#### DMCA

## Image registration methods: a survey (2003)

### Cached

### Download Links

- [adessowiki.fee.unicamp.br]
- [library.utia.cas.cz]
- [www.utia.cas.cz]
- [zoi.utia.cas.cz]
- [zoi.utia.cas.cz]
- [eecs.vanderbilt.edu]
- [www.andrew.cmu.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | IMAGE AND VISION COMPUTING |

Citations: | 761 - 10 self |

### Citations

4675 | A computational approach to edge detection
- Canny
- 1986
(Show Context)
Citation Context ... [114] or elongated anatomic structures [202] in medical imaging. Line correspondence is usually expressed by pairs of line ends or middle points. Standard edge detection methods, like Canny detector =-=[28]-=- or a detector based on the Laplacian of Gaussian [126], are employed for the line feature detection. The survey of existing edge detection method together with their evaluation can be found in [222].... |

3113 |
A method for registration of 3D shapes
- Besl, Mckay
- 1992
(Show Context)
Citation Context ...mploys also the pyramidal speed-up. Even that this overview does not intend to cover 3D registration methods, here the well-known Iterative Closest Point (ICP) algorithm, introduced by Besl and McKay =-=[18]-=- is mentioned, because it represents a key approach for registering 3D shapes (including free-form curves and surfaces). 4.2.2. Methods using invariant descriptors As an alternative to the methods exp... |

1272 |
Visual pattern recognition by moment invariants‖,
- Hu
- 1962
(Show Context)
Citation Context ...er developed this approach in Ref. [179]. A large group of methods uses moment-based invariants for description of closed-boundary region features. Considering the most often assumed deformations, Hu =-=[93]-=- introduced moment invariants to the similarity transform. Flusser and Suk derived the affine transform invariants [53] and used them successfully for registration of SPOT and Landsat images [54]. Hol... |

1151 | A fast algorithm for particle simulations
- Greengard, Rokhlin
- 1987
(Show Context)
Citation Context ...easonable accuracy. Flusser [50] proposed an adaptive approximation of the TPS ons992 square or triangular regions by simpler functions. Beatson and Newsam [15] adapted an earlier method by Greengard =-=[75]-=- for multipole expansion of radial functions. Powell [144] reduced the computational cost by the TPS tabulation. Barrodale et al. [13] paid attention to fast and robust calculation of the TPS coeffici... |

979 | A survey of image registration techniques
- Brown
- 1992
(Show Context)
Citation Context ...velopment and growing amount and diversity of obtained images invoked the research on automatic image registration. A comprehensive survey of image registration methods was published in 1992 by Brown =-=[26]-=-. The intention of our article is to cover relevant approaches introduced later and in this way map the current development of registration techniques. According to the database of the Institute of Sc... |

975 |
Principal warps: thin-plate splines and the decomposition of deformations
- Bookstein
- 1989
(Show Context)
Citation Context ... for the respective mathematical background). Although they had been used in mechanics and engineering for decades [84], they were introduced to image analysis community by Grimson [78] and Bookstein =-=[21]-=-. The TPS can be viewed as a very thin plate, which is fixed at the positions determined by the CPs in the reference image in the heights given by the x or y coordinates of the corresponding CPs in th... |

800 |
The Fourier Transform and Its Applications.
- Bracewell
- 1986
(Show Context)
Citation Context ...re preferred rather than the correlationlike methods. They exploit the Fourier representation ofsthe images in the frequency domain. The phase correlation method is based on the Fourier Shift Theorem =-=[23]-=- and was originally proposed for the registration of translated images. It computes the cross-power spectrum of the sensed and reference images and looks for the location of the peak in its inverse (s... |

552 |
A global optimisation method for robust affine registration of brain images. Med Image Anal.
- Jenkinson, Smith
- 2001
(Show Context)
Citation Context ...ker [174] tried to minimize a dissimilarity measure defined on point pairs by means of simulated annealing. Another optimization method, suited for multimodal data registration was introduced in Ref. =-=[97]-=- and its applicability was proved in combination with MI and correlation ratio. Again, the optimization methods can be speeded up by the pyramidal approach. There should be noted one thing with respec... |

369 | Cubic convolution interpolation for digital image processing
- Keys
- 1981
(Show Context)
Citation Context ...uch faster. The nearest neighbor function, the bilinear and bicubic functions (see Fig. 6), quadratic splines [42,191], cubic Bsplines [89], higher-order B-splines [108], Catmull–Rom cardinal splines =-=[100,184]-=-, Gaussians [7], and truncated sinc functions [182] belong to the most commonly used interpolants. Meijering et al. [129] investigated higherorder polynomial kernels (quintic and septic). However, the... |

347 |
Multiresolution elastic matching,
- Bajcsy, Kovacic
- 1989
(Show Context)
Citation Context ...rtions is not to use any parametric mapping functions, where the estimation of the geometric deformation is reduced to the search for the ‘best’ parameters. This idea were introduced by Bajcsy et al. =-=[10]-=- and is often called elastic registration. The images are viewed as pieces of a rubber sheet, on which external forces stretching the image and internal forces defined by stiffness or smoothness const... |

313 |
Hierarchical chamfer matching: A parametric edge matching algorithm.
- Borgefors
- 1988
(Show Context)
Citation Context ...row et al. [14] introduced the chamfer matching for image registration. Line features detected in images are matched by means of the minimalization of the generalized distance between them. Borgefors =-=[22]-=- proposed an B. Zitová, J. Flusser / Image and Vision Computing 21 (2003) 977–1000 improved version, where better measure of correspondence-the sequential distance transform together with the root mea... |

304 |
A fast operator for detection and precise location of distinct points, corners and centres of circular features.
- Förstner, Gülch
- 1987
(Show Context)
Citation Context ...er and Nagel [43] searched for the local extrema of the Gaussian curvature. However, corner detectors based on the second-order derivatives of the image function are sensitive to noise. Thus Förstner =-=[62]-=- developed a more robust, although time consuming, corner detector, which is based on the first-order derivatives only. The reputable Harris detector (also referred to as the Plessey detector) [135] i... |

229 |
Parametric correspondence and chamfer matching, IJCAI
- Barrow, Tenenbaum, et al.
- 1977
(Show Context)
Citation Context ...simultaneously with the feature correspondence. Local errors do not influence globally the registration process. The clustering technique was implemented, for example, in Refs. [30,72]. Barrow et al. =-=[14]-=- introduced the chamfer matching for image registration. Line features detected in images are matched by means of the minimalization of the generalized distance between them. Borgefors [22] proposed a... |

172 |
Cubic splines for image interpolation and digital filtering”,
- Hou, Andrews
- 1978
(Show Context)
Citation Context ...ace an m £ m 2D convolution by ðm þ 1Þ 1D convolutions which is much faster. The nearest neighbor function, the bilinear and bicubic functions (see Fig. 6), quadratic splines [42,191], cubic Bsplines =-=[89]-=-, higher-order B-splines [108], Catmull–Rom cardinal splines [100,184], Gaussians [7], and truncated sinc functions [182] belong to the most commonly used interpolants. Meijering et al. [129] investig... |

169 | Fast fluid registration of medical images,
- Bro-Nielsen, Gramkow
- 1996
(Show Context)
Citation Context ...e image is here modelled as a thick fluid that flows out to match the sensed image under the control of the derivative of a Gaussian sensor model. This approach is mainly used in medical applications =-=[25]-=-. The weakness of this approach is blurring introduced during the registration process. Lester and Arridge [110] proposed to use fluid model just for finding the correspondence of CPs and then process... |

155 |
Pattern recognition by affine moment invariants,
- Flusser, Suk
- 1993
(Show Context)
Citation Context ...ed-boundary region features. Considering the most often assumed deformations, Hu [93] introduced moment invariants to the similarity transform. Flusser and Suk derived the affine transform invariants =-=[53]-=- and used them successfully for registration of SPOT and Landsat images [54]. Holm [88] extracted closed boundary regions and proposed to represent them by their perimeter, area, compactness, moments,... |

150 |
Gray-level corner detection,
- Kitchen, Rosenfeld
- 1982
(Show Context)
Citation Context ...ation properties of the detectors. Corners are widely used as CPs mainly because of their invariance to imaging geometry and because they are well perceived by a human observer. Kitchen and Rosenfeld =-=[101]-=- proposed to exploit the second-order partial derivatives of the image function for corner detection. Dreschler and Nagel [43] searched for the local extrema of the Gaussian curvature. However, corner... |

117 |
Interpolation des fonctions de deux variables suivant le principe de la flexion des plaques minces,
- Duchon
- 1976
(Show Context)
Citation Context ...ainly locally. The most often used representatives of the radial basis functions are the thin-plate splines (TPS), where the radial terms have the form gðx; x iÞkx 2 x ik 2 lnðkx 2 x ikÞ (see Duchon =-=[44]-=- and Wahba [203] for the respective mathematical background). Although they had been used in mechanics and engineering for decades [84], they were introduced to image analysis community by Grimson [78... |

114 |
A class of algorithms for fast digital image registration,
- Barnea, Silverman
- 1972
(Show Context)
Citation Context ...ods deal with the images without attempting to detect salient objects. Windows of predefined size or even entire images are used for the correspondence estimation during the second registration step, =-=[4,12,145]-=-. The limitations of the area-based methods originate in their basic idea. Firstly, the rectangular window, which is most often used, suits the registration of images which locally differ only by a tr... |

105 |
Extension of phase correlation to subpixel registration
- Faroosh, Zerubia, et al.
- 2002
(Show Context)
Citation Context ...the original graylevel images. Extension of phase correlation to subpixel registration by means of the analytic expression of phase correlation on down sampled images was introduced by Foroosh et al. =-=[61]-=-. 4.1.3. Mutual information methods The mutual information (MI) methods are the last group of the area-based methods to be reviewed here. They have appeared recently and represent the leading techniqu... |

97 |
Medical Image Registration,
- Hajnal, Hill, et al.
- 2001
(Show Context)
Citation Context ...metimes next to the dissimilarity measure term the formula to be minimized contains as well so-called regularization or penalty term, which interconnects the transformation and data to be transformed =-=[82]-=-. These two terms together form the cost function (energy) associated with the registration and the aim of the optimization methods is to minimize it. In literature such methods can be referred to as ... |

93 |
Volumetric model and 3-D trajectory of a moving car derived from monocular TV-frame sequence of a street scene,
- Dreschler, Nagel
- 1981
(Show Context)
Citation Context ...cause they are well perceived by a human observer. Kitchen and Rosenfeld [101] proposed to exploit the second-order partial derivatives of the image function for corner detection. Dreschler and Nagel =-=[43]-=- searched for the local extrema of the Gaussian curvature. However, corner detectors based on the second-order derivatives of the image function are sensitive to noise. Thus Förstner [62] developed a ... |

92 | An algorithmic overview of surface registration techniques for medical imaging,”Medical Image Analysis
- Audette, Ferrie, et al.
- 2000
(Show Context)
Citation Context ... exhaustive review of the general-purpose image registration methods is in Ref. [26]. Registration techniques applied particularly in medical imaging are summarized in Refs. [86,111,123,195]. In Ref. =-=[9]-=- the surface based registration methods in medical imaging are reviewed. Volume-based registration is reviewed in Ref. [40]. The registration methods applied mainly in remote sensing are described and... |

91 | Image warping by radial basis functions : applications to facial expressions”, CVGIP:Graph Models Image Processing,
- Arad, Dyn, et al.
- 1994
(Show Context)
Citation Context ...nomials and multiquadrics when registering aerial images. The comparison of the TPS with the Gaussian radial basis functions and with multiquadrics considering their local properties was done in Ref. =-=[8]-=-. It was concluded that the TPS have favorable properties when used as mapping functions for image registration, while the other radial basis functions are more convenient for other applications such ... |

83 | Image registration based on boundary mapping,
- Davatzikos, Prince, et al.
- 1996
(Show Context)
Citation Context ...ch is often applied. The external forces can be derived from the local optimization of the similarity function which is defined by the intensity values or by the correspondence of boundary structures =-=[38]-=-, among others. In Ref. [140],no B. Zitová, J. Flusser / Image and Vision Computing 21 (2003) 977–1000 external forces were used and the prescribed displacements, derived from the correspondence of bo... |

79 |
An implementation of a computational theory of visual surface interpolation.
- Grimson
- 1983
(Show Context)
Citation Context ...44] and Wahba [203] for the respective mathematical background). Although they had been used in mechanics and engineering for decades [84], they were introduced to image analysis community by Grimson =-=[78]-=- and Bookstein [21]. The TPS can be viewed as a very thin plate, which is fixed at the positions determined by the CPs in the reference image in the heights given by the x or y coordinates of the corr... |

76 |
Interpolation using surface splines,”
- Harder, Desmarais
- 1972
(Show Context)
Citation Context ...ms have the form gðx; x iÞkx 2 x ik 2 lnðkx 2 x ikÞ (see Duchon [44] and Wahba [203] for the respective mathematical background). Although they had been used in mechanics and engineering for decades =-=[84]-=-, they were introduced to image analysis community by Grimson [78] and Bookstein [21]. The TPS can be viewed as a very thin plate, which is fixed at the positions determined by the CPs in the referenc... |

72 |
Rucklidge, "Comparing images using the Hausdorff distance
- Huttenlocher, Klanderman, et al.
- 1993
(Show Context)
Citation Context ...ges with certain characteristic (fixed-pattern noise), projection-based registration [27], working with accumulated image rows and columns, respectively, outperforms classical CC. Huttenlocher et al. =-=[95]-=- proposed a method working with other type of similarity measure. They registered binary images (the output of an edge detector) transformed by translation or translation plus rotation, by means of th... |

70 |
Predicting error in rigid-body point-based registration.
- Fitzpatrick, West, et al.
- 1998
(Show Context)
Citation Context ...ble. Otherwise, the exclusion of several CP’s may result in inaccurate estimation of mapping parameters. In medical applications, CP’s can be far from the region of interest. Thus, Fitzpatrick et al. =-=[47,49]-=- proposed to detect anatomical points within the region of interest and to use them as test points (the called them ‘target points’). The concept of TPE can be extended such that the distance between ... |

66 | Degraded image analysis: An invariant approach.
- Flusser, Suk
- 1998
(Show Context)
Citation Context ... transform was expected) using the circular moments of distribution of the line features orientation. They combined moments and the scale-space representation of the images. Recently, Flusser and Suk =-=[55]-=- introduced a new class of moment invariants that are invariant to image blurring and demonstrated their performance in registering SPOT and AVHRR satellite images. Bentoutou et al. [16] registered mu... |

62 |
Piecewise linear mapping functions for image registration, Pattern Recognition,
- Goshtasby
- 1986
(Show Context)
Citation Context ...e and weighted mean methods [69] gain the ability to register images locally by introducing the slight variation to the original least square method. The local methods called piecewise linear mapping =-=[67]-=- and piecewise cubic mapping [68], together with the Akima’s quintic approach [209], apply the combination of the CP-based image triangulation and of the collection of local mapping functions each val... |

61 |
Registration of translated and rotated images using finite Fourier transform,
- Castro, Morandi
- 1987
(Show Context)
Citation Context ...ependent noise and non-uniform, time varying illumination disturbances. The computational time savings are more significant if the images, which are to be registered, are large. De Castro and Morandi =-=[29]-=- introduced an extension of the phase correlation for additional rotation transform. B. Zitová, J. Flusser / Image and Vision Computing 21 (2003) 977–1000 983 Fig. 2. Area-based matching methods: regi... |

60 |
A physics-based coordinate transformation for 3-D image matching,”
- Davis, Khotanzad, et al.
- 1997
(Show Context)
Citation Context ...A linear combination of translated cubic B-splines was used for the registration of the echo planar images [103]. Another type of spline-based function, the elastic body spline (EBS), was proposed in =-=[39]-=-. It evolved from the description of the equilibrium displacements of homogeneous isotropic elastic material subjected to a load. Body tissues depicted in the image data to be registered often have pr... |

60 | Voxel similarity measures for 3-D serial MR brain image registration,”
- Holden, Hill, et al.
- 2000
(Show Context)
Citation Context ...sults then indicates (although does not guarantee) good registration accuracy. Different consistency check can be employed when a set of at least two sensed images is registered to the same reference =-=[34,63,87,215]-=-. The sensed images can be also registered among themselves using the same set of CP’s, which provides another set of mapping parameters. Usingstransitivity of mappings, we obtain for each sensed imag... |

55 |
Fast evaluation of radial basis functions,
- Beatson, Newsam
- 1992
(Show Context)
Citation Context ...ity of the TPS evaluation while preserving reasonable accuracy. Flusser [50] proposed an adaptive approximation of the TPS ons992 square or triangular regions by simpler functions. Beatson and Newsam =-=[15]-=- adapted an earlier method by Greengard [75] for multipole expansion of radial functions. Powell [144] reduced the computational cost by the TPS tabulation. Barrodale et al. [13] paid attention to fas... |

55 |
Registration of images with geometric distortions,”
- Goshtasby
- 1988
(Show Context)
Citation Context ...e viewed as a very thin plate, which is fixed at the positions determined by the CPs in the reference image in the heights given by the x or y coordinates of the corresponding CPs in the sensed image =-=[70]-=-. The TPS minimizes the quadratic variation functional of potential energy that reflects the amount of function variation and which should be small for a good mapping function. The type of registratio... |

53 |
A moment-based approach to registration of images with affine geometric distortion,
- Flusser, Suk
- 1994
(Show Context)
Citation Context ...tion changes are expected or multisensor analysis is demanded. Region features. The region-like features can be the projections of general high contrast closed-boundary regions of an appropriate size =-=[54,72]-=-, water reservoirs, and lakes [71,88], buildings [92], forests [165], urban areas [161] or shadows [24]. The general criterion of closedboundary regions is prevalent. The regions are often represented... |

52 | Registration techniques for multisensor remotely sensed imagery,
- Fonseca, Manjunath
- 1996
(Show Context)
Citation Context ...ased registration methods in medical imaging are reviewed. Volume-based registration is reviewed in Ref. [40]. The registration methods applied mainly in remote sensing are described and evaluated in =-=[59, 81,106]-=-. Big evaluation project of different registration methods was run in Vanderbilt university [206]. Registration methods can be categorized with respect to various criteria. The ones usually used are t... |

49 |
Image registration by local approximation methods,
- Goshtasby
- 1988
(Show Context)
Citation Context ... the local geometric distortion in mind. Several authors have shown the superiority of the local or at least locally sensitive registration methods above the global ones in such situations (Goshtasby =-=[69]-=-, Ehlers and Fogel [46], Wiemker [209], and Flusser [50], among others). The weighted least square and weighted mean methods [69] gain the ability to register images locally by introducing the slight ... |

47 |
A feature-based image registration algorithm using improved chain-code representation combined with invariant moments,
- Dai, Khorram
- 1999
(Show Context)
Citation Context ...d first two Hu’s moments as preselectors for matching of closed contours. The candidate matches were tested using the chain code representation of the contours. A similar method was described in Ref. =-=[35]-=-, where the moment invariants are used together with the chain codes. Sato and Cipolla [163] computed directly, without correspondence estimation, the parameters of the present geometric deformations ... |

47 | Coupling dense and landmark-based approaches for nonrigid registration.
- Hellier, Barillot
- 2003
(Show Context)
Citation Context ...thods for images with various contrast and sharpness is analyzed in Ref. [151]. Recently, registration methods using simultaneously both area-based and feature-based approaches have started to appear =-=[85]-=-. 4. Feature matching The detected features in the reference and sensed images can be matched by means of the image intensity values in their close neighborhoods, the feature spatial distribution, or ... |

46 | Radial basis functions with compact support for elastic registration of medical images,
- Fornefett, Rohr, et al.
- 2001
(Show Context)
Citation Context ...comparison to the Akima’s method is presented in Ref. [209]. The medical application of multiquadrics is shown in Ref. [117]. Wendland’s functions applied in medical image registration appear in Ref. =-=[60]-=-. These functions have very small global influence and even significant local deformations can be well registered by this approach. This property is advantageous for registering medical images, where ... |

43 | Quadratic interpolation for image resampling,
- Dodgson
- 1997
(Show Context)
Citation Context ...arability enables to replace an m £ m 2D convolution by ðm þ 1Þ 1D convolutions which is much faster. The nearest neighbor function, the bilinear and bicubic functions (see Fig. 6), quadratic splines =-=[42,191]-=-, cubic Bsplines [89], higher-order B-splines [108], Catmull–Rom cardinal splines [100,184], Gaussians [7], and truncated sinc functions [182] belong to the most commonly used interpolants. Meijering ... |

42 |
Spatial registration of multispectral and multitemporal digital imagery using Fast Fourier Transform,
- Anuta
- 1970
(Show Context)
Citation Context ...or by using the edge or vector correlation. Pratt [145] applied, prior to the registration, image filtering to improve the CC performance on noisy or highly correlated images. Van Wie [196] and Anuta =-=[6]-=- employed the edge-based correlation, which is computed on the edges extracted from the images rather than on the original images themselves. In this way, the method is less sensitive to intensity dif... |

42 | The distribution of target registration error in rigid-body point-based registration.
- Fitzpatrick, West
- 2001
(Show Context)
Citation Context ... acquisition (e.g. scanner dependent distortions and errors) and to the required accuracy of the registration (the analysis of errors990 for rigid-body point-based registration was introduced in Ref. =-=[48]-=-). In special situations when the geometric deformation is partially known, e.g. when there exists a model for the distortion caused by the acquisition device and/or the scene geometry, the pre-correc... |

37 | Fast Algorithm for Point Pattern Matching: Invariant to Translations, Rotations and Scale Changes.
- Chang
- 1997
(Show Context)
Citation Context ...ameters are thus found simultaneously with the feature correspondence. Local errors do not influence globally the registration process. The clustering technique was implemented, for example, in Refs. =-=[30,72]-=-. Barrow et al. [14] introduced the chamfer matching for image registration. Line features detected in images are matched by means of the minimalization of the generalized distance between them. Borge... |

34 |
An objective comparison of 3D image interpolation methods,
- Grevera, Udupa
- 1998
(Show Context)
Citation Context ... cost. Several survey papers on resampling techniques have been published in the last years. A detailed investigation and comparison of methods was carried out in Ref. [138] for 2D images and in Ref. =-=[76]-=- for 3D data. Thevenaz et al. [182] paid attention to the elimination of undesired interpolation artifacts. Lehman et al. [109] published a survey article covering main interpolation methods (various ... |

32 |
Point pattern matching using convex hull edges,
- Goshtasby, Stockman
- 1985
(Show Context)
Citation Context ...sor analysis is demanded. Region features. The region-like features can be the projections of general high contrast closed-boundary regions of an appropriate size [54,72], water reservoirs, and lakes =-=[71,88]-=-, buildings [92], forests [165], urban areas [161] or shadows [24]. The general criterion of closedboundary regions is prevalent. The regions are often represented by their centers of gravity, which a... |

31 |
Piecewise cubic mapping functions for image registration,
- Goshtasby
- 1987
(Show Context)
Citation Context ...gain the ability to register images locally by introducing the slight variation to the original least square method. The local methods called piecewise linear mapping [67] and piecewise cubic mapping =-=[68]-=-, together with the Akima’s quintic approach [209], apply the combination of the CP-based image triangulation and of the collection of local mapping functions each valid within one triangle. These app... |

31 |
Registration of Video to Geo-referenced Imagery,
- Sawhney, Asmuth, et al.
- 1998
(Show Context)
Citation Context ...ce between the images [184] and the MI maximization [187], respectively. Sharma and Pavel [166] used the multiresolution Laplacian pyramid for the infrared and radar images registration. Kumar et al. =-=[102]-=- combined different types of pyramids (Laplacian, Gaussian) with different similarity measures (CC, sum of squared differences) to register aerial video sequences. Non-linear min-max filters applied i... |

30 | Moment forms invariant to rotation and blur in arbitrary number of dimensions,
- Flusser, Boldys, et al.
- 2003
(Show Context)
Citation Context ...ted and blurred digital subtraction angiography images using these invariants. Flusser et al. further developed this approach in Ref. [56] by introducing the combined blurrotation invariants. In Ref. =-=[52]-=- they generalized the previous invariants to register 3D images. Invariant combination of the basic geometric properties of features can form geometrically oriented descriptors. Govindu et al. [74] re... |

30 |
Description and discrimination of planar shapes using shape matrices,
- Goshtasby
- 1985
(Show Context)
Citation Context ...imple and fast description by radial shape vector but the usage of this method is limited to star-shape regions only. A generalized shape description in a form of a binary matrix was proposed in Ref. =-=[65,180]-=-. In Ref. [72], the shape matrices were used for registration of rotated and scaled satellite images. In Ref. [112] a chain code representation of contours was proposed as the invariant description an... |

30 |
Template matching in rotated images.
- Goshtasby
- 1985
(Show Context)
Citation Context ... situations, when no precedent feature detection was done and the invariants are successively computed for the window sliding across the whole image [55]. For translated and rotated images, Goshtasby =-=[66]-=- proposed to calculate the moment invariants [93] from the circular-shaped windows and then to apply the CC criterion on the moment window representation. A similar idea was used earlier by Wong and H... |

28 |
A region-based approach to digital image registration with subpixel accuracy,‖
- Goshtasby, Stockman, et al.
- 1986
(Show Context)
Citation Context ...tion changes are expected or multisensor analysis is demanded. Region features. The region-like features can be the projections of general high contrast closed-boundary regions of an appropriate size =-=[54,72]-=-, water reservoirs, and lakes [71,88], buildings [92], forests [165], urban areas [161] or shadows [24]. The general criterion of closedboundary regions is prevalent. The regions are often represented... |

27 |
A rapid and automatic image registration algorithm with subpixel accuracy,”
- Althof, Wind
- 1997
(Show Context)
Citation Context ...ods deal with the images without attempting to detect salient objects. Windows of predefined size or even entire images are used for the correspondence estimation during the second registration step, =-=[4,12,145]-=-. The limitations of the area-based methods originate in their basic idea. Firstly, the rectangular window, which is most often used, suits the registration of images which locally differ only by a tr... |

27 |
Geometrical registration of images: the multiresolution approach,”
- Djamdji, Bijaoui, et al.
- 1993
(Show Context)
Citation Context ...cted the modulus maxima of LH and HL coefficients and looked for the maxima of the correlation coefficients, computed from LL coefficients in small surroundings of the detected maxima. Djamdji et al. =-=[41]-=- use just HH coefficients. Le Moigne [105] applied the Daubechies wavelet to register Landsat images and AVHRR data. They extracted LH and HL frequency coefficients and found the correspondence by mea... |

26 | Projection-based image registration in the presence of fixed-pattern noise,
- Cain, Hayat, et al.
- 2001
(Show Context)
Citation Context ...approach to several other algorithms developed for multimodal data can be found in Ref. [154]. In case of noisy images with certain characteristic (fixed-pattern noise), projection-based registration =-=[27]-=-, working with accumulated image rows and columns, respectively, outperforms classical CC. Huttenlocher et al. [95] proposed a method working with other type of similarity measure. They registered bin... |

26 |
Accurate registration of serial 3D MR brain images and its application to visualizing change in neurodegenerative disorders,
- Freeborough, Woods, et al.
- 1996
(Show Context)
Citation Context ...sults then indicates (although does not guarantee) good registration accuracy. Different consistency check can be employed when a set of at least two sensed images is registered to the same reference =-=[34,63,87,215]-=-. The sensed images can be also registered among themselves using the same set of CP’s, which provides another set of mapping parameters. Usingstransitivity of mappings, we obtain for each sensed imag... |

25 |
Performance evaluation of scene registration and stereo matching for cartographic feature extraction,
- Hsieh, McKeown, et al.
- 1992
(Show Context)
Citation Context ...anded. Region features. The region-like features can be the projections of general high contrast closed-boundary regions of an appropriate size [54,72], water reservoirs, and lakes [71,88], buildings =-=[92]-=-, forests [165], urban areas [161] or shadows [24]. The general criterion of closedboundary regions is prevalent. The regions are often represented by their centers of gravity, which are invariant wit... |

24 | Combined invariants to linear filtering and rotation,
- Flusser, Zitova
- 1999
(Show Context)
Citation Context ... (see Fig. 4). The description Fig. 4. Feature-based method using invariant descriptors: in these two satellite images, control points (corners) were matched using invariants based on complex moments =-=[56]-=-. The numbers identify corresponding CP’s. The bottom image shows the registration result.sshould fulfill several conditions. The most important ones are invariance (the descriptions of the correspond... |

23 |
An invariant approach for image registration in digital subtraction angiography.
- Bentoutou, Taleb, et al.
- 2002
(Show Context)
Citation Context ...usser and Suk [55] introduced a new class of moment invariants that are invariant to image blurring and demonstrated their performance in registering SPOT and AVHRR satellite images. Bentoutou et al. =-=[16]-=- registered mutually shifted and blurred digital subtraction angiography images using these invariants. Flusser et al. further developed this approach in Ref. [56] by introducing the combined blurrota... |

23 | Automatic Registration of Satellite Images.
- Fonseca, Costa
- 1997
(Show Context)
Citation Context ...oil and gas pads [190], high variance points [45], local curvature discontinuities detected using the Gabor wavelets [125, 219], inflection points of curves [3,11], local extrema of wavelet transform =-=[58,90]-=-, the most distinctive points with respect to a specified measure of similarity [115], and corners [20,92,204]. The core algorithms of feature detectors in most cases follow the definitions of the ‘po... |

20 | Alignment using distributions of local geometric properties,
- Govindu, Shekhar
- 1999
(Show Context)
Citation Context ...mutual image rotation can be estimated. They derived a similar type of descriptors for the affine transform, too. A detailed study of the practical aspects of the proposed method can be found in Ref. =-=[73]-=-. Wang and Chen [205] computed the histogram of line-length ratios and the histogram of angle differences of any two line segments in the reference and sensed images. They assumed the similarity trans... |

19 |
Registering coronal histological 2-D sections of a rat brain with coronal sections of a 3-D brain atlas using geometric curve invariants and B-spline representation,
- Ali, Cohen
- 1998
(Show Context)
Citation Context ...ings [79,161], centroids of water regions, oil and gas pads [190], high variance points [45], local curvature discontinuities detected using the Gabor wavelets [125, 219], inflection points of curves =-=[3,11]-=-, local extrema of wavelet transform [58,90], the most distinctive points with respect to a specified measure of similarity [115], and corners [20,92,204]. The core algorithms of feature detectors in ... |

18 |
Robust point correspondence applied to two- and three-dimensional image registration,
- Guest, Berry, et al.
- 2001
(Show Context)
Citation Context ...ood coefficients [51], which can better handle questionable situations, can be an appropriate solution. Guest et al. proposed to select features according to the reliability of their possible matches =-=[80]-=-. The simplest feature description is the image intensity function itself, limited to the close neighborhood of the feature [1,107]. To estimate the feature correspondence, authors computed the CC on ... |

17 |
Using selective correlation coefficient for robust image registration,
- Kaneko, Satoh, et al.
- 2003
(Show Context)
Citation Context ...t with the increase of the transformation complexity. In case the images/objects to be registered are partially occluded the extended CC method based on increment sign correlation [98] can be applied =-=[99]-=-. Similar to the CC methods is the sequential similarity detection algorithm (SSDA) [12]. It uses the sequential search approach and a computationally simpler distance B. Zitová, J. Flusser / Image an... |

16 |
Warping digital images using thin plate splines.
- Barrodale, Skea, et al.
- 1993
(Show Context)
Citation Context ...s. Beatson and Newsam [15] adapted an earlier method by Greengard [75] for multipole expansion of radial functions. Powell [144] reduced the computational cost by the TPS tabulation. Barrodale et al. =-=[13]-=- paid attention to fast and robust calculation of the TPS coefficients. Numerous papers have dealt with the comparison of the performance of thin-plate splines and other mapping functions. In Refs. [5... |

16 |
Automated assembling of images: Image montage preparation,
- Dani, Chaudhuri
- 1995
(Show Context)
Citation Context ...xploits a summing pyramid (the pixel value at a coarser level corresponds to the summation of the pixel values on the previous level), a median pyramid, and an averaging pyramid was proposed in Refs. =-=[37,208,219]-=-, respectively. Wong and Hall [214] combined the SSDA method with pyramidal speed-up. Wang and Chen [205] extracted features (centroids of closed boundary regions) at every resolution level and found ... |

16 | Image registration using a new edge-based approach,
- Hsieh, Liao, et al.
- 1997
(Show Context)
Citation Context ...he angle with the x axis and the distance from the centroid. All these methods skip the step of finding the feature correspondence and establish directly the mapping function parameters. Hsieh et al. =-=[91]-=- used the angle histogram computed on line feature points for the compensation of rotation difference. After the removal of the rotation difference, the feature point correspondence is found by means ... |

15 |
Affine correlation.
- Berthilsson
- 1998
(Show Context)
Citation Context ...assumed geometric transformation of the sensed image window [83] and are able to handle even more complicated geometric deformations than the translation-usually the similarity transform. Berthilsson =-=[17]-=- tried to register in this manner even affinely deformed images and Simper [170] proposed to use a divide and conquer system and the CC technique for registering images differing by perspective change... |

14 | Non-rigid registration by geometryconstrained diffusion,
- Andersen, Nielsen
- 2001
(Show Context)
Citation Context ...t contours and other features as membranes, setting the geometrical constraints. Three variations of this approach are described in Ref. [189]. Different solution was proposed by Andersen and Nielsen =-=[5]-=-. Vemuri et al. [199] introduced elastic registration method, based on evolution of level sets, moving along their respective normals. Finally, the optical flow approach was originally motivated by es... |

11 |
A new approach to the interpolation of sampled data,
- Appledorn
- 1996
(Show Context)
Citation Context ...st neighbor function, the bilinear and bicubic functions (see Fig. 6), quadratic splines [42,191], cubic Bsplines [89], higher-order B-splines [108], Catmull–Rom cardinal splines [100,184], Gaussians =-=[7]-=-, and truncated sinc functions [182] belong to the most commonly used interpolants. Meijering et al. [129] investigated higherorder polynomial kernels (quintic and septic). However, their experiments ... |

11 |
Defrise M, Deconinck F. Symmetric Phase-Only Matched Filtering Image Registration and Recognition
- Chen
- 1994
(Show Context)
Citation Context ...image scale is present too, the images can be registered using the combination of polar-log mapping of the spectral magnitude (which corresponds to Fourier–Mellin transform) and the phase correlation =-=[31, 150]-=- or cepstrum filter [107]. The applications of the extended algorithm in remote sensing (SPOT images) and medical imaging (MR images) are described in Ref. [31]. The testing of the accuracy of the met... |

11 |
Object matching by means of matching likelihood coefficients,
- Flusser
- 1995
(Show Context)
Citation Context ... matching feature pairs in the space of feature descriptors, the minimum distance rule with thresholding is usually applied. If a more robust algorithm is needed, the matching likelihood coefficients =-=[51]-=-, which can better handle questionable situations, can be an appropriate solution. Guest et al. proposed to select features according to the reliability of their possible matches [80]. The simplest fe... |

11 | Robust image registration by increment sign correlation,
- Kaneko, Murase, et al.
- 2002
(Show Context)
Citation Context ...ever, grows very fast with the increase of the transformation complexity. In case the images/objects to be registered are partially occluded the extended CC method based on increment sign correlation =-=[98]-=- can be applied [99]. Similar to the CC methods is the sequential similarity detection algorithm (SSDA) [12]. It uses the sequential search approach and a computationally simpler distance B. Zitová, J... |

10 |
The computation of optical flow,
- Beuchemin, Barron
- 1995
(Show Context)
Citation Context ...gistration method, based on evolution of level sets, moving along their respective normals. Finally, the optical flow approach was originally motivated by estimation of relative motion between images =-=[19]-=-. The class of optical flow registration covers very large number of methods and is beyond the scope of this survey. 6. Image resampling and transformation The mapping functions constructed during the... |

10 |
Invariance of stereo images via theory of complex moments,
- Bhattacharya, Sinha
- 1997
(Show Context)
Citation Context ...r wavelets [125, 219], inflection points of curves [3,11], local extrema of wavelet transform [58,90], the most distinctive points with respect to a specified measure of similarity [115], and corners =-=[20,92,204]-=-. The core algorithms of feature detectors in most cases follow the definitions of the ‘point’ as line intersection, centroid of closed-boundary region or local modulus maxima of the wavelet transform... |

10 | A knowledge based approach to automatic image registration,
- Growe, Tonjes
- 1997
(Show Context)
Citation Context ... detection in multimodal medical images is presented by Maintz et al. [121,122]. Point features. The point features group consists of methods working with line intersections [175,198], road crossings =-=[79,161]-=-, centroids of water regions, oil and gas pads [190], high variance points [45], local curvature discontinuities detected using the Gabor wavelets [125, 219], inflection points of curves [3,11], local... |

9 |
Point pattern matching algorithm invariant to geometrical transformation and distortion,
- Cheng
- 1996
(Show Context)
Citation Context ... algorithm by integrating the MergeSort concept. Their method works with shifted and rotated images. Relaxation based method even for similarity transformed images was proposed, for example, by Cheng =-=[32]-=-, Ogawa [136] and Li [113]. Different relaxation methods are compared in Ref. [147]. Another solution to the CLP problem and consequently to the image registration is backtracking, where consistent la... |

8 |
Registration of ocular fundus images,
- Cideciyan
- 1995
(Show Context)
Citation Context ...mages) are described in Ref. [31]. The testing of the accuracy of the method in simulated conditions (registration of deformed and noisy images of ocular fundus) was performed with satisfying results =-=[34]-=-. Affinely distorted images were registered by means of phase correlation and log-polar mapping in Ref. [210]. Application of phase correlation in 3D is described in Ref. [119]. Another application ex... |

8 |
A survey of new techniques for image registration and mapping,
- Ghaffary, Sawchuk
- 1983
(Show Context)
Citation Context ...Because of its importance in various application areas as well as because of its complicated nature, image registration has been the topic of much recent research. The historically first survey paper =-=[64]-=- covers mainly the methods based on image correlation. Probably the most exhaustive review of the general-purpose image registration methods is in Ref. [26]. Registration techniques applied particular... |

8 |
An automated method for registration of satellite remote sensing images,
- Hanaizumi, Fujimura
- 1993
(Show Context)
Citation Context ...rotation and scaling are present. There are generalized versions of CC for geometrically more deformed images. They compute the CC for each assumed geometric transformation of the sensed image window =-=[83]-=- and are able to handle even more complicated geometric deformations than the translation-usually the similarity transform. Berthilsson [17] tried to register in this manner even affinely deformed ima... |

7 |
Point pattern matching using centroid bounding,
- Griffin, Alexopoulos
- 1989
(Show Context)
Citation Context ... the histogram of line-length ratios and the histogram of angle differences of any two line segments in the reference and sensed images. They assumed the similarity transform. Griffin and Alexopoulos =-=[77]-=- used the ratio of the smallest enclosing circle radii, the difference of the locations of centroids of gravity, and the sorting of the neighbors lexicographically according to the angle with the x ax... |

7 |
Results of test on image matching
- Gülch
- 1991
(Show Context)
Citation Context ...ased registration methods in medical imaging are reviewed. Volume-based registration is reviewed in Ref. [40]. The registration methods applied mainly in remote sensing are described and evaluated in =-=[59, 81,106]-=-. Big evaluation project of different registration methods was run in Vanderbilt university [206]. Registration methods can be categorized with respect to various criteria. The ones usually used are t... |

6 |
Virtual circles: a new set of features for fast image registration,
- Alhichri, Kamel
- 2003
(Show Context)
Citation Context ...parameters. They claimed the subpixel accuracy of registration could be achieved. Recently, selection of region features invariant with respect to change of scale caught attention. Alhichri and Kamel =-=[2]-=- proposed the idea of virtual circles, using distance transform. Affinely invariant neighborhoods were described in [194], based on Harris corner detector [135] and edges (curved or straight) going th... |

6 |
Point landmarks for registration of
- Banerjee, Mukherjee, et al.
- 1995
(Show Context)
Citation Context ...ings [79,161], centroids of water regions, oil and gas pads [190], high variance points [45], local curvature discontinuities detected using the Gabor wavelets [125, 219], inflection points of curves =-=[3,11]-=-, local extrema of wavelet transform [58,90], the most distinctive points with respect to a specified measure of similarity [115], and corners [20,92,204]. The core algorithms of feature detectors in ... |

6 |
Towards automatic rectification of satellite images using feature based matching,
- Holm
- 1991
(Show Context)
Citation Context ...sor analysis is demanded. Region features. The region-like features can be the projections of general high contrast closed-boundary regions of an appropriate size [54,72], water reservoirs, and lakes =-=[71,88]-=-, buildings [92], forests [165], urban areas [161] or shadows [24]. The general criterion of closedboundary regions is prevalent. The regions are often represented by their centers of gravity, which a... |

5 |
Development of a feature-based approach to automated image registration for multitemporal and multisensor remotely sensed imagery,
- Dai, Khorram
- 1997
(Show Context)
Citation Context ...gions based on homogeneity of image intensities was presented by Matas et al. [127]. Line features. The line features can be the representations of general line segments [92,132,205], object contours =-=[36, 74,112]-=-, coastal lines, [124,168], roads [114] or elongated anatomic structures [202] in medical imaging. Line correspondence is usually expressed by pairs of line ends or middle points. Standard edge detect... |

5 |
Image rectification with radial basis functions: Application to RS/GIS data integration,
- Fogel
- 1996
(Show Context)
Citation Context ...3] paid attention to fast and robust calculation of the TPS coefficients. Numerous papers have dealt with the comparison of the performance of thin-plate splines and other mapping functions. In Refs. =-=[57,70,209]-=- they are compared with polynomials and multiquadrics when registering aerial images. The comparison of the TPS with the Gaussian radial basis functions and with multiquadrics considering their local ... |

5 |
A fast algorithm for image registration without predetermining correspondence,
- Hsieh, Liao, et al.
- 1996
(Show Context)
Citation Context ...oil and gas pads [190], high variance points [45], local curvature discontinuities detected using the Gabor wavelets [125, 219], inflection points of curves [3,11], local extrema of wavelet transform =-=[58,90]-=-, the most distinctive points with respect to a specified measure of similarity [115], and corners [20,92,204]. The core algorithms of feature detectors in most cases follow the definitions of the ‘po... |

5 |
A model-based approach for geometrical correction of optical satellite images,
- Huseby, Halck, et al.
- 1999
(Show Context)
Citation Context ... there exists a model for the distortion caused by the acquisition device and/or the scene geometry, the pre-correction based on the inverse of the deformation can be performed (for example, in Refs. =-=[94,168,181]-=-, the authors model the Earth’s shape and rotation, the satellite orbit and the scanning geometry of the sensor). Models of mapping functions can be divided into two broad categories according to the ... |

4 |
Matching and registration method for remote sensing images,
- Abdelsayed, Ionescu, et al.
- 1995
(Show Context)
Citation Context ... select features according to the reliability of their possible matches [80]. The simplest feature description is the image intensity function itself, limited to the close neighborhood of the feature =-=[1,107]-=-. To estimate the feature correspondence, authors computed the CC on these neighborhoods. Other types of similarity measures can be used, too. Zheng and Chellapa make use of the correlation coefficien... |

3 |
Automatic selection of control points from shadow structures,
- Brivio, Ventura, et al.
- 1992
(Show Context)
Citation Context ...an be the projections of general high contrast closed-boundary regions of an appropriate size [54,72], water reservoirs, and lakes [71,88], buildings [92], forests [165], urban areas [161] or shadows =-=[24]-=-. The general criterion of closedboundary regions is prevalent. The regions are often represented by their centers of gravity, which are invariant with respect to rotation, scaling, and skewing and st... |

2 |
High Precision Geometric Correction of Airborne Remote Sensing Revisited: The Multiquadric Interpolation
- Ehlers, Fogel
- 1994
(Show Context)
Citation Context ...stortion in mind. Several authors have shown the superiority of the local or at least locally sensitive registration methods above the global ones in such situations (Goshtasby [69], Ehlers and Fogel =-=[46]-=-, Wiemker [209], and Flusser [50], among others). The weighted least square and weighted mean methods [69] gain the ability to register images locally by introducing the slight variation to the origin... |

2 |
An adaptive method for image registration, Pattern Recognition 25
- Flusser
- 1992
(Show Context)
Citation Context ... have shown the superiority of the local or at least locally sensitive registration methods above the global ones in such situations (Goshtasby [69], Ehlers and Fogel [46], Wiemker [209], and Flusser =-=[50]-=-, among others). The weighted least square and weighted mean methods [69] gain the ability to register images locally by introducing the slight variation to the original least square method. The local... |

2 |
Using geometric properties for correspondence-less image alignement
- Govindu, Shekhar, et al.
- 1998
(Show Context)
Citation Context ...gions based on homogeneity of image intensities was presented by Matas et al. [127]. Line features. The line features can be the representations of general line segments [92,132,205], object contours =-=[36, 74,112]-=-, coastal lines, [124,168], roads [114] or elongated anatomic structures [202] in medical imaging. Line correspondence is usually expressed by pairs of line ends or middle points. Standard edge detect... |

1 |
Image registration by matching relational structures, Pattern Recognition 17
- Cheng, Huang
- 1984
(Show Context)
Citation Context ...tation distortions in the images, but it is computationally demanding. Medioni and Nevatia [128] used line features and their descriptors (coordinates, orientation, average contrast). Cheng and Huang =-=[33]-=- proposed a star-based registration which considers individual feature points along with all links to their neighbors. Ton and Jain [190] speeded up the algorithm by integrating the MergeSort concept.... |

1 |
Volume image registration by template matching, Image and Vision Computing 19
- Ding, Goshtasby, et al.
- 2001
(Show Context)
Citation Context ...icularly in medical imaging are summarized in Refs. [86,111,123,195]. In Ref. [9] the surface based registration methods in medical imaging are reviewed. Volume-based registration is reviewed in Ref. =-=[40]-=-. The registration methods applied mainly in remote sensing are described and evaluated in [59, 81,106]. Big evaluation project of different registration methods was run in Vanderbilt university [206]... |

1 |
Region-based matching for image registration in remote sensing databases
- Ehlers
- 1991
(Show Context)
Citation Context ...int features. The point features group consists of methods working with line intersections [175,198], road crossings [79,161], centroids of water regions, oil and gas pads [190], high variance points =-=[45]-=-, local curvature discontinuities detected using the Gabor wavelets [125, 219], inflection points of curves [3,11], local extrema of wavelet transform [58,90], the most distinctive points with respect... |

1 |
Detection failure, assessing success, in: J.V
- Fitzpatrik
(Show Context)
Citation Context ...ble. Otherwise, the exclusion of several CP’s may result in inaccurate estimation of mapping parameters. In medical applications, CP’s can be far from the region of interest. Thus, Fitzpatrick et al. =-=[47,49]-=- proposed to detect anatomical points within the region of interest and to use them as test points (the called them ‘target points’). The concept of TPE can be extended such that the distance between ... |