CN103440619B - A kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method - Google Patents
A kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method Download PDFInfo
- Publication number
- CN103440619B CN103440619B CN201310231478.0A CN201310231478A CN103440619B CN 103440619 B CN103440619 B CN 103440619B CN 201310231478 A CN201310231478 A CN 201310231478A CN 103440619 B CN103440619 B CN 103440619B
- Authority
- CN
- China
- Prior art keywords
- sampling
- resolution
- oblique pattern
- pattern sampling
- oblique
- 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
Links
Abstract
The invention discloses a kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method, to realize the reconstruction of high spatial resolution satellite remote sensing images, the above method includes:Cell coordinate system on oblique pattern sampling grid is mapped to the cell coordinate system on high-resolution routine sampling grid;In space coordinates, triangulation is carried out to the pixel point of oblique pattern sampling;For the unknown pixel in fine-resolution meshes, as the known pixel value of the triangular apex where it, the unknown pixel value in fine-resolution meshes is calculated using super resolution ratio reconstruction methods such as polynomial interopolations;Method disclosed by the invention, using the oblique pattern sampling imaging of single line battle array, avoids the problem of two rows of sensor linear array manufactures are difficult, registration accuracy is unmanageable;At the same time by controlling direction sampling interval of pushing broom to carry out image data acquiring, it is ensured that oblique pattern sampling institute is square into spatial sampling grid, avoids loss of significance problem caused by extra geometric correction and gray-level interpolation.
Description
Technical field
The invention belongs to Technique of Super-resolution Image Construction field, more particularly to a kind of tiltedly pattern sampling modeling and super-resolution
Rate method for reconstructing.
Background technology
The Optical remote satellite of French SPOT -5 of transmitting in 2002 employs Super-mode sampling technology first, utilizes two misarrangements
The line array sensor for opening half pixel obtains image, in the case of sensor photosensitive elemental size is not changed, by spatial resolution by
5m has brought up to 2.5m.Success of the sub-pixed mapping sampling technique on SPOT-5 satellites is embodied as lifting remote sensing images spatial resolution
Research introduce new thinking.Super-resolution rebuilding based on the sampling of oblique pattern is soft under a kind of specific imaging pattern is supported
The method of combination of hardware room for promotion resolution ratio, by varying the imaging pattern of sensor linear array, it is intended to above carried from hardware sampling
High effective resolution.This kind of method sets about realizing the super-resolution rebuilding of remote sensing images in terms of two, is on the one hand set on hardware
New sensor sample pattern is counted, hardware sampling nature is on the other hand considered on software and various image deterioration factors are set
The mathematical model and algorithm of super-resolution image reconstruction are counted, to make up the limitation of Software-only method.Normal mode sampling under into
As linear array direction is vertical with direction of pushing broom, and tiltedly pattern sampling is will to be imaged linear array to tilt certain angle, while control is pushed broom
Direction sampling interval is imaged.
In order to lift the spatial resolution of satellite remote sensing images, space office of country of France successively propose by height mode,
The method of hyper mode and oblique pattern spatial domain sampling skill upgrading remote sensing images spatial resolution.Height mode and Super-mode sampling master
If by improving the spatial sampling frequencies of imaging linear array, to reduce spectral aliasing effect.Hyper mode and Hiper-mode sampling are all
Imaging linear array of two mutual displacements for half of array element is loaded with remote sensing platform, except that Super-mode sampling is in the side of pushing broom
The upward time of integration is identical with normal mode, and Hiper-mode sampling is the 1/ of normal mode in the time of integration on direction of pushing broom
2, improve space sampling densities by improving time sampling frequency.For both sampling configurations, hyper mode and height mode
Space sampling densities are respectively 2 times and 4 times of normal mode.However, the problem of both sampling configurations, is:1) two rows of imaging lines
Battle array adds platform complex;2) when being clapped in satellite side, in order to ensure two rows of linear arrays stagger half of array element all the time, it is also necessary to increase
Extra hardware device.
Single imaging linear array is used only in oblique pattern sampling, its linear array direction is no longer vertical with direction of pushing broom, but tilts one
Fixed angle, while control direction sampling interval of pushing broom to carry out image data acquiring, in the case where not changing array element size, carry
The spatial resolution of hi-vision.The manufacture cost for as linear array, reducing imaging device is lined up using only one since oblique pattern is sampled,
The problem of two rows of imaging linear array manufactures are difficult, registration accuracy is unmanageable is avoided at the same time.It is limited to the hypersensitivity of the technology, state
The outer play-by-play for rarely having correlation technique.China is still in theoretical validation and imaging device in oblique pattern remote sensing satellite design aspect
Principle prototype development stage, only a small number of documents are related to the method.It is parallel that all peaks etc., which propose a kind of spatial sampling grid,
The oblique pattern sample mode of quadrangle, former spatial relationship is obtained after then carrying out simple geometry correction to oblique pattern sampled images
Under image, the problem of this sampling configuration is:1) gray-level interpolation causes geometric distortion;2) it is imaged the angle of inclination of linear array at the same time
Constrain sampling density and visual field breadth.
The content of the invention
The shortcomings that in order to overcome the above-mentioned prior art, it is an object of the invention to provide a kind of tiltedly pattern sampling modeling and surpass
Resolution reconstruction method, avoids loss of significance problem caused by extra geometric correction and gray-level interpolation, while ensure that and adopt
Sample density and visual field breadth.
To achieve these goals, the technical solution adopted by the present invention is:
A kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method, include the following steps:
Pushed broom pattern based on linear array, oblique mode data collection is carried out using single line battle array;
Cell coordinate system on oblique pattern sampling grid is mapped to the cell coordinate on high-resolution routine sampling grid
System;
In space coordinates, triangulation is carried out to the pixel point of oblique pattern sampling;
For the unknown pixel in fine-resolution meshes, as the known pixel value of the triangular apex where it, using super
Resolution reconstruction method calculates the unknown pixel value in fine-resolution meshes.
In data acquisition, it is imaged linear array and direction of pushing broom is angled, and by controlling direction of pushing broom to sample
Spacing so that oblique pattern sampling institute is square into spatial sampling grid.
26.56 ° are the angle ranging from, imaging linear array is along the sampling interval in direction of pushing broom, along and vertical direction of pushing broom
Sampling interval isAlternatively,
63.44 ° are the angle ranging from, imaging linear array is along the sampling interval in direction of pushing broomEdge and vertical direction of pushing broom
Sampling interval is
In cell coordinate system maps, by oblique pattern sampling grid with base vector
Continuation, alternatively, with base vectorContinuation.
In space coordinates, Delaunay Triangulation is carried out to the pixel point of oblique pattern sampling.
In super-resolution rebuilding, the unknown pixel value in fine-resolution meshes is calculated using polynomial interopolation.
Compared with prior art, the oblique pattern sampling of single line battle array of the present invention is tilted a certain angle line array sensor, so that
Avoid the problem of two rows of sensor linear array manufactures are difficult, registration accuracy is unmanageable.At the same time by controlling direction of pushing broom to sample
Spacing carries out image data acquiring, it is ensured that oblique pattern sampling institute is square into spatial sampling grid, avoids extra geometry
Loss of significance problem caused by correction and gray-level interpolation.
Brief description of the drawings
Fig. 1 is overview flow chart of the present invention.
Fig. 2 is tiltedly pattern sampling model figure of the invention.
Fig. 3 is another tiltedly pattern sampling model figure of the invention.
Fig. 4 is that the coordinate system of tiltedly pattern sampling grid of the invention represents.
Fig. 5 is that the coordinate system of square sample grid of the present invention represents.
Fig. 6 is Delaunay Triangulation process schematic of the present invention.
Fig. 7 is tiltedly pattern sample interpolation schematic diagram of the invention.
Fig. 8 is tiltedly pattern sampled light integration schematic diagram of the invention.
Embodiment
The embodiment that the present invention will be described in detail with reference to the accompanying drawings and examples.
As shown in Figure 1, the overall procedure of the present invention includes the following steps:Linear array data acquisition, coordinate system mapping, triangle cut open
Point and super-resolution rebuilding.Data acquisition uses the push-broom imaging mode based on linear array, line by line gathered data, and imaging linear array is with pushing away
Broom direction is angled, and by controlling direction sampling interval of pushing broom so that oblique pattern samples institute and is into spatial sampling grid
Square.
Linear array data acquisition uses the oblique pattern of single line battle array, if c represents the size of sensor array element,Represent resolution grid
Distance between middle pixel.The problem of causing in order to avoid geometric deformation, it is ensured that oblique pattern sampling institute is into spatial sampling grid for just
It is square, that is, require the point of oblique pattern sampling to fall in the fine-resolution meshes of interpolation, be expressed as with relationshipUnder this relation, angle of inclinationFig. 2 gives the present invention and builds
Vertical oblique pattern sampling model, imaging linear array and direction of pushing broom are sampled into 26.56 ° of angles, sampling of the imaging linear array along direction of pushing broom
Spacing isIt is along the sampling interval with vertical direction of pushing broomAlternatively, as shown in figure 3, imaging linear array and push broom direction into
63.44 ° of angle samplings, imaging linear array are along the sampling interval in direction of pushing broomIt is along the sampling interval with vertical direction of pushing broom
If s is the sampling interval pushed broom and linear array is imaged on direction, then there are s=υ t, wherein, υ is the speed of pushing broom of imaging linear array
Degree, t is the time of integration.Obviously, in the case of constant airspeed of pushing broom, can only generally be reduced by reducing the time of integration to reach
The purpose of pixel sampling interval, but reduce the time of integration and may result in that photosensitive member is under-exposed, causes signal noise ratio (snr) of image to decline
The problems such as.Therefore, in order to ensure the time of integration of linear array data acquisition, imaging needs imaging linear array twice in succession pushing broom on direction
With larger sampling interval.
Oblique pattern sampled images be defined onThe rule formed for base vector continuation
Then sampling grid, is represented by
In formula, { e1, e1It is spatial domainBase vector.Especially, the oblique pattern sampling grid shown in Fig. 3 be withFor base vector continuation.
The base vector of most common square sample grid isSquare sample net
Lattice are represented byFig. 4 gives the coordinate system expression of oblique pattern sampling grid, coordinate
About it is set to (n1, n2), the coordinate system that Fig. 5 gives square sample grid represents that coordinate is about set to (m1, m2).Oblique pattern sampling
Reconstruction is to estimate pixel point unknown on square sample grid by the known pixel point of oblique pattern sampling acquisition, by oblique pattern
Sampled images are redeveloped into high-definition picture.
The present embodiment obtains high-definition picture using the method for space scatterplot interpolation.Interpolation is a kind of with real-time
Property, suitable for hard-wired super resolution ratio reconstruction method, but the present embodiment is not limited using interpolation method, can also be used
Other increasingly complex super resolution ratio reconstruction methods.Interpolation problem in the present embodiment can be attributed to the triangulation of planar point set
Problem.Triangulation is carried out to space scatterplot on two dimensional surface, interpolation curved surface is then constructed on triangular domain, with interpolation curved surface
Based on estimate unknown point value.
1908, G.Voronoi mathematically defined the effective range of each discrete points data first, i.e., it has
The scope of reflecting regional information is imitated, and defines the Voronoi diagram on two dimensional surface.1934, B.Delaunay by
Voronoi diagram, which develops, to be easier to the Delaunay Triangulation that analysis is applied.Voronoi diagram is cutd open with Delaunay triangles
It is divided into the powerful of analyzed area discrete data.Delaunay triangulations and Voronoi diagram antithesis each other, connect institute
There is the organic centre of adjacent Voronoi polygons, the Delaunay Triangulation of antithesis can be formed by Voronoi diagram.
One critical nature of Delaunay Triangulation is sky circumscribed circle property, i.e., for any Delaunay triangles, the triangle
The interior zone of shape circumscribed circle does not include other any points, all Delaunay triangles non-overlapping copies, and intactly covers whole
A spatial domain.Fig. 6 gives the schematic diagram that Delaunay triangulations are carried out to the known pixel point of oblique pattern sampling, figure
In solid origin represent known to pixel point, the unknown pixel point of cross × expression.
For each triangle T of Delaunay Triangulationk, linear interpolation method is by 3 of known triangle tops
The pixel value f of pointi(xi, yi), i=0,1,2, construct a plane:
In formula,WithFor the coefficient of polynomial function.According to each triangle Tk3 vertex pixel value
fi(xi, yi), i=0,1,2 simultaneous establishes system of linear equations to solve multinomial coefficient.Then arbitrary point is obtained using formula (2)
The pixel value at (x, y) place.
Cubic interpolation method constructs cubic surface by 6 points of neighborhood.Linear interpolation has continuity, but it is not one
Secondary to lead, the derivative on the boundary line of all triangles can produce jump.Cubic interpolation has the once property led, but will
Ask the calculating time of higher.Fig. 7 gives the linear interpolation and cubic interpolation schematic diagram of oblique pattern sampling, is established in the present invention
In oblique pattern sampling model, line segment P0M1=P1P2, for this sampling grid with proportional spacing, linear interpolation is actually
It can simplify and be expressed as M1=(P1+P2)/2, M0=(P0+M1)/2。
Assuming that the point spread function of imaging system is uniform function, down-sampling model of the invention passes through to high resolution graphics
The luminous intensity that photosensitive first surface is simulated as carrying out double integral is integrated.If f (x1, x2) representing real consecutive image, Fig. 8 gives
Oblique pattern sampled light integration schematic diagram is gone out, for the luminous intensity ∫ ∫ of any photosensitive member collectionΩf(x1, x2)dx1dx2, by its stroke
It is divided into three plane domain Ω1、Ω2And Ω3, these three plane domains x1Axis and x2The range of integration of axis is
Wherein,Single photosensitive member collection
Luminous intensity can be modeled asWherein, c2Represent the area of photosensitive member covering.
The spatial sampling grid that oblique pattern sample mode is formed is parallelogram, therefore institute is spatially into image
Inclined, used super resolution ratio reconstruction method is only to carry out geometric correction to image according to linear array angle of inclination.It is this oblique
The serious problems that mode imaging mode is brought are that the angle of inclination for being imaged linear array can influence sampling density and visual field breadth at the same time,
And the influence of these two aspects mutually restricts, with the increase at imaging linear array angle of inclination, space sampling densities increase, but together
When visual field breadth reduce.And oblique pattern samples the spatial sampling grid to be formed as square in the present invention, therefore institute is into image
Extra geometric correction and gray-level interpolation is not required., must in order to ensure oblique pattern sampling institute into spatial sampling grid to be square
Direction sampling interval of pushing broom must be controlled to carry out image data acquiring.By jointly controlling acquisition speed and speed of pushing broom, make
Be imaged linear array is along the spacing between direction data acquisition twice of pushing broom。
The spatial resolution of image is usually weighed with the sampling number (dots per inch, dpi) of per inch, this reality
What is measured on border is sampling density.The sampling density r of space latticenomIt is defined as
In formula, the sampling number of the sampling density, i.e. unit area of ρ (Г) representation space grid Г.It is with dpi
Mode metric space resolution ratio.Nominal resolution (Nominal resolution) RnomIt is defined as
Nominal resolution represents the distance between neighbouring sample point.
Usually using the expression formula metric space resolution ratio of array element size c.The space lattice base vector of normal mode sampling
For e1=(c, 0)T、e2=(0, c)T, sampling density rnom=c-2, nominal resolution Rnom=c.The space lattice of Super-mode sampling
Base vector is e1=(c, 0)T、Sampling density rnom=2c-2, nominal resolutionGao Mo
Formula sampling space lattice base vector beSampling density rnom=4c-2, it is nominal to differentiate
RateTiltedly the space lattice base vector of pattern sampling is Sampling densityNominal resolutionTiltedly
Pattern sampling density is 1.25 times of normal mode sampling, but compared with height mode and Super-mode sampling, oblique pattern sampling is simultaneously
Do not increase spatial sampling frequencies.In these sampling techniques, oblique pattern sampling is improved in the case where not changing array element size
The ground pixel resolutions of remote sensing images.
Claims (1)
1. a kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method, it is characterised in that include the following steps:
Push-broom imaging mode based on linear array, oblique mode data collection is carried out using single line battle array, in data acquisition, imaging
Linear array is 4p or 2p by controlling direction sampling interval of pushing broom with pushing broom direction into 26.56 ° or 63.44 ° of angles, by oblique pattern
Sampling grid is with base vector e1=(4p, 0)T、e2=(2p, p)TContinuation, alternatively, with base vector e1=(2p, 0)T、e2=(p, 2p)T
Continuation;
Cell coordinate system on oblique pattern sampling grid is mapped to the cell coordinate system on high-resolution routine sampling grid, net
Distance in lattice between pixel is p;
In space coordinates, Delaunay Triangulation is carried out to the pixel point of oblique pattern sampling;
For the unknown pixel in fine-resolution meshes, as the known pixel value of the triangular apex where it, using multinomial
Unknown pixel value in interpolation calculation fine-resolution meshes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310231478.0A CN103440619B (en) | 2013-06-09 | 2013-06-09 | A kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310231478.0A CN103440619B (en) | 2013-06-09 | 2013-06-09 | A kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103440619A CN103440619A (en) | 2013-12-11 |
CN103440619B true CN103440619B (en) | 2018-04-17 |
Family
ID=49694311
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310231478.0A Expired - Fee Related CN103440619B (en) | 2013-06-09 | 2013-06-09 | A kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103440619B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR3033071B1 (en) * | 2015-02-25 | 2017-02-17 | Centre Nat D'etudes Spatiales (Cnes) | METHOD FOR DETERMINING A SET OF MESHES FOR IMAGING A GEOGRAPHICAL AREA |
CN105973205B (en) * | 2016-05-13 | 2019-04-02 | 中国人民解放军装备学院 | It is a kind of to intersect oblique sampling configuration high-resolution imaging method |
CN106791405B (en) * | 2016-12-26 | 2019-05-21 | 首都师范大学 | A kind of method and system obtaining Displacement sequence image |
CN106920213B (en) * | 2017-01-19 | 2021-02-12 | 首都师范大学 | Method and system for acquiring high-resolution image |
CN107220936A (en) * | 2017-05-26 | 2017-09-29 | 首都师范大学 | A kind of image super-resolution reconstructing method and system |
CN110401968B (en) * | 2019-05-09 | 2020-05-22 | 温州曼昔维服饰有限公司 | Wireless communication system based on density analysis |
-
2013
- 2013-06-09 CN CN201310231478.0A patent/CN103440619B/en not_active Expired - Fee Related
Non-Patent Citations (4)
Title |
---|
传输型光学遥感器超模式斜采样新方法研究;周峰 等;《航天返回与遥感》;20050930;第26卷(第3期);43-46页 * |
基于最佳倒易晶胞的斜采样遥感图像复原方法;王静,夏德深;《宇航学报》;20111130;第32卷(第11期);2451-2456页 * |
空域自适应滤波方法及其在斜模式遥感图像复原中的应用;郑钰辉;《中国博士学位论文全文数据库 信息科技辑》;20100715(第7期);图6.8 * |
超分辨率图像重建及其应用;赵荣椿;《测控技术 增刊》;20071231;第26卷;3.4部分 * |
Also Published As
Publication number | Publication date |
---|---|
CN103440619A (en) | 2013-12-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103440619B (en) | A kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method | |
US10475232B2 (en) | Three-dimentional plane panorama creation through hough-based line detection | |
Apollonio et al. | A photogrammetry-based workflow for the accurate 3D construction and visualization of museums assets | |
KR101396370B1 (en) | Technologies including system, method and computer-readable storage medium to extract high resolution texture | |
Shan et al. | Simulation-based lidar super-resolution for ground vehicles | |
US5579456A (en) | Direct rendering of textured height fields | |
Li et al. | An improved building boundary extraction algorithm based on fusion of optical imagery and LIDAR data | |
Pepe et al. | Techniques, tools, platforms and algorithms in close range photogrammetry in building 3D model and 2D representation of objects and complex architectures | |
CN102346922B (en) | Space remote sensing load imaging geometric distortion three-dimensional visualization simulation method | |
JP2011501301A (en) | Geospatial modeling system and related methods using multiple sources of geographic information | |
JP2009545034A (en) | Geospatial modeling system and related method providing non-linear repair of voids in frequency domain data of a geospatial model | |
CN104809759A (en) | Large-area unstructured three-dimensional scene modeling method based on small unmanned helicopter | |
US20090309877A1 (en) | Soft shadow rendering | |
CN113592711B (en) | Three-dimensional reconstruction method, system, equipment and storage medium for non-uniform point cloud data | |
CN115861527A (en) | Method and device for constructing live-action three-dimensional model, electronic equipment and storage medium | |
CN102306386A (en) | Method for quickly constructing third dimension tree model from single tree image | |
CN113436559A (en) | Sand table dynamic landscape real-time display system and display method | |
CN111598803A (en) | Point cloud filtering method based on variable resolution voxel grid and sparse convolution | |
CN110658524B (en) | Cloud radar echo three-dimensional display method based on full-airspace multi-azimuth scanning mode | |
CN114511659B (en) | Volume rendering optimization method under digital earth terrain constraint | |
Jenny et al. | Real-time raster projection for web maps | |
Pfeifer | A subdivision algorithm for smooth 3D terrain models | |
Haltigin et al. | Comparison of ground-and aerial-based approaches for quantifying polygonal terrain network geometry on Earth and Mars via spatial point pattern analysis | |
CN104217412A (en) | Airborne super-resolution image reconstruction device and reconstruction method | |
Frisken et al. | Efficient estimation of 3D euclidean distance fields from 2D range images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180417 Termination date: 20210609 |
|
CF01 | Termination of patent right due to non-payment of annual fee |