring system combining a target detection algorithm based on deep learning and a stereo matching algorithm.
Background
The height measuring instrument is mainly used for measuring the height of an object and can also measure the difference size of the shape and the position. Height measurement can be classified into a contact height measurement method and a non-contact height measurement method according to the measurement method. The contact type height measurement method requires that a target is erected on a bottom plate of a height measuring instrument, a horizontal plate is moved to the top of the target, and the height of the target is measured by reading the height value of the horizontal plate. The non-contact height measuring method comprises infrared measurement and ultrasonic measurement, the principle of the method is that the height value of a target is calculated by utilizing the time difference between light/waves emitted by an instrument and the top of the target, and the method can realize accurate measurement on a workpiece and is mainly applied to the production industries of machinery, molds, hardware and the like.
With the development of big data and the emergence of large-scale hardware acceleration equipment, the deep learning-based method is gradually applied to various fields. In 2012, a Hinton led team obtained champions on the ImageNet image classification competition using a multilayer convolutional neural network, and then deep learning developed explosively and had great success in tasks such as image classification, instance segmentation, stereo matching, object detection, etc., for example, the SENet model based on the deep learning method won the first one in the ImageNet classification task competition in 2017, and the M2S _ CSPN model based on deep learning occupies the first ranking in 2018 in the stereo matching algorithm of the KITTI website. Compared with the traditional image task computing method, the deep learning-based method can perform low-dimensional to high-dimensional information coding on image information through a large amount of convolution operation, so that the non-linear problem in the image field is better solved.
Binocular stereo vision is an important form of machine vision, and is a method for acquiring three-dimensional geometric information of an object by acquiring two images of the object to be measured from different positions by using imaging equipment based on a parallax principle and calculating position deviation between corresponding points of the images. The binocular stereo vision measuring method has the advantages of high efficiency, proper precision, simple system structure, low cost and the like, and is very suitable for online and non-contact product detection and quality control of a manufacturing site. Through the development of decades, the stereoscopic vision is more and more widely applied in the fields of robot vision, aerial surveying and mapping, reverse engineering, military application, medical imaging, industrial detection and the like.
Disclosure of Invention
The invention aims to provide a non-contact height measuring system based on a deep learning method for a large-scale target, which can stably and accurately measure the height of the target in the visual field of a binocular camera by using deep learning.
The main principle is that a target detection frame and depth information are obtained through a trained target detection algorithm and a trained stereo matching algorithm respectively, and then the height of a target in the real world is obtained by utilizing the conversion relation from an image coordinate system to a camera coordinate system. The system of the invention realizes the principle process through different functional modules.
The system of the invention mainly comprises six modules, including: an image acquisition module, a target detection module, a depth calculation module, a detection frame correction module, a height calculation module and an output display module, wherein
The image acquisition module is used for acquiring images in front by using a binocular camera, and the images comprise: each frame of image obtained by the left camera of the binocular camera and each frame of image obtained by the right camera of the binocular camera.
And the target detection module is used for identifying and framing a target in an image acquired by the left camera of the binocular camera by using a target detection algorithm based on deep learning under the condition that the binocular camera acquires a front image, so as to obtain a target detection frame.
The depth calculation module is used for performing parallax calculation on the left view by using a stereo matching method based on deep learning under the condition that the binocular camera acquires the left and right front views, and then calculating the depth information of the target by using the parallax according to a conversion formula from the parallax to the depth.
And the detection frame correction module is used for correcting the top edge and the bottom edge of the target detection frame by using the depth information of the target under the condition that the target detection frame and the depth information are obtained, so that the detection frame can accurately frame the target.
And the height calculating module is used for calculating the height of the target by utilizing the conversion relation from the image coordinate system to the camera coordinate system under the condition that the corrected target detection frame and the target depth information are obtained.
And the output display module is used for displaying the height information of the target in real time.
The invention is also characterized in that:
the target detection algorithm based on deep learning in the target detection module can better detect targets with different sizes by utilizing a hole-FPN structure, and simultaneously, the missing detection rate of the targets is effectively reduced by using Softer-NMS (NMS, namely non-maximum suppression).
And the stereo matching algorithm based on deep learning in the depth calculation module optimizes a network 3D convolution structure by utilizing an Encode-Decode network structure, so that the precision of the model is improved.
The discontinuity of the parallax in the left parallax map output by the depth calculation module among different target regions can be used for correcting the target detection frame output by the target detection module.
In the output display module, in order to improve the accuracy and stability of the height value measurement, the measurement results of the multiple frames of images are averaged and displayed.
The invention has the beneficial effects that: the method has the advantages that the influence of factors such as illumination and complex textures on the accuracy of the target detection frame can be avoided by utilizing a deep learning technology, the problem that multi-scale detection of the target is difficult can be effectively solved by utilizing a cavity-characteristic pyramid structure, the parallax calculation accuracy of a stereo matching network on the target can be effectively improved by utilizing an Encode-Decode structure, and the accuracy of the target frame can be effectively improved by utilizing target depth information to correct the detection frame.
Drawings
FIG. 1 is a schematic diagram of the system architecture of the present invention;
FIG. 2 is a schematic diagram of a deep learning model in the object detection module of the present invention;
FIG. 3 is a schematic diagram of a hole-FPN network structure model according to the present invention;
FIG. 4 is a schematic diagram of a deep learning model in the depth calculation module of the present invention;
FIG. 5 is a schematic diagram of an Encode-Decode network structure model according to the present invention;
FIG. 6 is a diagram illustrating a mapping relationship from an image coordinate system to a camera coordinate system according to the present invention.
Detailed Description
In order to explain technical contents, structural features, and objects and effects of the present invention in detail, the following detailed description is made with reference to the accompanying drawings in conjunction with the embodiments.
The invention provides a target height measuring system based on a deep learning method, and firstly, the deep learning system provided by the invention is introduced below.
Fig. 1 is a schematic structural diagram of a target height measuring system based on a deep learning method according to the present invention, as shown in fig. 1, the system includes: the device comprises an image acquisition module, a target detection module, a depth calculation module, a detection frame correction module, a height calculation module and an output display module. Wherein the content of the first and second substances,
the image acquisition module is used for acquiring images in front by using a binocular camera, and the images comprise: each frame of image obtained by the left camera of the binocular camera and each frame of image obtained by the right camera of the binocular camera.
And the target detection module is used for identifying and framing a target in an image acquired by the left camera of the binocular camera by using a target detection algorithm based on deep learning under the condition that the binocular camera acquires a front image, so as to obtain a target detection frame.
The depth calculation module is used for performing parallax calculation on the left view by using a stereo matching method based on deep learning under the condition that the binocular camera acquires the left and right front views, and then calculating the depth information of the target by using the parallax according to a conversion formula from the parallax to the depth.
And the detection frame correction module is used for correcting the top edge and the bottom edge of the target detection frame by using the depth information of the target under the condition that the target detection frame and the depth information are obtained, so that the detection frame can accurately frame the target.
And the height calculating module is used for calculating the height of the target by utilizing the conversion relation from the image coordinate system to the camera coordinate system under the condition that the corrected target detection frame and the target depth information are obtained.
And the output display module is used for displaying the height information of the target in real time.
Specifically, the image acquisition module mainly comprises a binocular camera, and the binocular camera can acquire left and right views of each frame of a front scene at the same time and transmit the images to the computer.
Specifically, the target detection module mainly identifies and frames a target in the left view by using a target detection algorithm based on deep learning, so as to obtain a target detection frame. The structure of the object detection network model is shown in fig. 2. The structure of the method is shown in FIG. 3, and Soft-NMS is used for reducing the missing rate of the target.
Specifically, the input of the depth calculation module is a left view and a right view, and parallax calculation is performed on information in the left view by using a stereo matching algorithm based on deep learning. The structure of the deep learning-based stereo matching network model is shown in fig. 3. The structure mainly comprises three parts: feature extraction block, matching cost volume and 3D volume block. An Encode-Decode structure is used in the 3D volume block, which can effectively improve the parallax precision of the output parallax map, and is shown in fig. 5.
The disparity map may be converted into a depth map using a conversion formula between disparity and depth. If the base line of the binocular camera is b and the focal length is f, the formula for converting the parallax value d into the depth value Z is as follows:
specifically, the detection frame correction module is used for further optimizing the target detection frame output by the target detection module, because the target detection frame output by the target detection module cannot perfectly frame the target itself. Under the condition that the output result of the depth calculation module is obtained, because the parallaxes in the same target area are continuous, the positions of the top and the bottom of the target can be obtained by utilizing the discontinuity of the target parallax value and the background parallax value at the edge of the target, and the boundary lines between the top and the bottom and the background are used as the top edge and the bottom edge of the target detection frame output by the target detection module, so that the corrected target detection frame is obtained.
The specific algorithm process for detecting the correction module is as follows:
1. and establishing a pixel coordinate system for the disparity map, wherein the lower left corner of the image is taken as an origin.
2. The vertex coordinates of the top left corner of the inspection box are (x1, y1) and the vertex coordinates of the bottom right corner are (x2, y 2). And calculating the coordinate of the central point of the detection frame, wherein the central coordinate is necessarily in the target area, and the central point is set as a base point O.
3. Starting from the point O, it is determined whether the absolute value of the parallax value of a certain point a in the neighborhood of the vertical coordinate of the point O and the parallax value of the point O is smaller than 2, and if so, the point a is set as a new base point O.
4. Repeating the step 3 until the ordinate value of the base point O is not changed.
5. By continuously updating the base point, the top pixel coordinate y _ top and the bottom pixel coordinate y _ bottom of the target area can be finally obtained.
6. And setting y _ top and y _ bottom as the vertical coordinate values of the upper side and the lower side of the detection frame, and obtaining the vertex coordinates of the upper left corner and the lower right corner of the corrected detection frame as (x1 and y _ top) and (x2 and y _ bottom).
It can be understood that, in the case that the corrected detection frame can perfectly frame the target, the actual height value of the target can be obtained by using the conversion relationship from the image coordinate system to the camera coordinate system. Fig. 4 is a schematic diagram showing a projection relationship between an object in a camera coordinate system and an image coordinate system, where P denotes a real object, P' denotes an object in a camera projection image, O denotes a camera optical center, f denotes a camera focal length, and Z denotes a distance from the object P to the camera optical center O. The display target scale in the camera coordinate system and the target scale in the image coordinate system can be found to be in a triangular similarity relation.
Specifically, the height calculation module may determine the actual height H of the target using the following formula. The height of a detection frame of a target in an image coordinate system is known to be h, the depth value from the target to a base line of a binocular camera is known to be Z, and the focal length of the binocular camera is known to be f.
Specifically, the output display module is configured to display the calculated height of the target. It can be understood that the target height calculated from each frame of image has errors and fluctuations. In order to improve the stability of the target height display, the output results of every ten frames of images in the height calculation module are averaged, and the average value is used as the display value in the output display module.
The invention relates to a method for measuring the height of a target based on a deep learning technology, which respectively carries out target detection and depth estimation on a scene in a binocular camera visual field range by introducing a target detection algorithm with a cavity-FPN structure and a stereo matching algorithm with an Encode-Decode structure, corrects a target detection frame by utilizing target depth information to improve the precision of the target detection frame, and finally obtains the real height value of the target by utilizing the conversion relation from an image coordinate system to a camera coordinate system, thereby realizing the non-contact measurement of the height of the target and requiring protection of the height measuring method. While the foregoing is directed to embodiments of the present invention, the true spirit and scope of the present invention is not limited thereto, and it will be apparent to those skilled in the art that various modifications, equivalent substitutions, improvements and the like can be made to achieve the desired height determination in various applications. The invention is defined by the claims and their equivalents.
Claims (10)
1. A target height determination system based on a deep learning method, the system comprising: an image acquisition module, a target detection module, a depth calculation module, a detection frame correction module, a height calculation module and an output display module, wherein,
the image acquisition module is used for acquiring images in front by using a binocular camera, and the images comprise: each frame of image acquired by the left camera of the binocular camera and each frame of image acquired by the right camera of the binocular camera;
the target detection module is used for identifying and framing a target in an image acquired by a left camera of the binocular camera by using a target detection algorithm based on deep learning under the condition that the binocular camera acquires a front image to obtain a target detection frame;
the depth calculation module is used for performing parallax calculation on a left view by using a stereo matching method based on deep learning under the condition that a binocular camera acquires a front left view and a front right view, and then calculating depth information of a target by using the parallax according to a conversion formula from the parallax to the depth;
the detection frame correction module is used for correcting the top edge and the bottom edge of the target detection frame by using the depth information of the target under the condition that the target detection frame and the depth information are obtained, so that the detection frame can accurately frame the target;
the height calculation module is used for calculating the height of the target by utilizing the conversion relation from the image coordinate system to the camera coordinate system under the condition that the corrected target detection frame and the target depth information are obtained;
and the output display module is used for displaying the height information of the target in real time.
2. The system for measuring the target height based on the deep learning method as claimed in claim 1, wherein the image acquisition module is used for acquiring left and right views of a current scene;
the left and right views are specifically for: the left view is used for the target detection module and used as input information of the module; both the left and right views are used in the depth calculation module as input information for the module.
3. The system for measuring the height of the target based on the deep learning method as claimed in claim 1, wherein the target detection module only identifies and frames the target in the left view by using a deep learning model, and only obtains a target detection frame in the left view;
the deep learning model in the target detection module is a target detection algorithm which is trained by samples in advance, the input of the algorithm is a left view, and the output of the algorithm is a target detection frame in the left view.
4. The system for measuring the height of the target based on the deep learning method as claimed in claim 1, wherein the depth calculation module takes left and right views as input, and utilizes a deep learning model to perform cost aggregation on the difference between the left and right views, so as to finally realize parallax calculation on the left view, and obtain the depth information of the target in the left view through a conversion formula of parallax and depth;
the deep learning model is a stereo matching algorithm which is trained by using a sample in advance, the input of the algorithm is a left view and a right view, and the output is parallax of the left view.
5. The system of claim 4, wherein the left-view disparity value outputted by the depth calculation module is continuous in the same target region and discontinuous in different target regions.
6. The system of claim 1, wherein the input of the detection and correction module is the disparity value of the target detection box and the left view.
7. The system of claim 5, wherein the detection and correction module corrects the top and bottom edges of the detection frame by using the inter-view discontinuity of different target areas.
8. The system for measuring the height of the target based on the deep learning method as claimed in claim 1, wherein the input information of the height calculation module is: the corrected target detection frame, the depth information of the target, the base line of the binocular camera and the focal length of the binocular camera.
9. The system of claim 8, wherein the height calculating module uses the input information and the transformation relationship from the image coordinate system to the camera coordinate system to obtain the actual height of the target.
10. The system of claim 1, wherein the output display module is configured to display the target detection frame and the target height information of the left view in real time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010029985.6A CN111260720A (en) | 2020-01-13 | 2020-01-13 | Target height measuring system based on deep learning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010029985.6A CN111260720A (en) | 2020-01-13 | 2020-01-13 | Target height measuring system based on deep learning method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111260720A true CN111260720A (en) | 2020-06-09 |
Family
ID=70953990
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010029985.6A Pending CN111260720A (en) | 2020-01-13 | 2020-01-13 | Target height measuring system based on deep learning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111260720A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111724434A (en) * | 2020-06-23 | 2020-09-29 | 江苏农牧科技职业学院 | Aquaculture body growth tracking method, device, system and medium |
| CN112733653A (en) * | 2020-12-30 | 2021-04-30 | 智车优行科技（北京）有限公司 | Target detection method and device, computer readable storage medium and electronic equipment |
| CN112800929A (en) * | 2021-01-25 | 2021-05-14 | 安徽农业大学 | On-line monitoring method for bamboo shoot quantity and high growth rate based on deep learning |
| CN113188465A (en) * | 2021-04-21 | 2021-07-30 | 中铁第四勘察设计院集团有限公司 | Drilling hole depth identification method and device based on video learning |
| CN113781543A (en) * | 2021-11-10 | 2021-12-10 | 北京中科慧眼科技有限公司 | Binocular camera-based height limiting device detection method and system and intelligent terminal |
| CN115917246A (en) * | 2020-08-07 | 2023-04-04 | 科磊股份有限公司 | 3D structure inspection or metrology using deep learning |
| CN116664657A (en) * | 2023-05-18 | 2023-08-29 | 宁波弗浪科技有限公司 | Expressway height limit measurement method and system based on binocular target detection |
| CN116704011A (en) * | 2023-08-04 | 2023-09-05 | 深圳市木兰轩科技有限公司 | LED display screen pixel correction analysis system based on data processing |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105956539A (en) * | 2016-04-27 | 2016-09-21 | 武汉大学 | Method for height measurement of human body based on background modeling and binocular vision |
| CN109086724A (en) * | 2018-08-09 | 2018-12-25 | 北京华捷艾米科技有限公司 | A kind of method for detecting human face and storage medium of acceleration |
| US20190164305A1 (en) * | 2016-06-13 | 2019-05-30 | Goertek Technology Co., Ltd. | Indoor distance measurement method |
| CN109977767A (en) * | 2019-02-18 | 2019-07-05 | 浙江大华技术股份有限公司 | Object detection method, device and storage device based on super-pixel segmentation algorithm |
| CN110288659A (en) * | 2019-05-27 | 2019-09-27 | 魏运 | A kind of Depth Imaging and information acquisition method based on binocular vision |
-
2020
- 2020-01-13 CN CN202010029985.6A patent/CN111260720A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105956539A (en) * | 2016-04-27 | 2016-09-21 | 武汉大学 | Method for height measurement of human body based on background modeling and binocular vision |
| US20190164305A1 (en) * | 2016-06-13 | 2019-05-30 | Goertek Technology Co., Ltd. | Indoor distance measurement method |
| CN109086724A (en) * | 2018-08-09 | 2018-12-25 | 北京华捷艾米科技有限公司 | A kind of method for detecting human face and storage medium of acceleration |
| CN109977767A (en) * | 2019-02-18 | 2019-07-05 | 浙江大华技术股份有限公司 | Object detection method, device and storage device based on super-pixel segmentation algorithm |
| CN110288659A (en) * | 2019-05-27 | 2019-09-27 | 魏运 | A kind of Depth Imaging and information acquisition method based on binocular vision |
Non-Patent Citations (1)
| Title |
|---|
| 杜若琪，谷宇章，张诚，王营冠: "基于双目立体视觉的行人身高测量方法", 《信息技术》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111724434A (en) * | 2020-06-23 | 2020-09-29 | 江苏农牧科技职业学院 | Aquaculture body growth tracking method, device, system and medium |
| CN115917246A (en) * | 2020-08-07 | 2023-04-04 | 科磊股份有限公司 | 3D structure inspection or metrology using deep learning |
| CN112733653A (en) * | 2020-12-30 | 2021-04-30 | 智车优行科技（北京）有限公司 | Target detection method and device, computer readable storage medium and electronic equipment |
| CN112800929A (en) * | 2021-01-25 | 2021-05-14 | 安徽农业大学 | On-line monitoring method for bamboo shoot quantity and high growth rate based on deep learning |
| CN113188465A (en) * | 2021-04-21 | 2021-07-30 | 中铁第四勘察设计院集团有限公司 | Drilling hole depth identification method and device based on video learning |
| CN113781543A (en) * | 2021-11-10 | 2021-12-10 | 北京中科慧眼科技有限公司 | Binocular camera-based height limiting device detection method and system and intelligent terminal |
| CN113781543B (en) * | 2021-11-10 | 2022-03-18 | 北京中科慧眼科技有限公司 | Binocular camera-based height limiting device detection method and system and intelligent terminal |
| CN116664657A (en) * | 2023-05-18 | 2023-08-29 | 宁波弗浪科技有限公司 | Expressway height limit measurement method and system based on binocular target detection |
| CN116704011A (en) * | 2023-08-04 | 2023-09-05 | 深圳市木兰轩科技有限公司 | LED display screen pixel correction analysis system based on data processing |
| CN116704011B (en) * | 2023-08-04 | 2024-03-05 | 深圳市凯立特光电科技有限公司 | LED display screen pixel correction analysis system based on data processing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111260720A (en) | Target height measuring system based on deep learning method | |
| CN108555908B (en) | Stacked workpiece posture recognition and pickup method based on RGBD camera | |
| CN109741405B (en) | Depth information acquisition system based on dual structured light RGB-D camera | |
| CN101697233B (en) | Structured light-based three-dimensional object surface reconstruction method | |
| CN109579695B (en) | Part measuring method based on heterogeneous stereoscopic vision | |
| CN103824318A (en) | Multi-camera-array depth perception method | |
| CN105046743A (en) | Super-high-resolution three dimensional reconstruction method based on global variation technology | |
| CN103337069B (en) | High-quality three-dimensional color image acquisition methods and device based on multiple camera | |
| CN102221331A (en) | Measuring method based on asymmetric binocular stereovision technology | |
| CN110966956A (en) | Binocular vision-based three-dimensional detection device and method | |
| CN107957246A (en) | Article geometrical size measuring method on conveyer belt based on binocular vision | |
| CN113888639B (en) | Visual odometer positioning method and system based on event camera and depth camera | |
| CN112017248B (en) | 2D laser radar camera multi-frame single-step calibration method based on dotted line characteristics | |
| CN112001973B (en) | Quick three-dimensional human head measuring method based on digital speckle correlation | |
| CN111105467B (en) | Image calibration method and device and electronic equipment | |
| CN113310433A (en) | Virtual binocular stereo vision measuring method based on line structured light | |
| CN102881040A (en) | Three-dimensional reconstruction method for mobile photographing of digital camera | |
| CN111738971B (en) | Circuit board stereoscopic scanning detection method based on line laser binocular stereoscopic vision | |
| CN114543787A (en) | Millimeter-scale indoor map positioning method based on fringe projection profilometry | |
| CN110728745A (en) | Underwater binocular stereoscopic vision three-dimensional reconstruction method based on multilayer refraction image model | |
| CN112712566B (en) | Binocular stereo vision sensor measuring method based on structure parameter online correction | |
| CN111429571B (en) | Rapid stereo matching method based on spatio-temporal image information joint correlation | |
| CN112525106A (en) | Three-phase machine cooperative laser-based 3D detection method and device | |
| CN104614372A (en) | Detection method of solar silicon wafer | |
| CN108180825A (en) | A kind of identification of cuboid object dimensional and localization method based on line-structured light |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200609 |
|
| RJ01 | Rejection of invention patent application after publication |