Abstract: General demand for the cultural and weeds all types of simple, precise methods of identifi cation. One potential tool that allows an objective kind of hope assing tickets to the subject, color composition of the test. The previous occurrenced SFD maps in color prevalence weight, considered as the dominant information is not broadcast. In practice, more adaptable, more complex structure of color information can be obtained if the RGB color space points suitably chosen probability distribution functions that assign values. The resulting four-dimensional quasi-continuous set of images is more faithful to carry information, which we show to species separation test. The main research task is to fi nd the best possible distribution functions HPC technology, which is based on the results of standard IT environments can be used
Keywords: SFD, Color information, HPC technology, Quasi-continuous
1. Introduction
Berke et al. (2005) introduced the defi nition of spectral fractal dimension and certain measuring methods for its determination (Berke et al. 2006, 2007). Hegedűs et al. (2007) studied the mathematical properties of this concept, and its invariance to geometrical transformation of the analyzed object. Csák et. al. (2008) attempted to apply SFD in characterization of potato breeds. The present article examines the distribution and applicability of spectral fractal dimension, which should count as a characterization principle, if it is generated from images stored by a lossy an a lossless compression format. The goal is in the continues to be that - now more faithful to reality - good color information in order to characterize the structure of plant varieties, as it has happened with the SFD.
2. Material and method The applied special SFD defi nition
A 24-bit digital raster pixels R, G, B making in coordinate system numerical values according to the fi rst fi eld in the eighth of 256 is obtained in frames (one vertex at the origin) from your body. The color of the pixel in the cube to be the same place.
The i-th iteration step of split up this cube edge length of disjoint sub-cubes 2i. In our report to the i-th iteration, the number of candidate sub-cubes, ni is the number of these are pieces that contain the color test image. Then, the digital image assigned to review the spectral fractal dimension (SFD) value of (a) defi nes the term.
(1) The extended phase-Spectral Fractal Dimension (ESFD) is a concept
At the starting point it is, assume that only a certain probability to say that the image of a digital image of the actual measured reproducibly position is the color cube. Value of the individual measurements, is expected to be considered, but we should assume that scatters around the point as well. Three-dimensional normal distribution using discrete positions in the surrounding will be (2) given by the positive employment probability.
(2)
The (rk, gk, bk) of the measured pixel position, (rj, gj, bj) is the scattering range (Tk) is a j-th point, and C is a member of the discrete approximation of the correction.
A, B and C are such that the range of Tk (rk, gk, bk) fi les belonging Fri probability (3) is met.
(3)
While runs through the pixels indices, according to the probability values increase in the color cube (rj, gj, bj)-equivalent points assigned (initially zero) values, which are the color cubes matches dimensional random collector array (M) is stored .
Then the tested digital picture of the value assigned to ESFD (4) defi nes the term, where qi is the i-th iteration level, the number of squares, cubes that rain reaches the sum of the elements M 1, we (1)-compliant content.
(4) Obviously, if there is no scattering, then (4) equal (1) value.
Required computational
The HSFD calculation is more complex and requires a completely different algorithm than the SFD. While is the SFD only one parameter of the image, than the HSFD calculations necessary to specify the range of scattering by A and B are setting up with. Problem of the increased operational storage requirements and the large array of additional calculations involving mean. The associated growth factor scatter range of the pixel size was obviously a lower limit of the fi rst multiplier The rare set of pixels and higher standard deviation, the greater the actual odd. The computer needs to function in both the upper limit of O (n2logn), but in practice it is much less because of the resulting cycle exit.
HSFD good separation ability to adjust shall be seised in optimal values of A and B, and the parameter range to run the multiple records need to multiply the calculated.
Separating capability
Let f and g be functions, which are of the two sample line function of expected values of the range are interpreted as follows: f (x) report a lower expected value of the sample, x is less than functional values of occurrence probability of g (x) is report a higher expected value of sample x-values greater than function, the probability of occurrence.
Function f (x) g (x) product of function than separating location can be a maximum value function for optimal separation value.
The ESFD and SFD measurements from separating the two-value function shows the expected values of the third mintasor chart data: A = 0.1, B = 2.9; 93rd Tk item number.
Fig.3. Separating value
This example shows that different values can be suitable for separating ESFD as SFD but it can not be generalized. Past studies have shown that – if too slightly - more likely to be able to separate the two varieties of the ESFD, in the example shown by 3%.
3. Conclusion
The new defi nition could improve the ability of separating. Though just a little bit, the study only two types of potato tubers current state of limited series. Further testing should decide whether the sound practices developed HSFD or not.
4. Acknowledgement
This work was supported by Developing Competitiveness of Universities in the South Transdanubian Region (SROP-4.2.1.B-10/2/KONV-2010-0002).
5. References
Berke J. (2005): Spectral fractal dimension (Spektrális fraktáldimenzió), Proceedings of the 7th WSEAS Telecommunications and Informatics (TELE-INFO ’05), Prague, pp.23-26, ISBN 960 8457 11
Berke J. (2006): Measuring of Spectral Fractal Dimension (A spektrális fraktáldimenzió mérése), Advances in Systems, Computing Sciences and Software Engineering, Springer pp. 397–
402., ISBN 10 1- 4020-5262-6.
Berke J. (2007): Measuring of Spectral Fractal Dimension (A spektrális fraktáldimenzió mérése).
Journal of New Mathematics and Natural Computation, Vol. III/3, pp. 409-418., ISSN: 1793-0057.
Csák M. Hegedűs G. (2008): Az SFD mérésként való alkalmazhatósága a burgonyanemesítési kutatásokban (The DFD opportunity of applicable in a potatoimprovement research), Acta Agraria Kaposváriensis Vol 12 No 2, pp. 177-191.
Hegedűs G. (2007): Spectral fractal dimension – invariant transformations and shifting rules (Spektrális fraktáldimenzió – invariáns transzformációk és eltolási szabályok) Erdei Ferenc IV. Tudományos Konferencia, Kecskemét, 2007. augusztus 27-28. II. kötet, pp. 671 674., ISBN 978 963 7294 65 5.