CN100458560C - Methd for generating spherical panorama based on full frame image - Google Patents

Methd for generating spherical panorama based on full frame image Download PDF

Info

Publication number
CN100458560C
CN100458560C CNB2004100931866A CN200410093186A CN100458560C CN 100458560 C CN100458560 C CN 100458560C CN B2004100931866 A CNB2004100931866 A CN B2004100931866A CN 200410093186 A CN200410093186 A CN 200410093186A CN 100458560 C CN100458560 C CN 100458560C
Authority
CN
China
Prior art keywords
full frame
frame image
spherical panorama
spatial model
image generation
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.)
Expired - Fee Related
Application number
CNB2004100931866A
Other languages
Chinese (zh)
Other versions
CN1707353A (en
Inventor
刘小华
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.)
SHANGHAI JIETU SOFTWARE TECHN CO Ltd
Original Assignee
SHANGHAI JIETU SOFTWARE TECHN CO Ltd
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 SHANGHAI JIETU SOFTWARE TECHN CO Ltd filed Critical SHANGHAI JIETU SOFTWARE TECHN CO Ltd
Priority to CNB2004100931866A priority Critical patent/CN100458560C/en
Publication of CN1707353A publication Critical patent/CN1707353A/en
Application granted granted Critical
Publication of CN100458560C publication Critical patent/CN100458560C/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Abstract

The present invention discloses method of creating spherical panorama based on full frame images. Through constituting two-dimensional spherical space model in two kinds of space structures, spherical panorama is created with eight, including six horizontal, one overhead and one ground, full frame images, or created with six, including four horizontal, one overhead and one ground, full frame images. According to the characteristics of full frame images, the present invention creates one new optimized formula through parametrizing the full frame images. The method of the present invention has fast image make-up speed, high automation, and high image definition.

Description

A kind of method that generates spherical panorama based on full frame image
Technical field
The present invention relates to a kind of panorama image generation method, relate in particular to a kind of method that generates spherical panorama based on full frame (fullframe) image.
Background technology
Panoramic technique is a kind of rendering technique based on image that development in recent years is got up, and the characteristics of this technology are that its drafting and the complexity of scene have nothing to do, and can reach the sense of reality of photo one-level.Use this technology, the user at first photographs one group of sample image of a certain scene by photographic equipment, then sample image is carried out amalgamation and obtain final panoramic picture.The spherical model panorama is meant that when playing it is panorama on the sphere at center that view data is mapped in the viewpoint, and the cube panorama then is that when playing view data to be mapped in the viewpoint be panorama on cube dignity at center.Whole spherical panorama is meant angular field of view direction 360 degree that is up to the standard, and vertical direction 180 degree do not have the panoramic picture of observing dead angles.Lead to the column model panorama is called the cylindricality panorama, and spherical model panorama and cube model panorama are referred to as spherical panorama.
The generation method of panoramic picture is more at present, but the shooting process complexity, amalgamation difficulty height, weak effect.Generate panoramic picture the multiple technologies scheme can be arranged, one of them: is based on the scheme of soft-shelled turtle eye or drum shaft images generation panorama, provide any N earlier and open soft-shelled turtle eye (N 〉=3) or drum shaft images (N 〉=4), the regeneration panoramic picture, this scheme can effectively solve existing algorithm 4 the above soft-shelled turtle eye pattern pictures of amalgamation and the problem of amalgamation drum shaft images automatically automatically, but weak point is, the panoramic picture sharpness that the traditional round fish eye images generates is not as drum shaft images, and The Cloud Terrace can not generate seamless global type panorama along horizontally rotating captured drum shaft images, have only 180 degree owing to be used to take the vertical visual angle (VFov) of the sigma camera lens of drum shaft images, this means, measuring error when the systematic error of camera lens and camera and photographer take will cause the top at panoramic picture, the bottom has a spot of information dropout.The another kind of technical scheme that generates the spherical panorama image based on six sheets of drum shaft images, can solve the problem of above-mentioned information dropout effectively, this scheme provides " three degree spherical spaces " model, based on this model a kind of style of shooting is proposed, can be used for solving the problem of mending the world in the panoramic picture amalgamation, but from definition effect, the clear effect of drum shaft images is not as full frame image, concerning full frame (full frame) camera lens, its vertical visual angle (VFov) is 150 degree, horizontal view angle (HFov) is 100 degree, is characterized in that the utilization factor of CCD imaging apparatus or egative film is 100%, shown among Fig. 1 103.If digital camera resolution is the 2000*3008 pixel, it is the panoramic picture of 3000*6000 pixel that the utilization full frame image generates resolution, then a picture element in the panoramic picture zone on average will 1.4 picture element interpolation be calculated from the sample image zone, and for the sigma camera lens, generate identical image, on average to calculate from 0.96 picture element interpolation, therefore can generate the higher panoramic picture of sharpness based on full frame image, its effect is than the panoramic picture better effects if that generates based on drum shaft images.Pursue the user of high definition, may be more prone to use full frame (full frame) camera lens, " three degree spherical spaces " model requires every width of cloth image adjacent with other four width of cloth image, and exists optimum right between the adjacent image.With six full frame images that adopt portrait (Portrait) pattern to take is example, because its horizontal view angle (HFov) is 100 degree, vertical angle of view (VFov) is 150 degree, it is about 10 degree that minimum between horizontal direction (being that longitude is 0 degree direction) four images in place overlaps the zone, practice shows, when the coincidence zone of adjacent image is 10 when spending, can not guarantee that the matching characteristic point between them can be come out by accurate coupling, to be respectively 60 degree and 35 then passable when spending for overlapping the zone.For adopting landscape (Landscape) pattern, its result and roughly the same last.
Existing panorama image generation method exists original image and takes complicated, low, the consuming time length of amalgamation process automation degree, the shortcoming that sharpness is low.
Summary of the invention
The technical problem to be solved in the present invention provides a kind of spherical panorama image generation method based on full frame image, improves the amalgamation problem of full frame image, generates the higher panoramic picture of sharpness, has proposed a kind of new optimization equation.
Method based on full frame image generation spherical panorama of the present invention comprises the steps: to set up spatial model; Adjacent feature point extracts and coupling; Set up and the solving-optimizing equation; Generate panoramic picture.
Setting up spatial model is to set up two kinds of different space structures, to handle different input sample images: three-dimensional model is out of shape, the bulbec face that only keeps two direction generates a kind of spatial model twice and adds top, low full frame image generation whole spherical panorama to support four levels; The new construction of setting up a kind of spatial model twice adds top, low full frame image generation whole spherical panorama to support six levels.
Vow basic coplane at six pairing centers of horizontal full frame image, the angle between adjacent center is vowed is 60 degree; Basic coplane is vowed at the center of certain two width of cloth full frame image in top, low image and the horizontal image, and the angle between the adjacent center arrow is 90 degree.
Setting up spatial model also comprises every full frame image 5 privately owned positional parameters of setting and 3 publicly-owned transformation parameters; 5 privately owned positional parameters comprise 3 space orientation parameters (rotate, wave, pitching) and 2 picture centre positional parameters (x, y); 3 publicly-owned transformation parameters comprise: α and β in effective visual angle (FOV), the camera lens nonlinearities change (r=α * sin (β * θ)), wherein r is the distance that point-of-interest arrives the regional center of circle of optimum envelope circle, θ be camera lens axially with from the optical center point to this point-of-interest actual scenery the angle of institute's corresponding point line, α is a zoom factor, thereby realize the mapping one by one of putting on actual scene and soft-shelled turtle eye or the drum shaft images plane, β is mapping parameters radially.
The present invention proposes a kind of new optimization equation, because systematic error or measuring error, make along the light of the axial incident of camera lens, at the center on CCD or film not of the projection on CCD or the film, be the center of image, and this small deviation lacks certain standard and judges and measure, can position by the garden, border as the center of soft-shelled turtle eye and drum shaft images, the present invention sets up the optimization equation, is meant based on the optimum of adjacent image collection is set up the optimization equation: min Σ i N Σ k k i | | f i ( F ik , V ) - g i ( B ik , V ) | | , Wherein, f i(, V) and g i(, V) be the variation function that acts on the adjacent back width of cloth figure of i figure respectively with it; In the optimization equation central value is made as unknown quantity, it is finely tuned, with the result who obtains more to optimize by matching characteristic point.The V that optimizes in the equation is the parameter space of spatial model, and it is the publicly-owned parameter of all images and the set of privately owned parameter.
Since adopt method of the present invention, image amalgamation process automation degree height, and the user only need import original image, can generate panorama sketch automatically; Amalgamation speed is fast, at AMD Athlon (tm) XP2800+, and on the loom of 1.024G internal memory, the drum shaft images that eight resolution of amalgamation are 3008 * 2000 pixels, generating resolution is the panoramic picture of 3000 * 6000 pixels, whole amalgamation process is no more than 3 minutes; The image resolution ratio height.Not only improve picture quality, and save time, reduce cost.
Description of drawings
Fig. 1 is used sample image among the present invention;
Fig. 2 is an ideal structure synoptic diagram of model that the present invention sets up;
Fig. 3 is another ideal structure synoptic diagram of model that the present invention sets up;
Fig. 4 is a realization flow synoptic diagram of the present invention.
Embodiment
The present invention is further detailed explanation below in conjunction with accompanying drawing.
Fig. 1 has provided the sample image exemplary plot that relates among the present invention.Wherein 101 is under portrait (Portrait) pattern, utilization full frame (full frame) lens shooting gained full frame image, and 102 is under landscape (Landscape) pattern, the gained full frame image.
As shown in Figure 2, it is an ideal structure of two degree spherical space models described in the invention, its principle and structure are identical with three degree spherical space models, but because the image of being analyzed is a full frame image, the image information of this type can not guarantee that all there are enough coincidence zones in all adjacent spherical interior surfaces of its correspondence, therefore, when this structural drawing of utilization, might have only four secondary bulbec masks of both direction place bulbec face to have adjacent matching characteristic point right, at this moment, this structural map deteriorates to spherical space model twice.
As shown in Figure 3, it is another ideal structure of model that the present invention sets up, 301,302,303,304,305,306,307,308 represent eight different directions of sphere in true coordinate system, the intersection point of they and sphere is respectively A, B, C, D, E, F, G, H.Here A, B, C, D, E, F coplane, same for latitude on the sphere is 0 position, A, G, D, H coplane are set at longitude on the sphere and are 0 position.Photographer adopts Portrait pattern camera lens axially to begin along 301 directions, horizontal direction is rotated 60 degree successively and is taken six full frame images, and vertically upward and be rotated down camera lens 90 degree along the some directions in the horizontal six direction (for example 301 directions among the figure), make camera lens axially be positioned at 307 directions and top and bottom image of 308 directions bat, during shooting, the center that should keep camera lens is same point in the space roughly.The image that uses this image pickup method to obtain exists optimum right between six spherical interior surface adjacent images of its horizontal direction, exists optimum right between four spherical interior surface adjacent images of vertical direction.
Referring to Fig. 4, behind the full frame image of packing into, at first according to the difference of input picture, different space structures is set up in utilization 401, when input picture is four horizontal images when adding low two images in top (being called for short 4+2), sets up spatial model as shown in Figure 2; When input picture is six horizontal images when adding low two images in top (being called for short 6+2), set up spatial model as shown in Figure 3.Every width of cloth image possesses 3 private room positional parameters, 2 privately owned centralized positioning parameters and 3 publicly-owned transformation parameters.Wherein 3 private room positional parameters are meant rotation (roll), wave (yaw), pitching (pitch), 2 privately owned centralized positioning parameters, be meant centre coordinate (x with the sample input picture, y) be made as known variables, this is because because the influence of systematic error and test error, the projection of the axial place of camera lens light on image may off-center, thereby cause the error accumulation in model parameter is found the solution, influence the amalgamation quality, therefore need be with the picture centre coordinate parametersization, so that be optimized, 3 publicly-owned transformation parameters are meant effective visual angle (FOV), α and β among the camera lens nonlinearities change r=α * sin (β * θ), wherein r is the distance that point-of-interest arrives the regional center of circle of optimum envelope circle, θ be camera lens axially with from the optical center point to this point-of-interest actual scenery the angle of institute's corresponding point line, α is a zoom factor, thereby realize the mapping one by one of putting on actual scene and soft-shelled turtle eye or the drum shaft images plane, β is mapping parameters radially.402 pairs of adjacent images carry out feature point extraction and coupling, to find out optimum to collection.At first the image to horizontal direction carries out Feature Points Matching, and the image to vertical direction carries out Feature Points Matching again.Match above-mentioned unique point to after, therefrom select out again optimum to the collection ∪ i = 1 N R i = ∪ i N ∪ k = 1 K i ( F ik , B ik ) , Wherein, R iThe optimum that is i group adjacent image is right, F IkBe the k unique point of last secondary figure in the i group adjacent image, B IkUnique point for the back secondary figure of correspondence.403 set up and the solving-optimizing equation, at first set up based on the optimization equation of optimum to collection, find the solution this optimization equation again, draw model unknown parameter value.Principle that set up to optimize equation is, sets up the transformation equation about parameter space, uses this equation that optimum is carried out conversion to collection, make optimum to the arbitrfary point of collection to transforming to the same position in the space.Thus, optimizing the equation expression formula is: min Σ i N Σ k k i | | f i ( F ik , V ) - g i ( B ik , V ) | | , Wherein, f i(, V) and g i(, be respectively the variation function that acts on the adjacent back width of cloth figure of i figure V) with it, V is the parameter space of model, and it is the publicly-owned parameter of all images and the set of privately owned parameter, and its number of parameters is 5*N+3.‖ ‖ is an Euclidean distance.404 based on above parameter value, generates panoramic picture.

Claims (6)

1. the method based on full frame image generation spherical panorama is characterized in that, comprises the steps:
(1) sets up spatial model;
(2) adjacent feature point extracts and coupling;
(3) set up also solving-optimizing equation;
(4) generate panoramic picture;
Wherein set up spatial model, be to set up two kinds of different spatial models twice, to handle different input sample images, a kind of spatial model twice is that three-dimensional model is out of shape, the bulbec face that only keeps two direction generates a kind of and four horizontal Jia Ding, end full frame image generates the corresponding spatial model twice of whole spherical panorama; Another kind of spatial model twice is to set up a kind of and six horizontal Jia Ding, the corresponding spatial model twice of end full frame image generation whole spherical panorama.
2. according to claims 1 described method, it is characterized in that the angle between wherein six pairing adjacent center of horizontal full frame image are vowed is 60 degree based on full frame image generation spherical panorama.
3. according to claims 1 described method, it is characterized in that the angle between wherein the adjacent center of certain two width of cloth full frame image in top, base map picture and the horizontal image is vowed is 90 degree based on full frame image generation spherical panorama.
4. according to claims 1 described method, it is characterized in that, wherein set up spatial model and also comprise every full frame image 5 privately owned positional parameters of setting and 3 publicly-owned transformation parameters based on full frame image generation spherical panorama; 5 privately owned positional parameters comprise rotation, wave, 3 space orientation parameters of pitching and x, y2 picture centre positional parameter; 3 publicly-owned transformation parameters comprise: α and β among effective visual angle FOV, the camera lens nonlinearities change r=α * sin (β * θ), wherein r is the distance that point-of-interest arrives the regional center of circle of optimum envelope circle, θ be camera lens axially with from the optical center point to this point-of-interest actual scenery the angle of institute's corresponding point line, α is a zoom factor, thereby realize the mapping one by one of putting on actual scene and soft-shelled turtle eye or the drum shaft images plane, β is mapping parameters radially.
5. according to claims 1 described method, it is characterized in that, wherein set up and optimize equation, be meant based on the optimum of adjacent image collection is set up the optimization equation based on full frame image generation spherical panorama: min Σ i N Σ k k i | | f i ( F ik , V ) - g i ( B ik , V ) | | , Wherein, f i(, V) and g i(, V) be the variation function that acts on the adjacent back width of cloth figure of i figure respectively with it; In the optimization equation central value is made as unknown quantity, it is finely tuned, with the result who obtains more to optimize by matching characteristic point.
6. according to claims 5 described methods based on full frame image generation spherical panorama, it is characterized in that the V that wherein optimizes in the equation is the parameter space of spatial model, it is the publicly-owned parameter of all images and the set of privately owned parameter.
CNB2004100931866A 2004-12-17 2004-12-17 Methd for generating spherical panorama based on full frame image Expired - Fee Related CN100458560C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CNB2004100931866A CN100458560C (en) 2004-12-17 2004-12-17 Methd for generating spherical panorama based on full frame image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CNB2004100931866A CN100458560C (en) 2004-12-17 2004-12-17 Methd for generating spherical panorama based on full frame image

Publications (2)

Publication Number Publication Date
CN1707353A CN1707353A (en) 2005-12-14
CN100458560C true CN100458560C (en) 2009-02-04

Family

ID=35581328

Family Applications (1)

Application Number Title Priority Date Filing Date
CNB2004100931866A Expired - Fee Related CN100458560C (en) 2004-12-17 2004-12-17 Methd for generating spherical panorama based on full frame image

Country Status (1)

Country Link
CN (1) CN100458560C (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2011155361A (en) * 2010-01-26 2011-08-11 Sony Corp Imaging apparatus, imaging control method, and program
CN102291527B (en) * 2011-08-11 2014-02-12 杭州海康威视数字技术股份有限公司 Panoramic video roaming method and device based on single fisheye lens
CN102426703B (en) * 2011-08-11 2014-04-09 重庆市勘测院 Course and pitch angle speed acquisition method of vehicle panoramic image acquisition platform
CN106899840B (en) * 2017-03-01 2018-06-05 北京大学深圳研究生院 Panoramic picture mapping method
CN107274455A (en) * 2017-07-07 2017-10-20 东北林业大学 Mix the three-dimensional rebuilding method of panoramic image in vision system
CN113329182A (en) 2017-08-31 2021-08-31 深圳市大疆创新科技有限公司 Image processing method, unmanned aerial vehicle and system

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001133256A (en) * 1999-11-05 2001-05-18 Asia Air Survey Co Ltd Method of calculating position and attitude of aerial photograph
CN1437165A (en) * 2003-01-24 2003-08-20 上海杰图软件技术有限公司 Intelligent panorama creating process based on two fish-eye images
CN1452730A (en) * 2000-03-22 2003-10-29 爱格斯露申光电子有限公司 Panoramic image acquisition device

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001133256A (en) * 1999-11-05 2001-05-18 Asia Air Survey Co Ltd Method of calculating position and attitude of aerial photograph
CN1452730A (en) * 2000-03-22 2003-10-29 爱格斯露申光电子有限公司 Panoramic image acquisition device
CN1437165A (en) * 2003-01-24 2003-08-20 上海杰图软件技术有限公司 Intelligent panorama creating process based on two fish-eye images

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
一种全自动稳健的图像拼接融合算法. 赵向阳,杜利民.中国图象图形学报,第9卷第4期. 2004
一种全自动稳健的图像拼接融合算法. 赵向阳,杜利民.中国图象图形学报,第9卷第4期. 2004 *

Also Published As

Publication number Publication date
CN1707353A (en) 2005-12-14

Similar Documents

Publication Publication Date Title
CN101000461B (en) Method for generating stereoscopic panorama by fish eye image
JP7162933B2 (en) Method, apparatus and system for establishing an internal spatial model of an object, and computer apparatus and computer readable storage medium
CN105243637B (en) One kind carrying out full-view image joining method based on three-dimensional laser point cloud
JP5054971B2 (en) Digital 3D / 360 degree camera system
Zhang et al. A UAV-based panoramic oblique photogrammetry (POP) approach using spherical projection
CN107492069B (en) Image fusion method based on multi-lens sensor
CN109314752A (en) Effective determination of light stream between image
Huang et al. Panoramic imaging: sensor-line cameras and laser range-finders
CN110213475A (en) A kind of image pickup method of panoramic picture
CN102984453A (en) Method and system of real-time generating hemisphere panoramic video images through single camera
CN107454377B (en) A kind of algorithm and system carrying out three-dimensional imaging using camera
Lin et al. A low-cost portable polycamera for stereoscopic 360 imaging
CN207218846U (en) Circulating type mixes light field imaging device
CN114926612A (en) Aerial panoramic image processing and immersive display system
Koeva 3D modelling and interactive web-based visualization of cultural heritage objects
CN100458560C (en) Methd for generating spherical panorama based on full frame image
CN106683057A (en) Method for automatically geometrically correcting multi-projection display walls with optional continuous screens with curved surfaces
CN100390666C (en) Intelligent method for fast generating high-definition panorama based on round fish eye or drum shape image
JP2004139294A (en) Multi-viewpoint image processing program, system, and marker
CN114332383B (en) Scene three-dimensional modeling method and device based on panoramic video
CN108510537B (en) 3D modeling method and device
Gonzalez-Aguilera et al. Application of non-destructive techniques to the recording and modelling of palaeolithic rock art
CN111031259B (en) Inward type three-dimensional scene acquisition virtual compound eye camera
CN100426139C (en) Method for generating whole spherical panorama based on six sheets of drum shaft images
TW413796B (en) A method and system for establishing environment image

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
EE01 Entry into force of recordation of patent licensing contract

Assignee: Shanghai Zendai jietu Network Technology Co.,Ltd.

Assignor: SHANGHAI JEITU SOFTWARE Co.,Ltd.

Contract fulfillment period: 2009.9.22 to 2023.1.24

Contract record no.: 2009310000282

Denomination of invention: Methd for generating spherical panorama based on full frame image

Granted publication date: 20090204

License type: Exclusive license

Record date: 20091015

LIC Patent licence contract for exploitation submitted for record

Free format text: EXCLUSIVE LICENSE; TIME LIMIT OF IMPLEMENTING CONTACT: 2009.9.22 TO 2023.1.24; CHANGE OF CONTRACT

Name of requester: SHANGHAI ZHENGDA JIETU NETWORK TECHNOLOGY CO., LTD

Effective date: 20091015

EC01 Cancellation of recordation of patent licensing contract

Assignee: Shanghai Zendai jietu Network Technology Co.,Ltd.

Assignor: SHANGHAI JEITU SOFTWARE Co.,Ltd.

Contract record no.: 2009310000282

Date of cancellation: 20130327

LICC Enforcement, change and cancellation of record of contracts on the licence for exploitation of a patent or utility model
EE01 Entry into force of recordation of patent licensing contract

Application publication date: 20051214

Assignee: BEIJING BAIDU NETCOM SCIENCE AND TECHNOLOGY Co.,Ltd.

Assignor: SHANGHAI JEITU SOFTWARE Co.,Ltd.

Contract record no.: 2014310000043

Denomination of invention: Methd for generating spherical panorama based on full frame image

Granted publication date: 20090204

License type: Common License

Record date: 20140321

LICC Enforcement, change and cancellation of record of contracts on the licence for exploitation of a patent or utility model
DD01 Delivery of document by public notice

Addressee: Patent of Shanghai jietu Software Technology Co.,Ltd. The person in charge

Document name: payment instructions

DD01 Delivery of document by public notice
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20090204

Termination date: 20211217

CF01 Termination of patent right due to non-payment of annual fee