Invariant object recognition with discriminant features based on local fast-Fourier Mellin transform

Götze, Nicolai; Drüe, Siegbert; Hartmann, Georg

doi:10.1109/icpr.2000.905607

Cited by 15 publications

(9 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where ν ∈ R is the "log-radial frequency," m the polar frequency ‡ ‡ and I the Fourier-Mellin transform [17][18][19][20][21] (FMT) of I:…”

Section: Computational Aspects -The Fourier-mellin Transformmentioning

confidence: 99%

Secure graphical data storage by full spectrum image coding

Oltmans¹

2006

SPIE Proceedings

View full text Add to dashboard Cite

Full Spectrum coding is a method of providing a picture with a machine readable code image. The transparency and robustness of the code can be influenced by the embedding parameters, according to the intended application (e.g. robust or fragile marking). The code image contains hidden information, identifying the document in some way. In principle, it can be any binary pattern or grey-scale picture. Alternatively, it may be desirable to encode (binary) data, e.g. ID numbers or (biometric) templates. Such data should be transformed into a code image suitable for the proposed kind of embedding. In the presented implementation, the code image contains a binary block structure, similar to a 2-dimensional (bar) code. Even in the presence of noise, this code can be extracted from the magnitude spectrum of the captured image by iterated down-sampling. Full Spectrum is well-suited for application to printed documents. It survives graphical processes (halftone screening, printing, digitizing) and the reconstructed code image is invariant to shifting and robust to cropping. A technique is presented to perform registration for rotation and scaling of the captured image with respect to the original. The approach is to embed a marking signal based on perfect correlation sequences in the radial-polar representation of the code image. In this domain, the rotation angle and (under certain conditions) the scale factor, can be determined by linear cross-correlation. Additionally, techniques are proposed to obtain resistance to more general distortions.

show abstract

“…where ν ∈ R is the "log-radial frequency," m the polar frequency ‡ ‡ and I the Fourier-Mellin transform [17][18][19][20][21] (FMT) of I:…”

Section: Computational Aspects -The Fourier-mellin Transformmentioning

confidence: 99%

Secure graphical data storage by full spectrum image coding

Oltmans¹

2006

SPIE Proceedings

View full text Add to dashboard Cite

show abstract

“…15 Another approach entails transforming the image by an invariant mapping such as the Fourier-Mellin transform. 16,17 Scale-invariant segmentation can also be obtained by training a classifier based on features that eliminate changes in scale. Such scale-invariant features include wavelets, 5 features from the linear scale space 18 and different statistical moments.…”

Section: Introductionmentioning

confidence: 99%

Scale‐invariant segmentation of dynamic contrast‐enhanced perfusion MR images with inherent scale selection

Janssen¹,

Egmont-Petersen

Hendriks³

et al. 2002

J. Visual. Comput. Animat.

View full text Add to dashboard Cite

Selection of the best set of scales is problematic when developing signal-driven approaches for pixel-based image segmentation. Often, different possibly conflicting criteria need to be fulfilled in order to obtain the best trade-off between uncertainty (variance) and location accuracy. The optimal set of scales depends on several factors: the noise level present in the image material, the prior distribution of the different types of segments, the class-conditional distributions associated with each type of segment as well as the actual size of the (connected) segments. We analyse, theoretically and through experiments, the possibility of using the overall and class-conditional error rates as criteria for selecting the optimal sampling of the linear and morphological scale spaces. It is shown that the overall error rate is optimized by taking the prior class distribution in the image material into account. However, a uniform (ignorant) prior distribution ensures constant class-conditional error rates. Consequently, we advocate for a uniform prior class distribution when an uncommitted, scale-invariant segmentation approach is desired. Experiments with a neural net classifier developed for segmentation of dynamic magnetic resonance (MR) images, acquired with a paramagnetic tracer, support the theoretical results. Furthermore, the experiments show that the addition of spatial features to the classifier, extracted from the linear or morphological scale spaces, improves the segmentation result compared to a signal-driven approach based solely on the dynamic MR signal. The segmentation results obtained from the two types of features are compared using two novel quality measures that characterize spatial properties of labelled images.

show abstract

“…The Zernike moments were applied to image recognition by transforming the image pixel values [36]. On the other hands, invariant recognition algorithm was proposed based on the Fourier -Mellin transform that is a Fourier transformation on a polar coordinate [38]. Because both of the algorithms were defined on the polar coordinate, rotation invariance is achieved and they are applied only to image retrieval or recognition.…”

Section: Modified Zernike Moments For Rotation-invariancementioning

confidence: 99%