CN114898218A - Monocular lookout tower monitoring video target space positioning method of high-dimensional overdetermined equation - Google Patents
Monocular lookout tower monitoring video target space positioning method of high-dimensional overdetermined equation Download PDFInfo
- Publication number
- CN114898218A CN114898218A CN202210711311.3A CN202210711311A CN114898218A CN 114898218 A CN114898218 A CN 114898218A CN 202210711311 A CN202210711311 A CN 202210711311A CN 114898218 A CN114898218 A CN 114898218A
- Authority
- CN
- China
- Prior art keywords
- camera
- view
- space
- ground
- rotation angle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000012544 monitoring process Methods 0.000 title claims abstract description 28
- 238000013507 mapping Methods 0.000 claims abstract description 45
- 230000002457 bidirectional effect Effects 0.000 claims abstract description 23
- 238000006243 chemical reaction Methods 0.000 claims description 37
- 230000000007 visual effect Effects 0.000 claims description 31
- 239000011159 matrix material Substances 0.000 claims description 14
- 230000009466 transformation Effects 0.000 claims description 13
- 238000005457 optimization Methods 0.000 claims description 9
- 238000012937 correction Methods 0.000 claims description 5
- 238000007476 Maximum Likelihood Methods 0.000 claims description 3
- 230000009897 systematic effect Effects 0.000 claims description 3
- 230000002265 prevention Effects 0.000 abstract description 4
- 238000004422 calculation algorithm Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000007637 random forest analysis Methods 0.000 description 5
- 238000009826 distribution Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/24—Aligning, centring, orientation detection or correction of the image
- G06V10/242—Aligning, centring, orientation detection or correction of the image by image rotation, e.g. by 90 degrees
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/24—Aligning, centring, orientation detection or correction of the image
- G06V10/245—Aligning, centring, orientation detection or correction of the image by locating a pattern; Special marks for positioning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a monocular lookout tower monitoring video target space positioning method based on a high-dimensional over-determined equation. The method is based on the real-time return angle data of the monocular lookout tower camera cloud deck and the geographic space coordinates of the video image ground objects, an over-determined equation set is constructed, the view space bidirectional mapping is realized, the positioning precision requirement is met, the problem of real-time bidirectional positioning of the video target monitored by the monocular lookout tower camera and the geographic space is solved, and the method has the advantages of strong applicability, wide positioning range, high positioning precision, real-time data return and the like, and has higher application value in scenes such as urban security, forest fire prevention and the like.
Description
Technical Field
The invention belongs to the technical field of monitoring video target space positioning, and particularly relates to a monocular watchtower monitoring video target space positioning method based on a high-dimensional overdetermined equation.
Background
Surveillance videos contain rich visual information, but do not have geospatial attributes. Therefore, the spatial expression of the target of the surveillance video is always the key point and the difficulty in the fusion direction of the surveillance video and the geographic information.
Monocular PTZ camera surveillance video is currently the most widespread one in the field of video surveillance. The traditional monocular PTZ camera monitoring video and the geographic scene are fused in two modes of visual perspective transformation and pixel characteristic matching. The visual perspective transformation specifically comprises a perspective transformation matrix and an inverse perspective transformation matrix so as to realize the projection and the inverse projection of the video to the geographic space, and the method has high requirements on the accuracy of the position coordinates and the actual posture parameters of the camera in the physical world. At present, a large number of monocular watchtower monitoring facilities have the problems of geographic space coordinate loss, relative parameter of attitude parameters and the like, are difficult to adapt to actual engineering production work, and have limited practical popularization. The pixel feature matching is realized by adopting a video and geographic scene picture pixel feature matching approach, such as a SIFT + RANSAC method, a three-dimensional geographic scene needs to be rendered on a computer in real time, a scene simulation camera is used for framing, and feature matching solving is carried out between the three-dimensional geographic scene and actually acquired video data. Therefore, the invention is a high-precision positioning method suitable for field remote distance and has important significance for popularization of engineering application.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to construct a high-dimensional overdetermined equation set based on real-time return angle data of a single-view lookout tower camera tripod head and the geographic space coordinates of video image ground objects, and realize the visual space bidirectional mapping, thereby solving the problem of real-time bidirectional positioning between a video target monitored by the single-view lookout tower camera and the geographic space, expanding the positioning range and improving the positioning precision.
The technical scheme is as follows: the invention discloses a monocular lookout tower monitoring video target space positioning method of a high-dimensional overdetermined equation, which comprises the following steps of:
step 1, constructing a visual reference point: and adjusting the posture of the camera, aligning the central point of the picture of the camera to the ground object with clear boundary, finding the ground object on the map, recording the posture of the camera and the geographic space coordinates of the ground object at the moment, and taking the coordinates as a reference point of the visual ground and recording the coordinates as P (x, y, z, P, t).
and 2.1, constructing a coordinate conversion relation from (x, y, z) to (p, t), establishing a corresponding high-dimensional over-determined equation, substituting the view ground reference points, solving unknown numbers of the high-dimensional over-determined equation, and obtaining a ground view mapping model from the geographic space coordinates to the camera rotation angle.
And 2.2, constructing a coordinate conversion relation from (p, t) to (x, y, z), establishing a corresponding high-dimensional over-determined equation, substituting the reference points into the view ground, solving unknown numbers of the high-dimensional over-determined equation, and obtaining a view ground mapping model from the rotation angle of the camera to the geographic space coordinate.
Step 3, error assessment and model optimization: and substituting the geographic space coordinates of the visual reference points into the model to obtain corresponding rotation angles, calculating a ground-view conversion error, and adjusting or deleting points with larger errors until the root-mean-square error meets the positioning requirement. And substituting the rotation angle of the visual reference point into the model to obtain a corresponding geographic space coordinate, calculating a visual conversion error, and adjusting or deleting the point with the larger error until the root mean square error meets the positioning requirement.
Step 5, converting according to the circumstances: randomly selecting a target point in a camera monitoring picture, acquiring corresponding camera external parameters (p, t) when the target point is calibrated by a simple pinhole model method, substituting the (p, t) into a monocular camera visual space bidirectional mapping model, solving corresponding geographic space coordinates (x, y, z), and identifying on a map to realize visual conversion.
Further, in the step 1, the number of the acquired visual reference points should be not less than 14 groups, so as to meet the resolving requirement of the unknown number in the high-dimensional over-determined equation. The reference points should be distributed as uniformly as possible in all directions of the camera to balance the positioning accuracy within the positioning range.
Further, in step 2, the substance of the monocular camera view-space two-way mapping model is a function relationship of one-to-one correspondence between the camera rotation angle (p, t) and the geospatial coordinates (x, y, z), and can be disassembled into a geospatial coordinate to camera rotation angle view mapping model and a camera rotation angle to geospatial coordinate view mapping model, and the specific steps are as follows:
step 2.1, constructing a coordinate conversion relation from (x, y, z) to (p, t), establishing a corresponding high-dimensional overdetermined equation shown in a formula (1),
converting the formula (1) into a matrix form, see formula (2):
and (3) substituting any 13 groups of ground-view reference points acquired in the step (1) into an equation (2), and solving a matrix by adopting a least square method to obtain a ground-view mapping model from the geospatial coordinates to the rotation angle of the camera.
Step 2.2, constructing a coordinate conversion relation from (x, y, z) to (p, t), establishing a corresponding high-dimensional overdetermined equation, see formula (3),
converting the above formula into matrix form, see formula (4)
And (3) substituting any 14 groups of ground reference points acquired in the step (1) into an equation (4), and solving a matrix by adopting a least square method to obtain a ground mapping model from the rotation angle of the camera to the geographic space coordinate.
Further, in step 3, the error is visually converted to reference point geospatial coordinates and a distance measure between predicted coordinates. The geodesic transformation error is measured in terms of the reference point rotation angle and the predicted rotation angle deviation.
Further, in the step 4, after the camera rotation angle (p', t) is calculated from the ground view model, the arbitrary posture (p) of the camera is determined 0 ,t 0 ) A geometric correction method is adopted to construct the three-dimensional space relation between the camera and the selected point and evaluate the coordinates (x, y, z) of the selected point and the coordinates (x) of the camera 0 ,y 0 ,z 0 ) To reduce the systematic error caused by the view conversion model, and to obtain the higher precision camera rotation angle (p, t) corresponding to the geographic space coordinate, the p value is calculated by formula (5),
wherein S represents the setpoint coordinates (x, y, z) and the camera coordinates (x) 0 ,y 0 ,z 0 ) WorkshopThe distance of (c).
Further, in the step 5, a perspective projection model method is adopted, that is, under the condition that the internal parameters of the camera are fixed, a plurality of calibration images are acquired through angle conversion, the relationship between the internal parameters of the camera and the external parameters of any calibration image is determined, through linear model analysis, the camera parameter optimization solution of any image can be calculated, and nonlinear refinement is performed by using a maximum likelihood method to obtain the required camera parameters (p, t).
Has the advantages that: compared with the prior art, the invention has the following advantages:
(1) according to the invention, the mapping relation is established by associating the rotation angle of the camera with the geographic coordinate for the first time through researching the data characteristics and the positioning conditions of the video monitoring of the monocular lookout tower. The camera is remotely controlled through linkage of the monitoring picture and map data to realize remote accurate monitoring of the target, so that the monitoring and supervision capacity of the monocular high-altitude observation tower video is greatly improved, and the video monitoring and supervision system is effectively used for businesses such as city security, forest fire prevention and the like.
(2) The invention constructs a monocular camera ground space bidirectional mapping model based on a high-dimensional overdetermined equation, and can realize high-precision ground space bidirectional positioning in a larger range through a limited number of ground reference point identifiers.
Drawings
FIG. 1 is a flow chart of the present invention
FIG. 2 is a view of a reference point distribution diagram
FIG. 3 is an error analysis chart
FIG. 4 is a schematic view of a camera in space of a rotation angle
FIG. 5 is a ground view of the conversion effect
FIG. 6 is a schematic view of a camera calibration image
FIG. 7 is a view of the effect of the visual conversion
Detailed Description
The technical solution of the present invention is described in detail below, but the scope of the present invention is not limited to the embodiments.
As shown in fig. 1, the method for positioning a monocular watchtower monitoring video target space by using a high-dimensional over-determined equation in this embodiment outputs a positioning parameter by constructing a bidirectional mapping relationship of a visual reference point, so as to implement visual bidirectional positioning, and specifically includes the following steps:
step 1, constructing a visual reference point: adjusting the posture of the camera, aligning the central point of the camera picture with the ground object with clear boundary, finding the ground object on the map, recording the posture of the camera and the geographic space coordinates of the ground object at the moment, taking the coordinates as a ground reference point and recording the coordinates as P 1 (x 1 ,y 1 ,z 1 ,p 1 ,t 1 )、P 2 (x 2 ,y 2 ,z 2 ,p 2 ,t 2 )、……、 P n (x n ,y n ,z n ,p n ,t n ) The number of the visual reference points is not less than 14 groups, and the visual reference points are uniformly distributed in all directions of the camera as much as possible so as to balance the positioning precision in the positioning range. Fig. 2 shows a view reference point distribution diagram, in which the shaded portion is the invisible area where the camera view is blocked, and therefore there is no optional reference point.
step 2.1, constructing a coordinate conversion relation from (x, y, z) to (p, t), establishing a corresponding high-dimensional overdetermined equation shown in a formula (1),
converting the formula (1) into a matrix form, see formula (2):
and (3) substituting any 13 groups of ground-view reference points acquired in the step (1) into an expression (2), and solving a matrix by adopting a least square method to obtain a ground-view mapping model from the geographic space coordinates to the rotation angle of the camera.
Step 2.2, constructing a coordinate conversion relation from (p, t) to (x, y, z), establishing a corresponding high-dimensional overdetermined equation, see formula (3),
converting the above formula into matrix form, see formula (4)
And (3) substituting any 14 groups of ground reference points acquired in the step (1) into an equation (4), and solving a matrix by adopting a least square method to obtain a ground mapping model from the rotation angle of the camera to the geographic space coordinate.
Step 3, error assessment and model optimization: and substituting the geographic space coordinates of the visual reference points into the model to obtain corresponding rotation angles, calculating a ground-view conversion error, and adjusting or deleting points with larger errors until the root-mean-square error meets the positioning requirement. And substituting the rotation angle of the visual reference point into the model to obtain a corresponding geographic space coordinate, calculating a visual conversion error, and adjusting or deleting the point with the larger error until the root mean square error meets the positioning requirement. The error is visually transformed to reference point geospatial coordinates and a distance measure between predicted coordinates. The geodesic transformation error is measured in terms of the reference point rotation angle and the predicted rotation angle deviation. As shown in fig. 3, the single-point positioning algorithm can exhaust the target position on the three-dimensional terrain profile to realize target object positioning, is a common watchtower monitoring video positioning method, and is widely applied to the field of forest fire prevention. The random forest is used as a machine learning algorithm with high accuracy, the core idea is integrated learning, unbiased estimation of reverse internal generation errors is obtained by integrating a plurality of decision trees, and the unbiased estimation can be used for solving the regression problem. Therefore, two algorithms of single-point positioning and random forest fitting are adopted as comparison experiments. Fig. 3(a) is a distribution diagram of predicted geographic coordinates, and it can be seen from the diagram that the error of the predicted result of the single-point positioning algorithm is large, mainly because it has a high requirement on the elevation accuracy, and is affected by the installation inclination of the pan-tilt, and the error increase areas are symmetrically distributed. In fig. 3(b), as the distance between the verification point and the tower increases, the random forest method error gradually increases, the single-point positioning method error distribution is unstable, and the algorithm error is stably distributed and has a downward trend. The phenomenon shows that the mapping model provided by the algorithm is not influenced by the prediction distance and has superiority in apparent fitting. In FIG. 3(c), the random forest prediction error is larger for the pan value due to the limited number of reference points, while the random forest fitting is more suitable for large data prediction scenarios. In the area where the reference points are sparsely distributed, the error of the algorithm tends to rise. Finally, the algorithm is obviously superior to other algorithms in the aspect of computational efficiency.
Table 1 is the error of the validation point prediction calculated by the visually transformed mapping model after error evaluation and model optimization, where the visually transformed error is measured as the distance d-error between the reference point geographic coordinates and the predicted coordinates. The conversion error is looked at to predict the rotation angle deviation p-error, t-error metric.
TABLE 1 results of the positioning experiment
Wherein S represents the setpoint coordinates (x, y, z) and the camera coordinates (x) 0 ,y 0 ,z 0 ) The distance between them.
And controlling the posture transformation of the camera according to the rotation angle to enable the target point to be positioned at the center of the monitoring picture, thereby realizing the ground vision conversion. As shown in fig. 5, the ground view conversion effect map is obtained by taking a camera as a center, uniformly selecting 16 points around the camera according to a rule, calculating the rotation angle of the camera and marking the feature corresponding to the map point in a video picture. In the figure, the prediction results of 16 positioning points are all positioned in the picture, more than 60 percent of the prediction results are close to the center of the picture, and the positioning effect is good.
And 5, visually converting: selecting a target point in a camera monitoring picture at will, collecting a plurality of calibration images through angle transformation under the condition that internal parameters of the camera are fixed by adopting a pinhole model method, determining the relation between the internal parameters of the camera and external parameters of any calibration image, calculating a camera parameter optimization solution of any image through linear model analysis, and performing nonlinear refinement by utilizing a maximum likelihood method to obtain required camera parameters (p, t), wherein the camera calibration image schematic diagram is shown in fig. 6. And (p, t) is substituted into the monocular camera view space bidirectional mapping model, so that the corresponding geographic space coordinates (x, y, z) can be solved, and the coordinates are marked on a map, thereby realizing view conversion. Fig. 7 is a view showing the effect of visual conversion. The visual ground conversion can be widely applied to target monitoring and positioning, three-dimensional information is added to a two-dimensional target detection result, a pond electronic fence is drawn in a video picture, a drawing point prediction result is displayed in a map, and the prediction result is basically matched with the contour of the pond.
In conclusion, the method and the device construct the over-determined equation set based on the real-time returned angle data of the tripod head of the monocular lookout tower camera and the geographic space coordinates of the ground objects of the video pictures, realize the bidirectional mapping of the view space, meet the positioning precision requirement, solve the problem of real-time bidirectional positioning of the video target monitored by the monocular lookout tower camera and the geographic space, and have the advantages of strong applicability, wide positioning range, high positioning precision, real-time returned data and the like. The invention does not depend on accurate camera external reference information, such as whether the camera is installed horizontally or not and whether the geographic coordinates are absolutely accurate or not, and has stronger adaptability to outdoor complex environment positioning. The positioning distance can reach more than 500 meters, the positioning precision is within 20 meters, and the method has high application value in scenes such as city security, forest fire prevention and the like.
Claims (6)
1. A monocular watchtower monitoring video target space positioning method based on a high-dimensional overdetermined equation is characterized by comprising the following steps: the method is based on a high-dimensional over-determined equation, a video camera view space mapping model is constructed through a small number of reference point calibration, and high-precision bidirectional positioning between a field large-scale video picture target point location and a geographic space coordinate is achieved. The method comprises the following steps:
step 1, constructing a visual reference point: and adjusting the posture of the camera, aligning the central point of the picture of the camera to a ground feature with clear boundary, finding the ground feature on the map, recording the geographic space coordinates of the ground feature and the rotation angles Pan and Tilt of the camera at the moment as a visual reference point, and marking as P (x, y, z, P and t).
Step 2, constructing a monocular camera view space bidirectional mapping model: the essence of the monocular camera view-space bidirectional mapping model is a corresponding functional relationship between a camera rotation angle (p, t) and a geographic space coordinate (x, y, z), and the monocular camera view-space bidirectional mapping model can be disassembled into a view-space mapping model from the geographic space coordinate to the camera rotation angle and a view-space mapping model from the camera rotation angle to the geographic space coordinate, and the specific steps are as follows:
and 2.1, constructing a coordinate conversion relation from (x, y, z) to (p, t), establishing a corresponding high-dimensional over-determined equation, substituting the view ground reference points, solving unknown numbers of the high-dimensional over-determined equation, and obtaining a ground view mapping model from the geographic space coordinates to the camera rotation angle.
And 2.2, constructing a coordinate conversion relation from (p, t) to (x, y, z), establishing a corresponding high-dimensional over-determined equation, substituting the reference points into the view ground, solving unknown numbers of the high-dimensional over-determined equation, and obtaining a view ground mapping model from the rotation angle of the camera to the geographic space coordinate.
Step 3, error assessment and model optimization: and substituting the geographic space coordinates of the visual reference points into the model to obtain corresponding rotation angles, calculating a ground-view conversion error, and adjusting or deleting points with larger errors until the root-mean-square error meets the positioning requirement. And substituting the rotation angle of the visual reference point into the model to obtain a corresponding geographic space coordinate, calculating a visual conversion error, and adjusting or deleting the point with the larger error until the root mean square error meets the positioning requirement.
Step 4, ground vision conversion: the method comprises the steps of selecting a point P at random within a map positioning range, obtaining geographic space coordinates (x, y, z) of the point, substituting the (x, y, z) into a selected ground-view positioning model, calculating corresponding camera rotation angles (P', t), adopting a geometric correction method for optimization due to the fact that a ground-view conversion model has certain system errors, calculating and obtaining the camera rotation angles (P, t) with higher precision, and then connecting a network to remotely control camera posture transformation to enable a target point position to be located in the center of a monitoring picture so as to achieve ground-view conversion.
And 5, visually converting: randomly selecting a target point in a camera monitoring picture, acquiring a corresponding camera rotation angle parameter (p, t) when the target is positioned at the center of the picture according to a pixel coordinate of the target point through a perspective projection model, substituting the (p, t) into a monocular camera view-space two-way mapping model, solving to obtain a corresponding geographic space coordinate (x, y, z), and identifying on a map to realize view conversion.
2. The monocular watchtower monitoring video target space positioning method based on the high-dimensional overdetermined equation as recited in claim 1, wherein: in the step 1, the number of the acquired visual reference points is not less than 14 groups, so that the resolving requirement of the unknown number in the high-dimensional over-determined equation is met. The reference points should be distributed as uniformly as possible in all directions of the camera to balance the positioning accuracy within the positioning range.
3. The monocular watchtower monitoring video target space positioning method based on the high-dimensional overdetermined equation as recited in claim 1, wherein: in the step 2, the monocular camera view space bidirectional mapping model is substantially a corresponding functional relationship between the camera rotation angle (p, t) and the geospatial coordinates (x, y, z), and can be disassembled into a view mapping model from the geospatial coordinates to the camera rotation angle and a view mapping model from the camera rotation angle to the geospatial coordinates, and the specific steps are as follows:
step 3.1, a geographical mapping model from the geographic space coordinates to the rotation angle of the camera, namely, a coordinate conversion relation from (x, y, z) to (p, t) is constructed, a corresponding high-dimensional overdetermined equation is established, see formula (1),
converting the formula (1) into a matrix form, see formula (2),
and (3) substituting any 13 groups of ground-view reference points acquired in the step (1) into an expression (2), and solving a matrix by adopting a least square method to obtain a ground-view mapping model from the geographic space coordinates to the rotation angle of the camera.
Step 3.2, a visual mapping model of the camera rotation angle to the geographic space coordinate is established, namely a coordinate conversion relation from (p, t) to (x, y, z) is established, a corresponding high-dimensional overdetermined equation is established, see formula (3),
the above formula is converted into a matrix form, see formula (4),
and (3) substituting any 14 groups of ground reference points acquired in the step (1) into an equation (4), and solving a matrix by adopting a least square method to obtain a ground mapping model from the rotation angle of the camera to the geographic space coordinate.
4. The monocular watchtower monitoring video target space positioning method based on the high-dimensional overdetermined equation as recited in claim 1, wherein: in said step 3, the error is visually converted to measure the distance between the reference point geospatial coordinates and the predicted coordinates. The geodesic transformation error is measured in terms of the reference point rotation angle and the predicted rotation angle deviation.
5. The monocular watchtower monitoring video target space positioning method based on the high-dimensional overdetermined equation as recited in claim 1, wherein: in the step 4, after the camera rotation angle (p', t) is calculated by the ground view model, an arbitrary posture (p) of the camera is determined 0 ,t 0 ) A geometric correction method is adopted to construct the three-dimensional space relation between the camera and the selected point and evaluate the coordinates (x, y, z) of the selected point and the coordinates (x) of the camera 0 ,y 0 ,z 0 ) So as to reduce the systematic error brought by the visual transformation model, the p value is calculated by formula (5),
wherein S represents the setpoint coordinates (x, y, z) and the camera coordinates (x) 0 ,y 0 ,z 0 ) The distance between them.
6. The monocular watchtower monitoring video target space positioning method based on the high-dimensional overdetermined equation as recited in claim 1, wherein: in the step 5, according to the perspective projection model, namely under the condition that the internal parameters of the camera are fixed, a plurality of calibration images are acquired through angle transformation, the relationship between the internal parameters of the camera and the external parameters of any calibration image is determined, through linear model analysis, the camera parameter optimization solution of any image can be calculated, and nonlinear refinement is performed by using a maximum likelihood method to obtain the required camera parameters (p, t).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210711311.3A CN114898218A (en) | 2022-06-22 | 2022-06-22 | Monocular lookout tower monitoring video target space positioning method of high-dimensional overdetermined equation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210711311.3A CN114898218A (en) | 2022-06-22 | 2022-06-22 | Monocular lookout tower monitoring video target space positioning method of high-dimensional overdetermined equation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114898218A true CN114898218A (en) | 2022-08-12 |
Family
ID=82728241
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210711311.3A Pending CN114898218A (en) | 2022-06-22 | 2022-06-22 | Monocular lookout tower monitoring video target space positioning method of high-dimensional overdetermined equation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114898218A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116433756A (en) * | 2023-06-15 | 2023-07-14 | 浪潮智慧科技有限公司 | Surface object space analysis method, device and medium of monocular camera |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226838A (en) * | 2013-04-10 | 2013-07-31 | 福州林景行信息技术有限公司 | Real-time spatial positioning method for mobile monitoring target in geographical scene |
| CN106803270A (en) * | 2017-01-13 | 2017-06-06 | 西北工业大学深圳研究院 | Unmanned aerial vehicle platform is based on many key frames collaboration ground target localization method of monocular SLAM |
| WO2020155616A1 (en) * | 2019-01-29 | 2020-08-06 | 浙江省北大信息技术高等研究院 | Digital retina-based photographing device positioning method |
| CN111915678A (en) * | 2020-07-17 | 2020-11-10 | 哈尔滨工程大学 | Underwater monocular vision target depth positioning fusion estimation method based on depth learning |
| CN114332385A (en) * | 2021-11-23 | 2022-04-12 | 南京国图信息产业有限公司 | Monocular camera target detection and spatial positioning method based on three-dimensional virtual geographic scene |
-
2022
- 2022-06-22 CN CN202210711311.3A patent/CN114898218A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226838A (en) * | 2013-04-10 | 2013-07-31 | 福州林景行信息技术有限公司 | Real-time spatial positioning method for mobile monitoring target in geographical scene |
| CN106803270A (en) * | 2017-01-13 | 2017-06-06 | 西北工业大学深圳研究院 | Unmanned aerial vehicle platform is based on many key frames collaboration ground target localization method of monocular SLAM |
| WO2020155616A1 (en) * | 2019-01-29 | 2020-08-06 | 浙江省北大信息技术高等研究院 | Digital retina-based photographing device positioning method |
| CN111915678A (en) * | 2020-07-17 | 2020-11-10 | 哈尔滨工程大学 | Underwater monocular vision target depth positioning fusion estimation method based on depth learning |
| CN114332385A (en) * | 2021-11-23 | 2022-04-12 | 南京国图信息产业有限公司 | Monocular camera target detection and spatial positioning method based on three-dimensional virtual geographic scene |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116433756A (en) * | 2023-06-15 | 2023-07-14 | 浪潮智慧科技有限公司 | Surface object space analysis method, device and medium of monocular camera |
| CN116433756B (en) * | 2023-06-15 | 2023-08-18 | 浪潮智慧科技有限公司 | Surface object space analysis method, device and medium of monocular camera |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105096382B (en) | A kind of method and device that real-world object information is associated in video monitoring image | |
| CN109523471B (en) | Method, system and device for converting ground coordinates and wide-angle camera picture coordinates | |
| CN109685855B (en) | Camera calibration optimization method under road cloud monitoring platform | |
| CN104463969B (en) | A kind of method for building up of the model of geographical photo to aviation tilt | |
| CN109520500A (en) | One kind is based on the matched accurate positioning of terminal shooting image and streetscape library acquisition method | |
| CN109191533B (en) | Tower crane high-altitude construction method based on fabricated building | |
| CN113390514B (en) | Three-dimensional infrared temperature measurement method based on multi-sensor array | |
| CN112419425B (en) | Anti-disturbance high-precision camera group measuring method for structural deformation measurement | |
| CN113902809A (en) | Method for jointly calibrating infrared camera and laser radar | |
| CN109297426B (en) | Large-scale precision industrial equipment deformation and servo angle detection method | |
| CN113077519A (en) | Multi-phase external parameter automatic calibration method based on human skeleton extraction | |
| CN114898218A (en) | Monocular lookout tower monitoring video target space positioning method of high-dimensional overdetermined equation | |
| CN108154535B (en) | Camera calibration method based on collimator | |
| US11703820B2 (en) | Monitoring management and control system based on panoramic big data | |
| CN106643735A (en) | Indoor positioning method and device and mobile terminal | |
| CN110991306A (en) | Adaptive wide-field high-resolution intelligent sensing method and system | |
| CN111240617B (en) | Video delivery method and system based on three-dimensional map and environment monitoring method and system | |
| CN116152471A (en) | Factory safety production supervision method and system based on video stream and electronic equipment | |
| CN112347974A (en) | Human head posture estimation algorithm and operator working state recognition system | |
| CN116755104A (en) | Method and equipment for positioning object based on three points and two lines | |
| CN116228888A (en) | Conversion method and system for geographic coordinates and PTZ camera coordinates | |
| CN108592789A (en) | A kind of steel construction factory pre-assembly method based on BIM and machine vision technique | |
| CN115326025A (en) | Binocular image measuring and predicting method for sea waves | |
| CN105635698B (en) | A method of Optimal Supervisory Control position is generated according to monitoring objective position | |
| CN113487677B (en) | Outdoor medium-long distance scene calibration method based on multi-PTZ camera with random distributed configuration |
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 |