CN107909637A - A kind of magnanimity monitor video uses and presentation mode - Google Patents

A kind of magnanimity monitor video uses and presentation mode Download PDF

Info

Publication number
CN107909637A
CN107909637A CN201711043476.3A CN201711043476A CN107909637A CN 107909637 A CN107909637 A CN 107909637A CN 201711043476 A CN201711043476 A CN 201711043476A CN 107909637 A CN107909637 A CN 107909637A
Authority
CN
China
Prior art keywords
image
video
registration
presentation mode
feature
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
Application number
CN201711043476.3A
Other languages
Chinese (zh)
Inventor
宋昌江
杨东亮
***
李昕迪
费磊
周钰琢
胡玥明
吴冈
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Institute of Automation of Heilongjiang Academy of Sciences
Original Assignee
Institute of Automation of Heilongjiang Academy of Sciences
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Institute of Automation of Heilongjiang Academy of Sciences filed Critical Institute of Automation of Heilongjiang Academy of Sciences
Priority to CN201711043476.3A priority Critical patent/CN107909637A/en
Publication of CN107909637A publication Critical patent/CN107909637A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T15/003D [Three Dimensional] image rendering
    • G06T15/005General purpose rendering architectures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/30Determination of transform parameters for the alignment of images, i.e. image registration
    • G06T7/33Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
    • G06T7/337Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving reference images or patches
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00Television systems
    • H04N7/18Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10016Video; Image sequence
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20016Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20212Image combination
    • G06T2207/20221Image fusion; Image merging

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Graphics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Processing Or Creating Images (AREA)

Abstract

The invention discloses a kind of use of magnanimity monitor video and presentation mode, comprise the steps of:A, video image intercepts.B, image registration:Piece image possesses different extractible features, what selection was wherein easily extracted, and can characterize the feature of image similarity subject to registration to a certain extent as registering foundation.Based on the present invention proposes a kind of threedimensional model by reality scene, merged with reference to multi-channel video, real-time video monitoring new method, and development system model machine, development system modular software.Its feature is shown:1st, with real three-dimensional model foundation virtual environment, and in this, as video fusion carrier and final presentation main body;2nd, using the RANSAC algorithms of belt restraining realize more video matchings with it is registering;3rd, the real-time processing and display of more video fusions are realized using gaussian pyramid algorithm.

Description

A kind of magnanimity monitor video uses and presentation mode
Technical field
The present invention relates to a kind of video image processing technology, is specifically a kind of magnanimity monitor video use and presentation mode.
Background technology
For at present, the method for more commonly used full-view video image collection has two kinds, first, being imaged using fish eye lens Machine obtains large-scale video image, but its obtained anamorphose can be very serious, it is difficult to combined with threedimensional model, it is necessary to The image of generation is corrected with algorithm, so as to obtain panoramic plane video image.Second, the method for multiple-camera is used, so Obtained image is spliced afterwards and fusion treatment.Such as the Command View series of products of IPIX companies of the U.S., fish is utilized The wide viewing angle of glasses head directly acquires 360 ° of scenes, recycles algorithm to carry out image flame detection and obtains the plane of an annular extent Full-view video image;And the PARASCAN panoramic cameras that Honeywell companies release are then by multiple fixed shootings Machine shoots different spaces scene, then obtained image is spliced, so as to fulfill panoramic video.
At present, the external theory and practice research in terms of more camera lens overall view monitorings is the field of an opposite blank, Also rarely have the fairly perfect and convenient practical technique and solution realized at present.Similar is some panoramic videos or empty The solution of plan display is eventful to be applied to some large-scale displayings or video conference, such as the online generation of 2010 Shanghai World's Fairs It is rich to visit platform, it is exactly a very typical application of panorama virtual reality technology.In addition, in the National University of Defense technology, Zhejiang University and software study Suo Deng colleges and universities of the Chinese Academy of Sciences also carried out the research in similar direction with scientific research institutions, and achieve some into Fruit, but embodiment feasible also without complete set is proposed so far.
In terms of technical research, the panoramic mosaic technology of video can be divided into two major classes:Mode based on image registration;Base In the connecting method of manifold thought.Method for registering images can be summarized as the select permeability to three elements:Feature space, similitude Criterion and search strategy.Feature space defines the basic feature information for registration, and similarity criterion is tagsort Basic foundation, and search strategy is then determined from feature space according to similitude according to the method for finding homogenous characteristics.Specifically Image registration mode can be divided into following two major class again:
(1) method for registering images based on gray value information.
This method directly calculates relation between image using the half-tone information of image, the tightness degree of its relation is with similar Property metric function judges.This method can be divided into two major classes by different according to similarity function again:Image based on time domain is matched somebody with somebody Quasi- method and the method for registering images based on frequency domain.On the whole, this method is fairly simple, it is not necessary to complicated pretreatment, but The quality of similarity measurement degree directly affects the precision of image registration.Need to provide suitable phase when therefore, using this method Like property measurement.
(2) mode of distinguished point based.The image split-joint method of distinguished point based needs the characteristic point of extraction image in advance, The correspondence between image is found out further according to characteristic point, determines the overlapping region of image, then each width image is melted again Close.The difficult point of this method is the feature for how extracting and selecting robust, and how the matching relationship between feature determines.It is special That levies is various informative, such as has Harris corner features, Sift features, Gist features etc., it is necessary to be selected according to actual conditions The feature of robust.
Generally speaking, no matter from technical research situation, or practical situations, for grinding for three-dimensional panorama integration technology Study carefully all in development phase is explored, educational circles proposes many methods, but is currently available using and can meet the side that people require Method is also very limited.Particularly in panorama sketch effect and realize that the requirement of speed is extremely difficult to the demand of people.
The content of the invention
It is an object of the invention to provide a kind of magnanimity monitor video using and presentation mode to solve above-mentioned background technology The problem of middle proposition.
To achieve the above object, the present invention provides following technical solution:
A kind of magnanimity monitor video uses and presentation mode, comprises the steps of:
A, video image intercepts.
B, image registration:Piece image possesses different extractible features, what selection was wherein easily extracted, and can be The feature of image similarity subject to registration is characterized in a way as registering foundation.
C, image co-registration:Gray scale in the same Pixel-level of Same Scene, target to multi-source image carries out integrated treatment, The new images ultimately produced can accommodate the information of all pixels point in source images.
D, texture mapping is to threedimensional model.
As further scheme of the invention:Described image registration specifically includes:1. extract SIFT characteristics;2. calculate away from Matched from angle;3. RANSAC is purified.
As further scheme of the invention:Described image fusion specifically includes:A, single and to sort, b, full images melt Close.
Compared with prior art, the beneficial effects of the invention are as follows:The present invention propose it is a kind of using reality scene threedimensional model as Basis, merged with reference to multi-channel video, real-time video monitoring new method, and development system model machine, development system modularization are soft Part.Its feature is shown:1st, with real three-dimensional model foundation virtual environment, and in this, as video fusion carrier and final presentation Main body;2nd, using the RANSAC algorithms of belt restraining realize more video matchings with it is registering;3rd, realized using gaussian pyramid algorithm more The real-time processing and display of video fusion.
Brief description of the drawings
Fig. 1 is the work flow diagram of the present invention.
Fig. 2 is image registration flow chart.
Embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.It is based on Embodiment in the present invention, those of ordinary skill in the art are obtained every other without making creative work Embodiment, belongs to the scope of protection of the invention.
Please refer to Fig.1-2, in the embodiment of the present invention, a kind of magnanimity monitor video uses and presentation mode, includes following step Suddenly:
A, video image intercepts.
B, image registration:Piece image possesses different extractible features, what selection was wherein easily extracted, and can be The feature of image similarity subject to registration is characterized in a way as registering foundation.
C, image co-registration:Gray scale in the same Pixel-level of Same Scene, target to multi-source image carries out integrated treatment, The new images ultimately produced can accommodate the information of all pixels point in source images.
D, texture mapping is to threedimensional model.
Image registration specifically includes:1. extract SIFT characteristics;2. calculate distance to be matched with angle;3. RANSAC is carried It is pure.
Image co-registration specifically includes:A, it is single and to sort, b, full images fusion.
The present invention operation principle be:The technology path of this problem includes following three aspects:
(1) image registration
The image registration of feature based is most common method in registration.Piece image possesses different extractible spies Sign, selection are wherein easily extracted, and can characterize to a certain extent the feature of image similarity subject to registration as registration according to According to.Process description is illustrated in fig. 2 shown below.
This problem uses SIFT algorithms, which is exactly primarily to generate metric space, for more rulers of simulated image data It is necessary to spend characteristic dimension space, and realize that the unique linear core of change of scale is Gaussian convolution core, then piece image Metric space is defined as:
Wherein it is changeable scale Gaussian function., it is space coordinate, is scale coordinate.
Vector description after normalization of each SIFT feature comprising one 128 dimension, in this way, these features can Similitude is done in vector space to have detected.And the computing of the similarity measure based on distance and angle is basic operation, Speed is fast, so being suitable for thick matching stage.
Because thick matching algorithm, to there is the feature pair mismatched in obtained matching characteristic point, together all the time in set When extraction when also have certain trueness error, so needing a kind of very strong algorithm of robustness further to purify characteristic point It is right, so being exactly herein to use the RANSAC algorithm for estimating of robust in the first step work of thin matching stage.
(2) image co-registration
Image co-registration is a pith in image mosaic later stage, fusion it is bad, ghost just occurs, it is in addition digital Image co-registration is an important technology of graphical analysis, and the technology is in numerous necks such as panorama sketch, virtual reality, numerical map splicings Domain suffers from important application.
Fusion based on pixel, thought be to the gray scale in the Same Scene of multi-source image, the same Pixel-level of target into Row integrated treatment, the new images ultimately produced can accommodate the information of all pixels point in source images.Currently used method has Weighted mean method, be fade-in and gradually go out method and pyramid decomposition method.
Weighted mean method can be expressed as:The weighted average of pixel grey scale is to use to divide Analysis, obtains and characterizes each source images InThe weights W of certain feature power in (i, j)=(n=1,2 ..., N)n(i, j) (n=1, 2,...,N)。
The progressive fusion method gradually gone out is to be transitioned into back panel image in overlapped part the past width image uniform, It is exactly by the gray value of the overlapped area pixel of image, according to the image of some weights additive synthesis Cheng Xin, and prunes away The image section being staggered in vertical direction.The rgb value of pixel in corresponding two images lap is each denoted as r1, g1, b1And r2, g2, b2, then respective pixel RGB component value r in the image after fusion3, g3b3It can pass through
r3=d × r1+(1-d)×r2
g3=d × g1+(1-d)×g2
b3=d × b1+(1-d)×b2
Calculate and obtain, wherein d is gradual change coefficient.
Thought based on pyramid decomposition fusion is to do pyramid decomposition to the source images for participating in fusion first, then Using the pyramid of source images, suitable parameter is selected to merge the pyramid on each layer, finally by each gold after fusion Word tower carries out inverse transformation and can obtain fusion results image.
Laplace pyramids to establish image, first have to carry out Gauss pyramid decompositions to image:
Counted since the pyramidal top layers of Laplace, successively mode from top to bottom, carrying out recursion as the following formula can be extensive Its multiple corresponding Gauss pyramid, and finally obtain original image G0
(3) the 3D engines exploitation based on OGRE
In Three-dimensional Display software, rendering efficiency highly impacts the performance of display system very much, and rendering batch is Minimum rendering unit in display system, greatly influences rendering efficiency again in the middle in rendering for this monitoring system.Such as:Render 1000 units, if point 10 batches, per batch render 100, scene is averagely per second to render 80.2 two field pictures;If divide 100 batches It is secondary, it is averagely per second per lO scene of batch render to render 36.8 two field pictures, it is seen that to render the object of identical quantity, batch is fewer to be rendered Speed is faster, therefore renders batch and should try one's best and lack.
Common load mode, static geometry load mode is respectively adopted in this problem in OGRE three-dimensional graphics renderer engines Three-dimensional scene models are loaded and rendered with example geometry load mode, are compared by analysis, common loading method is worked as Rendering efficiency can be met the requirements when model quantity is seldom, and when quantity is on the increase, efficiency constantly declines, and Static geometry all keeps high efficiency to render from example geometry load mode in the case of different model quantity.Due to the original of mechanism Cause, all objects are unable to relative motion in static solid, it is necessary to as an overall action, and the single thing in example geometry Body can be acted freely.If the relatively motionless object that is much fixed up in magnanimity scene, it is possible to using quiet The mode of state geometry loads it, and if also to be acted to object later, can using several stress models of example Rendering efficiency is improved, and can meet functional requirement.
Although OGRE engine functions are powerful, without oneself dedicated database, this project model is needed by the OGRE Software of the third party establishes the model of model or utilization to establish.

Claims (3)

1. a kind of magnanimity monitor video uses and presentation mode, it is characterised in that comprises the steps of:
A:Video image intercepts;
B:Image registration:Piece image possesses different extractible features, what selection was wherein easily extracted, and can be at certain The feature of image similarity subject to registration is characterized in degree as registering foundation;
C:Image co-registration:Gray scale in the same Pixel-level of Same Scene, target to multi-source image carries out integrated treatment, finally The new images of generation can accommodate the information of all pixels point in source images;
D:Texture mapping is to threedimensional model.
2. a kind of magnanimity monitor video according to claim 1 uses and presentation mode is it is characterized in that, described image is matched somebody with somebody Standard specifically includes:1. extract SIFT characteristics;2. calculate distance to be matched with angle;3. RANSAC is purified.
3. a kind of magnanimity monitor video according to claim 1 uses and presentation mode is it is characterized in that, described image is melted Conjunction specifically includes:A, it is single and to sort, b, full images fusion.
CN201711043476.3A 2017-10-31 2017-10-31 A kind of magnanimity monitor video uses and presentation mode Pending CN107909637A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711043476.3A CN107909637A (en) 2017-10-31 2017-10-31 A kind of magnanimity monitor video uses and presentation mode

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711043476.3A CN107909637A (en) 2017-10-31 2017-10-31 A kind of magnanimity monitor video uses and presentation mode

Publications (1)

Publication Number Publication Date
CN107909637A true CN107909637A (en) 2018-04-13

Family

ID=61843013

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711043476.3A Pending CN107909637A (en) 2017-10-31 2017-10-31 A kind of magnanimity monitor video uses and presentation mode

Country Status (1)

Country Link
CN (1) CN107909637A (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090169102A1 (en) * 2007-11-29 2009-07-02 Chao Zhang Multi-scale multi-camera adaptive fusion with contrast normalization
CN103136739A (en) * 2011-11-29 2013-06-05 北京航天长峰科技工业集团有限公司 Registering method of controllable vidicon surveillance video and three-dimensional model in complex scene
CN105245841A (en) * 2015-10-08 2016-01-13 北京工业大学 CUDA (Compute Unified Device Architecture)-based panoramic video monitoring system
CN105516654A (en) * 2015-11-25 2016-04-20 华中师范大学 Scene-structure-analysis-based urban monitoring video fusion method
CN106530407A (en) * 2016-12-14 2017-03-22 深圳市金大象文化发展有限公司 Three-dimensional panoramic splicing method, device and system for virtual reality
CN107045730A (en) * 2017-03-06 2017-08-15 青岛鹏海软件有限公司 A kind of multidimensional exhibition system and method for digital culture scene or image

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090169102A1 (en) * 2007-11-29 2009-07-02 Chao Zhang Multi-scale multi-camera adaptive fusion with contrast normalization
CN103136739A (en) * 2011-11-29 2013-06-05 北京航天长峰科技工业集团有限公司 Registering method of controllable vidicon surveillance video and three-dimensional model in complex scene
CN105245841A (en) * 2015-10-08 2016-01-13 北京工业大学 CUDA (Compute Unified Device Architecture)-based panoramic video monitoring system
CN105516654A (en) * 2015-11-25 2016-04-20 华中师范大学 Scene-structure-analysis-based urban monitoring video fusion method
CN106530407A (en) * 2016-12-14 2017-03-22 深圳市金大象文化发展有限公司 Three-dimensional panoramic splicing method, device and system for virtual reality
CN107045730A (en) * 2017-03-06 2017-08-15 青岛鹏海软件有限公司 A kind of multidimensional exhibition system and method for digital culture scene or image

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王立楠: "多路视频流图像拼接***的研究与实现", 《中国优秀硕士学位论文全文数据库信息科技辑》 *

Similar Documents

Publication Publication Date Title
Zakharchenko et al. Quality metric for spherical panoramic video
Naemura et al. Real-time video-based modeling and rendering of 3D scenes
JP6201476B2 (en) Free viewpoint image capturing apparatus and method
CN105323252A (en) Method and system for realizing interaction based on augmented reality technology and terminal
CN103345736A (en) Virtual viewpoint rendering method
US8494254B2 (en) Methods and apparatus for image rectification for stereo display
US20220148297A1 (en) Image fusion method based on fourier spectrum extraction
CN106127680A (en) A kind of 720 degree of panoramic video fast browsing methods
US20100302234A1 (en) Method of establishing dof data of 3d image and system thereof
CN101916455A (en) Method and device for reconstructing three-dimensional model of high dynamic range texture
CN101729920A (en) Method for displaying stereoscopic video with free visual angles
CN109788270B (en) 3D-360-degree panoramic image generation method and device
Suenaga et al. A practical implementation of free viewpoint video system for soccer games
Wan et al. Drone image stitching using local mesh-based bundle adjustment and shape-preserving transform
An et al. A real-time three-dimensional tracking and registration method in the AR-HUD system
CN109902675A (en) The method and apparatus of the pose acquisition methods of object, scene reconstruction
CN108616746A (en) The method that 2D panoramic pictures based on deep learning turn 3D panoramic pictures
CN107909637A (en) A kind of magnanimity monitor video uses and presentation mode
CN114004773A (en) Monocular multi-view video synthesis method based on deep learning and reverse mapping
Lo et al. Acquiring 360° Light Field by a Moving Dual-Fisheye Camera
JP2021047468A (en) Image processing equipment, image processing method, and image processing program
Salehi et al. Alignment of cubic-panorama image datasets using epipolar geometry
Zhang et al. Design of a 3D reconstruction model of multiplane images based on stereo vision
Li et al. Omnidirectional neural radiance field for immersive experience
Naweed et al. Enhancement of 3D capture of room-sized dynamic scenes with Pan-Tilt-Zoom cameras

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: 20180413

RJ01 Rejection of invention patent application after publication