In this paper a rotation, scale and translation (RST) invariant image descriptor based on 1D signatures is presented. The position invariant is obtained using the amplitude spectrum of the Fourier transform of the image. That spectrum is introduced in the analytical Fourier-Mellin transform (AFMT) to obtain the scale invariance. From the normalized AFMT amplitude spectrum two 1D signatures are constructed. To build a 1D circular signature, circular path binary masks are used to filter the spectrum image. On the other hand, ray path binary filters are utilized in the construction of the 1D ray signature. These 1D signatures are RST invariant image descriptors. The Latin alphabet letters in Arial font style were used to test the descriptor efficiency. According with the statistical analysis of bootstrap with a constant replacement B = 1000 and normal distribution, the descriptor has a confidence level at least of 95%.
In this work is presented a pattern recognition image descriptor invariant to rotation, scale and translation (RST), which classify images using the Z-Fisher transform. A binary rings mask is generated using the Fourier transform. The normalized analytic Fourier-Mellin amplitude spectrum is filtered with that mask to build 1D signature. The signatures comparison of the problem image and the target are done by the Pearson correlation coefficient (PCC). In general, those PCC values do not satisfy a normal distribution, hence the Fisher’s Z distribution is employed to determine the confidence level of the RST invariant descriptor. The descriptor presents a confidence level of 95%.
The effects of illumination variations in digital images are a trend topic of the pattern recognition field. The luminance information of the objects help to classify them, however the environment illumination could cause a lot of problem if the system is not illumination invariant. Some applications of this topic include image and video quality, biometrics classification, etc. In this work an illumination analysis for a digital system invariant to position and rotation based on Fourier transform, Bessel masks, one-dimensional signatures and linear correlations are presented. The digital system was tested using a reference database of 21 fossil diatoms images of gray-scale and 307 x 307 pixels. The digital system has shown an excellent performance in the classification of 60,480 problem images which have different non-homogeneous illumination.
Digital systems of invariant non-linear correlation to position and scale based on adaptive binary mask of concentric
rings and unidimensional signatures are useful tool in pattern recognition. With the modulus of the Fourier transform of
the image we obtain the invariance to translation. Using the Scale transformation and adaptive binary ring masks the
scale invariant is calculated. The discrimination between objects is done by non-linear correlation of the unidimensional
signatures assigned to the problem image and the target. In addition, working with unidimensional signatures reduce the
computational time considerably, achieving a step toward the ultimate goal, which is developing a simple digital system
that accomplishes recognition in real time at low cost.
A nonlinear correlation digital algorithm invariant to position, rotation and scale using a binary mask is presented. In
order to analyze this new identification digital system binary and gray images are used. The problem images had a
±30% of maximum scale variation with respect to the target. Some composite filters had a very good performance in
this range. The rotation goes from 0° to 359°. Concentric binary rings masks were elaborated, from the Fourier
transform, using the real or the imaginary part. The signatures of the problem image and the target were obtained from
the ring mask. The objective is identifying a specific target no matter the position, rotation or scale presented in the
problem image. A statistical analysis was done to know the mean correlation confidence level. In this work, a new, fast
and functional position, scale and rotation invariance pattern recognition digital system was obtained.
A new rotational invariance computational filter is presented. The filter was applied to a problem image, in this case, an image of 256 by 256 pixels of black background with a centered white Arial letter. The complete alphabet is represented in those images. The image is rotated one degree by one degree until complete 360 degrees; hence, for each alphabet letter we are generating 360 images. To achieve the rotational invariance, first of all, a translational invariance is applied and then a 256 by 256 binary mask of concentric circular rings of three pixels of thickness and separation is used. The sum of the information in the circular rings represents the signature of the image. The average of the signature of the 360 images of a selected letter is the filter used to compute the phase correlation with all alphabet letter and their rotated images. The confidence level is
calculated by the mean value with two standard errors (2SE) of those 360 correlation values for each letter. The confidence level shows that this system works efficiently on the discrimination between letters.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.