CN106331625A - Indoor single human body target PTZ tracking method - Google Patents
Indoor single human body target PTZ tracking method Download PDFInfo
- Publication number
- CN106331625A CN106331625A CN201610753304.4A CN201610753304A CN106331625A CN 106331625 A CN106331625 A CN 106331625A CN 201610753304 A CN201610753304 A CN 201610753304A CN 106331625 A CN106331625 A CN 106331625A
- Authority
- CN
- China
- Prior art keywords
- human body
- image
- target
- body target
- frame image
- 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 19
- 230000033001 locomotion Effects 0.000 claims abstract description 62
- 238000001514 detection method Methods 0.000 claims abstract description 21
- 239000013598 vector Substances 0.000 claims abstract description 21
- 239000013589 supplement Substances 0.000 claims description 4
- 230000008878 coupling Effects 0.000 claims description 3
- 238000010168 coupling process Methods 0.000 claims description 3
- 238000005859 coupling reaction Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 description 7
- 238000013459 approach Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000003703 image analysis method Methods 0.000 description 1
- 230000009017 pursuit movement Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000032258 transport Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
- G06T5/92—Dynamic range modification of images or parts thereof based on global image properties
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The invention discloses an indoor single human body target PTZ tracking method comprising the following steps: carrying out global motion estimation on a previous frame of image and a current frame of image; utilizing a global motion vector acquired by calculation by global motion estimation to carry out motion compensation on the previous image frame; acquiring an absolute difference of the image after the motion compensation and the current frame of image to acquire a residual image; confirming a moving target position area according to the energy distribution of the residual image; carrying out target detection on the position area to acquire an accurate position of a human body target and a picture-covering proportion; and then calculating a PTZ control instruction. According to the indoor single human body target PTZ tracking method disclosed by the invention, the global motion estimation is carried out on two continuous frames of video images, the motion compensation is carried out on the image according to the estimated global motion vector, the residual image of the two continuous frames is acquired by acquiring the absolute difference, then the position area of the target is confirmed, and fine positioning is performed on the target by a target detection method to acquire the accurate position and the size of the target.
Description
Technical field
The invention belongs to field of video monitoring, be specifically related to a kind of indoor single human body target PTZ tracking.
Background technology
Along with science and technology and the development of society, it is more and more extensive that video monitoring system is applied, and existing monitoring system is more come
More cannot meet the demand of actual application, be mainly reflected in, current demand needs not only to see the target of monitoring scene, also
Needing to see the action trail of the details of target and target, common camera only one scene of fixing monitoring, when target is left
During monitoring scene, just cannot monitor the behavior of target the most continuously, even and high-resolution video camera also cannot accurately see
The details of clear target farther out.
The Cloud Terrace or spherical camera can solve this problem to a certain extent, can pass through Artificial Control The Cloud Terrace or spherical
The ptz motion of video camera reaches follow the tracks of target and see the purpose of details clearly.In numerous monitoring application, by artificial side
Formula realizes the scheme that pursuit movement target clearly cannot be implemented for a long time, so having occurred as soon as the scheme from motion tracking, passes through
The method such as video image analysis method or infrared target detection realizes automatic motion target tracking.
The tracking of automatic target it is generally desirable to the size in picture of tracked target and can be maintained at and can see target clearly
Details, can take into account again the steady statue of certain background.This be accomplished by when target away from camera lens when, shooting function automatic
By zoom, pushing away nearly zoom, when target walks close to video camera when, zoom is zoomed out in camera lens energy autozoom.Common automatic tune
The algorithm of whole zoom is the zoom value that the T coordinate of the setting height(from bottom) according to ball machine and ball machine calculates needs, or by demarcating
Follow the tracks of the zoom value in solstics, calculated the zoom value of needs by the T coordinate of present image.But both approaches is required for
Video camera certain altitude (usually less than 4 meters) is installed, for the indoor target that setting height(from bottom) is limited, both approaches
Bigger error can be introduced owing to the height difference of target and Target Station are in the first-class problem of ladder, cause zoom excessive, it is impossible to take into account
Background even object boundary exceeds picture, or target is too small cannot see details clearly.
Summary of the invention
The present invention solves that problem that prior art exists proposes, its objective is to provide a kind of indoor single human body target
PTZ tracking.
The technical scheme is that a kind of indoor single human body target PTZ tracking, comprise the following steps:
() carries out overall motion estimation to previous frame image and current frame image;
() utilizes the calculated global motion vector of step () overall motion estimation, previous frame image carries out motion and mends
Repay;
() takes absolute difference to the image after step () motion compensation and current frame image, obtains residual image;
(), according to the Energy distribution of step () residual image, confirms moving target position region;
() carries out target detection to the step () band of position, obtains the exact position of human body target and shared aspect ratio;
() calculates PTZ control instruction according to aspect ratio shared by the band of position and human body target.
In described step (), the method for overall motion estimation is: previous frame image is divided into the uniform grid of 16 × 16,
Selected characteristic point in each grid, carries out the coupling search of these characteristic points in current frame image.
The coordinate of characteristic point selected in described each characteristic point corresponding point in current frame image and previous frame image
All motion vectors for the motion vector of current signature point, are iterated computing, obtain the unique of convergence by the vector formed
The global motion vector that motion vector is current frame image.
Determine that the method detecting the band of position is by residual image: previous frame image is carried out motion compensation, then supplement
Image after repaying and the absolute difference of present image, by the Energy distribution of absolute difference, calculate the detection region of human body target.
By residual image, step () determines that the method detecting the band of position is: previous frame image carries out motion and mends
Repay, then supplement repay after image and the absolute difference of present image, by the Energy distribution of absolute difference, calculate the detection of human body target
Region.
In region, carry out human detection, determine exact position and the ratio of shared picture of human body in present frame.
Account for the ratio of picture according to human body width, calculate desired zoom value Zaim, further according to the position of human body detected
Determine the control parameter to the motion of PT direction.
The invention has the beneficial effects as follows:
By front and back two frame video image is carried out overall motion estimation, according to the global motion vector estimated, image is entered
Row motion compensation, then by the way of seeking absolute difference, obtain the residual image of before and after two frame, and then confirm the band of position of target,
Again target is finely positioned by the method for target detection, obtain accurate location and the size of target.The present invention is predominantly
Solve indoor PTZ when following the tracks of, owing to the setting height(from bottom) of equipment receives the restriction of roof height, it is impossible to reach general PTZ and follow the tracks of
The setting height(from bottom) requirement of equipment, thus cause the zoom size determined according to general PTZ tracking, error is relatively big, causes
Zoom is excessive, it is impossible to take into account background even object boundary beyond picture, or target is too small cannot see details clearly, the most all can lead
Cause invalid tracking.
Accompanying drawing explanation
Fig. 1 is the method flow diagram of the present invention.
Detailed description of the invention
Hereinafter, referring to the drawings and embodiment the present invention is described in detail:
As it is shown in figure 1, a kind of indoor single human body target PTZ tracking, it is characterised in that: comprise the following steps:
() carries out overall motion estimation to previous frame image and current frame image.
Previous frame image and current frame image are carried out overall motion estimation, and the method for overall motion estimation is, by previous
Two field picture is divided evenly into the grid of 16 × 16, chooses 1-2 characteristic point in each grid, carries out this in current frame image
The coupling search of a little characteristic points.The purpose that one two field picture is divided into 16 × 16 grids equably is that the regional making image is all joined
With the calculating process to global motion vector, it is to avoid the global motion vector that the motion component of local moving objects causes the most greatly is repeatedly
Cannot restrain for process.
() utilizes the calculated global motion vector of step () overall motion estimation, transports previous frame image
Dynamic compensation.
The specific formula for calculation of each parameter of global motion vector is:
Wherein a1 is the zoom factor of horizontal direction, and a2 is the shift factor of horizontal direction, and a3 is the zoom factor of vertical direction,
A4 is the shift factor of vertical direction;K represents the line number of characteristic point place image block, and L represents the columns of characteristic point place block, v
Representing motion vector, vklx i.e. represents the motion vector in the x direction of the block that row k l arranges, and vkly is in like manner;S is the seat of characteristic point
Mark, sksl is horizontal coordinate and vertical coordinate.
() takes absolute difference to the image after step () motion compensation and current frame image, obtains residual image.
The formula of global motion compensation is:
Wherein u is coordinate Si, and Sj carries out the new coordinate position after motion compensation.
(), according to the Energy distribution of step () residual image, confirms moving target position region.Residual image is entered
The method that row is analyzed is: residual image is carried out binaryzation, white pixel carries out connected domain detection and obtains the extraneous square of agglomerate
Shape region, extends out 15% respectively by wide for this region height, and the region after extending out is as detection region.
() carries out target detection to the step () band of position, obtains exact position and the shared picture ratio of human body target
Example.
By image based on detection region, this image being carried out target detection, detection method can use common schema
The algorithm identified, the feature such as HOG, LBP, DPM coordinates common grader, it is possible to use convolutional neural networks even depth learns
Etc. method, concrete grammar does not do and specifies, the testing result of these methods can return target location and size accurately.
() calculates PTZ control instruction according to aspect ratio shared by the band of position and human body target.
The side-play amount at the positional distance picture center according to target, determines the movement velocity in P and T direction, or calculating is wanted
Say that target moves to the PT motion amplitude at picture center, if control the motion of PT according to motion amplitude, then need to send PT fortune
The instruction of dynamic amplitude is to interrupt at any time, to guarantee after new instruction arrives, it is not necessary to the instruction before waiting has performed
Can start to perform, so that the motion in PT direction is stable.The change of the ratio-dependent Z coordinate accounting for picture according to target diminishes greatly
And amplitude, finally making human body target be in picture central area, and keep suitable aspect ratio, this ratio can basis
Application demand determines, but needs to consider that overall motion estimation is affected greatly by target too conference, it is proposed that value is human body width
Account for about the 25% of picture width.
The present invention by front and back two frame video image is carried out overall motion estimation, according to the global motion estimated to
Amount, carries out motion compensation to image, then obtains the residual image of before and after two frame by the way of seeking absolute difference, and then confirms target
The band of position, then target is finely positioned by the method for target detection, obtains accurate location and the size of target.This
When inventing mainly for solving indoor PTZ tracking, owing to the setting height(from bottom) of equipment receives the restriction of roof height, it is impossible to reach one
As PTZ follow the tracks of the setting height(from bottom) requirement of equipment, thus cause the zoom size determined according to general PTZ tracking, error
Relatively big, cause zoom excessive, it is impossible to take into account background even object boundary beyond picture, or target is too small cannot see details clearly,
Finally all can cause invalid tracking.
Claims (7)
1. an indoor single human body target PTZ tracking, it is characterised in that: comprise the following steps:
() carries out overall motion estimation to previous frame image and current frame image;
() utilizes the calculated global motion vector of step () overall motion estimation, previous frame image carries out motion and mends
Repay;
() takes absolute difference to the image after step () motion compensation and current frame image, obtains residual image;
(), according to the Energy distribution of step () residual image, confirms moving target position region;
() carries out target detection to the step () band of position, obtains the exact position of human body target and shared aspect ratio;
() calculates PTZ control instruction according to aspect ratio shared by the band of position and human body target.
The indoor single human body target PTZ tracking of one the most according to claim 1, it is characterised in that: described step
In (), the method for overall motion estimation is: previous frame image is divided into the uniform grid of 16 × 16, chooses spy in each grid
Levy a little, in current frame image, carry out the coupling search of these characteristic points.
The indoor single human body target PTZ tracking of one the most according to claim 2, it is characterised in that: described each
The vector that the coordinate of characteristic point selected in characteristic point corresponding point in current frame image and previous frame image is formed, for working as
All motion vectors are iterated computing by the motion vector of front characteristic point, and the unique motion vector obtaining convergence is current
The global motion vector of two field picture.
The indoor single human body target PTZ tracking of one the most according to claim 1, it is characterised in that: pass through residual error
Image determines that the method for the detection band of position is: previous frame image carries out motion compensation, then supplement repay after image and current
The absolute difference of image, by the Energy distribution of absolute difference, calculates the detection region of human body target.
The indoor single human body target PTZ tracking of one the most according to claim 1, it is characterised in that: step ()
The method being determined the detection band of position by residual image is: previous frame image carries out motion compensation, then supplement repay after figure
Picture and the absolute difference of present image, by the Energy distribution of absolute difference, calculate the detection region of human body target.
The indoor single human body target PTZ tracking of one the most according to claim 5, it is characterised in that: in region
Carry out human detection, determine exact position and the ratio of shared picture of human body in present frame.
The indoor single human body target PTZ tracking of one the most according to claim 6, it is characterised in that: according to human body
Width accounts for the ratio of picture, calculates desired zoom value Zaim, determines further according to the position of human body detected and moves PT direction
Control parameter.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610753304.4A CN106331625A (en) | 2016-08-30 | 2016-08-30 | Indoor single human body target PTZ tracking method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610753304.4A CN106331625A (en) | 2016-08-30 | 2016-08-30 | Indoor single human body target PTZ tracking method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106331625A true CN106331625A (en) | 2017-01-11 |
Family
ID=57789059
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610753304.4A Pending CN106331625A (en) | 2016-08-30 | 2016-08-30 | Indoor single human body target PTZ tracking method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106331625A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108377364A (en) * | 2018-02-07 | 2018-08-07 | 深圳市亿联智能有限公司 | A kind of high efficiency video monitoring mobile object trace mode |
CN112330720A (en) * | 2020-11-12 | 2021-02-05 | 北京环境特性研究所 | Tracking method and device for moving weak and small target |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1808498A (en) * | 2006-02-23 | 2006-07-26 | 上海交通大学 | Real-time automatic moving portrait tracking method incorporating time domain differential and spatial domain diversity |
CN102074016A (en) * | 2009-11-24 | 2011-05-25 | 杭州海康威视软件有限公司 | Device and method for automatically tracking motion target |
CN103325112A (en) * | 2013-06-07 | 2013-09-25 | 中国民航大学 | Quick detecting method for moving objects in dynamic scene |
CN104778726A (en) * | 2015-04-29 | 2015-07-15 | 深圳市保千里电子有限公司 | Motion trail tracing method and system based on human body characteristics |
-
2016
- 2016-08-30 CN CN201610753304.4A patent/CN106331625A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1808498A (en) * | 2006-02-23 | 2006-07-26 | 上海交通大学 | Real-time automatic moving portrait tracking method incorporating time domain differential and spatial domain diversity |
CN102074016A (en) * | 2009-11-24 | 2011-05-25 | 杭州海康威视软件有限公司 | Device and method for automatically tracking motion target |
CN103325112A (en) * | 2013-06-07 | 2013-09-25 | 中国民航大学 | Quick detecting method for moving objects in dynamic scene |
CN104778726A (en) * | 2015-04-29 | 2015-07-15 | 深圳市保千里电子有限公司 | Motion trail tracing method and system based on human body characteristics |
Non-Patent Citations (1)
Title |
---|
朱娟娟,郭宝龙: "复杂场景中基于变块差分的运动目标检测", 《光学精密工程》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108377364A (en) * | 2018-02-07 | 2018-08-07 | 深圳市亿联智能有限公司 | A kind of high efficiency video monitoring mobile object trace mode |
CN108377364B (en) * | 2018-02-07 | 2020-09-01 | 深圳市亿联智能有限公司 | High-efficiency video monitoring moving object tracking mode |
CN112330720A (en) * | 2020-11-12 | 2021-02-05 | 北京环境特性研究所 | Tracking method and device for moving weak and small target |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113269098B (en) | Multi-target tracking positioning and motion state estimation method based on unmanned aerial vehicle | |
CN110222581B (en) | Binocular camera-based quad-rotor unmanned aerial vehicle visual target tracking method | |
US11461912B2 (en) | Gaussian mixture models for temporal depth fusion | |
CN112505065B (en) | Method for detecting surface defects of large part by indoor unmanned aerial vehicle | |
CN110796010B (en) | Video image stabilizing method combining optical flow method and Kalman filtering | |
CN104215239B (en) | Guidance method using vision-based autonomous unmanned plane landing guidance device | |
CN109191504A (en) | A kind of unmanned plane target tracking | |
US11906983B2 (en) | System and method for tracking targets | |
CN103778645B (en) | Circular target real-time tracking method based on images | |
CN102447835A (en) | Non-blind-area multi-target cooperative tracking method and system | |
CN106910206B (en) | Target tracking method and device | |
JP6858681B2 (en) | Distance estimation device and method | |
CN103149939A (en) | Dynamic target tracking and positioning method of unmanned plane based on vision | |
CN103106659A (en) | Open area target detection and tracking method based on binocular vision sparse point matching | |
CN104704384A (en) | Image processing method, particularly used in a vision-based localization of a device | |
CN101511022A (en) | Method for implementing machine-carried video compression and target tracking unitedly | |
CN109949364B (en) | Vehicle attitude detection precision optimization method based on road side monocular camera | |
CN106780560A (en) | A kind of feature based merges the bionic machine fish visual tracking method of particle filter | |
CN103391430A (en) | DSP (digital signal processor) based relevant tracking method and special device | |
JP2017524932A (en) | Video-assisted landing guidance system and method | |
CN111915651A (en) | Visual pose real-time estimation method based on digital image map and feature point tracking | |
CN106331625A (en) | Indoor single human body target PTZ tracking method | |
CN111028272A (en) | Object tracking method and device | |
Wang et al. | An attitude estimation method based on monocular vision and inertial sensor fusion for indoor navigation | |
CN113916221A (en) | Self-adaptive pedestrian track calculation method integrating visual odometer and BP network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170111 |