Research Article  Open Access
A New Definition of Fractional Derivatives Based on Truncated LeftHanded GrünwaldLetnikov Formula with and Median Correction
Abstract
We propose a new definition of fractional derivatives based on truncated lefthanded GrünwaldLetnikov formula with and median correction. Analyzing the difficulties to choose the fractional orders and unsatisfied processing results in signal processing using fractionalorder partial differential equations and related methods; we think that the nonzero values of the truncated fractional order derivatives in the smooth regions are major causes for these situations. In order to resolve the problem, the absolute values of truncated parts of the GL formula are estimated by the median of signal values of the remainder parts, and then the truncated GL formula is modified by replacing each of the original signal value to the differences of the signal value and the median. Since the sum of the coefficients of the GL formula is zero, the median correction can reduce the truncated errors greatly to proximate GL formula better. We also present some simulation results and experiments to support our theory analysis.
1. Introduction
Partial differential equations (PDEs) and related methods are very important tools for signal processing [1–13]. Especially, in recent years progress was achieved in the theory of fractional calculus as a useful tool to handle applications in the area of physics, chemistry, and engineering sciences [6–20].
However, unlike integerorder derivatives with zeros or small values in the smooth regions and with big values near singularities, the values of truncated fractional derivatives are with very large absolute values. The direct result is that the derivative values cannot be used to measure the degrees of singularities. That is, small derivative values perhaps relate to singularity regions and big derivative values perhaps relate to smooth regions. Therefore, some integerorder PDE methods cannot be modified to their fractionalorder counterparts directly. Hu noticed the problem in 2013 and proposed a new fractionalorder PDE with different orders of different parts for fractionalorder PM method [13]. However, he did not give the analysis in theory.
In this paper, we study behaviors of fractionalorder derivatives of truncated lefthanded GrünwaldLetnikov formula with . Based on the analysis, a new fractional derivative formula is proposed based on truncated lefthanded GrünwaldLetnikov formula with and median correction. The median correction is used to reduce the truncated errors of GL formula.
The rest of this paper is as follows. Section 2 introduces some basic theory backgrounds in fractional derivatives; we also deduce some useful results based on these theory backgrounds. In Section 3 we introduce the truncated GL formula and its numerical approximation. Section 4 presents the new fractional derivative formula and gives properties and numerical methods for the new fractional definition. The simulation results and experiments are presented in Section 5. We also give conclusions and acknowledgments finally.
2. Fractional Derivatives
In this section, we will introduce some preparations for the new method, that is, GrünwaldLetnikov formula and its matrix approximation.
2.1. GrünwaldLetnikov Formula
Fractionalorder derivatives are defined as operators whose orders have been extended to noninteger numbers. There are a number of definitions of fractional derivatives. One usual way of representing the discrete fractional derivatives is by the GrünwaldLetnikov (GL) formula [21, 22], which is where , denotes the uniform space step, and represents the normalized GL weights which are given by
For , (1) becomes the classical 1st derivative, and for any , is a positive integer; they are classical th derivatives of . Note that for when is a positive integer, equations are with limit support whose support lengths are . However, for when is not an integer, fractional derivatives are nonlocal operators. That is, the value of the fractional derivative at a point depends on the function values at all the points to the left of the point .
Therefore, in order to handle fractional derivative numerically, it is necessary to compute the coefficients , where is the order of the fractional derivative. For that we can use the recurrence relationships:
Some useful properties for lefthanded GL formula are presented as follows.
Lemma 1. The nonlocal operator defined in (1) is a linear operator.
Proof. and are two functions, and ; is a real number. We have
Lemma 2. .
Proof. Since , let ; we have
Lemma 3. For , is a positive integer; one has
Proof. For , .
Assume that .
According to (3), we have . Since and , . Thus, . Here
Then, we have . Thus, .
Lemma 4. For , is an integer. One has
Proof. From Lemma 3, for , , .
Thus, for . We have , .
Since , we have , .
2.2. Numerical Method of GL Formula
For GL formula in (1) in signal processing, the uniform space step is set to for easy description; is the variant whose support is . That is, the signal is compact support. Therefore, equations can be specified as
The coefficients can also be obtained recursively from (3). We can discretize (1) into a finite difference on a grid on the axis, where the th lattice is ; th lattice is , ; and the th lattice is . That is, is the length of the signal.
Therefore, using the matrix approximate method, we have where represents truncated GL formula, , represents the transposed vector, and is an lower triangular strip matrix defined as
3. Truncated GrünwaldLetnikov Formula
Fractional integration and fractional differentiation are generalizations of notions of integerorder integration and differentiation and include th derivatives ( denotes an integer number) as particular cases. One usual way of representing the discrete fractional derivatives is by the GrünwaldLetnikov (GL) formula introduced in Section 2 (see (1)).
However, for digital signals, we have to discuss truncated GL formula rather than the GL formula itself because of the limited supports of digital signals. In this section, the definition of truncated GL formula and its properties are discussed firstly and then we will give its numerical scheme.
3.1. Truncated GrünwaldLetnikov Formula
The truncated GL formula is where , the uniform space step is set to , is the length of the support, represents the normalized GL weights, and their recurrence relationship is given by (3).
Just as above sections, we will discuss some properties of truncated GL formula.
Lemma . For , is an integer. One has
Proof. According to Lemma 2, we have According to Lemma 3, . Therefore, .
Lemma . For , is a positive integer, . One has
Proof. For , .
Assume that .
According to Lemma 1, we have . Since and , . Thus, . Here
Then, we have . Thus, .
Lemma . For , is an integer and is a positive integer. One has
Proof. From Lemma 3, for , , .
Thus, for . We have , , .
3.2. Numerical Method of Truncated GL Formula
We can discretize (12) into a finite difference on a grid on the axis, where the th lattice is ; th lattice is , ; and the th lattice is . That is, is length of signals.
Therefore, using the matrix approximate method, we have where , represents the transposed vector, and is an matrix defined as
Notice that is the length of the signal and is the length of the support. That is, . Since is sparser than for , the computation cost of the truncated GL formula is lower than that of the GL formula.
4. New Truncated GrünwaldLetnikov Formula
Although Lemma 2 tells us the values of fractional derivatives for a constant function defined by GL formula equal to zeros, for truncated GL formula, Lemma 2′ shows that it is not true.
The main awkwardness for this situation comparing with integerorder derivatives is that the fractional derivatives cannot be used to measure the strength of singularities. Therefore, estimation methods based on the strength of singularities measured by the modula of 1order derivatives cannot be generalized to their fractional counterparts directly. These estimation methods include many popular and stateofart frameworks, such as anisotropic diffusion, nonlocal means, and bilateral filtering.
Therefore, in order to generalize fractional derivatives to these frameworks, the truncated GL formula should be modified as follows: for , , .
We start from the requirement to obtain the definition and properties of the new truncated GL formula, and then the numerical method of the new model by matrix is presented.
4.1. Motivations and Definitions
The discussion is from the error of truncated GL formula.
Definition 5. The error of the truncated GL formula is
where represents truncated GL formula, is the support length, and is the signal, .
Especially, for , we have
that is, the sum of all terms after . From Lemma 2′, we have , for , and from Lemma 3, we have , for , which implies that the error will become smaller as becomes bigger.
Moreover, from Lemma 2, we have ; thus, the truncated error can be changed as follows.
Proposition 6. The truncated error of is
Proof. From Lemma 2 and the above equation, Thus, For , is a constant real number; we can get similar results.
Proposition 7. The truncated error of is
Similarly, for , values from to are not known and we only know the samples from to . Thus, we should estimate errors defined in (20) by the values from to . For , estimate errors defined in (20) by the values from to have been accomplished through Proposition 7, which reminds us that the problem can be solved by assuming a constant. Thus, times the constant and the sum of , can approximate the error well. This constant also should be estimated from values from to .
One alternative scheme is that the median of values from to is used as the estimate value for the constant since median is an estimate with good performance in flexibility and reliability.
Definition 8. The new truncated GL formula is where represents the new truncated GL formula, is the median of the vector , the uniform space step is set to , is the length of the support, and represents the normalized GL weights.
4.2. Numerical Method of New Truncated GL Formula
We can discretize (26) into a finite difference on a grid on the axis, where the th lattice is ; th lattice is , ; and the th lattice is . That is, is length of signals.
Therefore, using the matrix approximate method, we have where represents new truncated GL formula, , and represents the transposed vector. is the corrected vector of by the median of , and is an matrix defined as
Notice that is the length of the signal and is the length of the support. That is, . Since is sparser than for , the computation cost of the truncated GL formula is lower than the GL formula's.
4.3. Properties of the New Truncated GL Formula
In this subsection, we will give some important properties of the new truncated GL formula.
Theorem 9. The nonlocal operator defined in (27) is a linear operator.
Proof. and are two functions, and ; is a real number. We have
Here, are corrected vectors of , , which are defined as the corrected vector in (27).
Theorem 10. and are two matrixes defined in (27), that is, the approximation matrixes of truncated GL formula with fractional orders and , respectively. Thus, where , .
Lemma 12 can be proved easily by times two matrixes. That is, two operators of the truncated GL formula with different fractional orders are commutative.
We guess the following equations in Guess 1 are correct. However, we cannot prove it.
Guess 1. We have
5. Numerical Simulations
For the numerical approximation, although longer memory is more precise computation for the fractional derivatives, a fixed number for is adopted for reducing computation complexity, for example, or and so forth. But these truncated forms will lead to some unsatisfied results. In this section, we will give error analysis of truncated GL formulas firstly and then present experiments using test signals.
5.1. Error Analysis of Truncated GL Formula
The error analysis of truncated GL formula is very important in the applications of fractional derivatives. Some efforts discuss the problem in theory [23]. In this subsection we will discuss the truncated errors by considering the signal since the most serious effect of truncated errors is that the values of fractional derivatives are not equal to zeros when .
According to Lemma 2, the values of untruncated GL formula for are equal to zeros, which are coincident to the 1order derivatives. For the truncated GL formula when fractional order satisfies , according to Lemma 2′, the remainder part of is more than zero. Thus, the truncated part is less than zero. That is, it is a negative number. The truncated errors of truncated GL formula for with different support lengths and different fractionalorders , , are shown in Table 1, which demonstrate the above conclusions.

Moreover, we will compare the changes of absolute values of truncated errors with different and different support lengths.
Lemma 11. The absolute values of truncated errors , where is the support length, for , become smaller as the support lengths become larger.
Proof. According to Lemma 3, , , for is an integer.
According to Lemma 2′, , for is an integer and .
Here, can be considered as the support length of the truncated GL formula. When , where and are two support lengths and and , we have
Lemma 12. The absolute values of truncated errors for become smaller as the fractional orders become larger.
Proof. Since and , for , , and , where and are two fractional orders, we have
According to Lemma 2, then .
For and , we have . Thus,
where represents the absolute value. Thus, for , , and , we have
If , for , , , and .
Thus, we can assume that , , and , . We have
Since and , we have
Thus, and .
Summary of above two lemmas: we have that the long support and large fractional orders of truncated GL formulas will have small absolute values of truncated errors. Experiments shown in Table 1 and Figure 1 also support these theory analysis results. Note that the truncated errors are negative numbers. Therefore, discussing their absolute values can show the differences between zeros and errors.
5.2. Experiments
In order to test if the new truncated method can reduce the truncated errors in real signals, two test signals, blocks, and bumps, with samples, are used for analysis of the performance of our new framework (see Figures 2 and 3).
(a)
(b)
(c)
(d)
(a)
(b)
(c)
(d)
From Figure 2, we can see both integerorder derivatives (green lines) and fractional derivatives ( are represented by red lines and are represented by black lines) have the properties that the singularities are related to the local extrema. Just as discussed above, the values of truncated GL formula are not zeros in smooth regions (see Figures 2(a) and 2(c)). Moreover, coinciding Lemma 11, the signals with longer truncated length will be nearer to zeros in the smooth segments than the shorter length signals. The bigger fractional orders will also have better performance in sharper impulses in singularities and much more near zeros in the smooth segments than the smaller fractional orders, which coincides Lemma 12 (see Figures 2(a) and 2(c)).
When truncated length for the new definition of fractional derivatives (see Figure 2(b)), the values of new definition at singularities have very high impulses comparing to the corresponding truncated GL formula (Figure 2(a)), which is a very impressive nature to detect, locate, and preserve singularities. Moreover, near singularities, the values are decreased/increased gradually to zeros and the values of the smooth segments are zeros. These characters together ensure that the new fractional derivatives can obtain good results in signal processing.
Observing Figure 3, we can find that 1order derivatives cannot locate “bump like” singularities: (1) there are two extremums of 1order derivatives even for a very narrow bump, for example, the first bump; (2) some weak singularities cannot be detected, for example, the left singularity of the fifth bump.
Fortunately, all fractional derivatives have better performance in above two sides. That is, each narrow bump only has one extremum of the fractional derivatives.
The best performance of the new definition for block signal is the derivative with truncated length and fractional order , which also has the best natures for the bump signal. That is, it has very high impulses in singularities and equals to zeros in smooth segments. Weak singularity in the left of the fifth bump has a very obvious high impulse, which can be detected and located easily. It is very interesting thing that the weak impulses in two signals using truncated GL formula are enhanced by new definitions of fractional derivatives with small and by comparing Figures 2(a) and 2(b) and Figures 3(a) and 3(b).
In summary, from the simulation results of two test signals, we can conclude that the new definition of fractional derivatives has the best performance in three type derivatives including 1order derivatives, truncated GL formula and themselves.
6. Conclusions
In this paper, we study errors of truncated GrünwaldLetnikov formula with and then the errors are corrected by the medians of remainder parts of signals, which has some very impressive natures in signal processing. That is, it has very high impulses in singularities, which can detect and locate singularities easily; the values of the new definition are equal to zeros in smooth segments, which can be used efficiently in signal smoothing and filtering. Moreover, it also has good performance in weak singularity detection and location.
Conflict of Interests
The author declares that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
This paper is supported by the National Natural Science Foundation of China (nos. 60873102 and 60873264), Major State Basic Research Development Program (no. 2010CB732501), and Open Foundation of Visual Computing and Virtual Reality Key Laboratory of Sichuan Province (no. J2010N03). This work was supported by a grant from the National High Technology Research and Development Program of China (no. 2009AA12Z140).
References
 S. Hu, Z. Liao, D. Sun, and W. Chen, “A numerical method for preserving curve edges in nonlinear anisotropic smoothing,” Mathematical Problems in Engineering, vol. 2011, Article ID 186507, 14 pages, 2011. View at: Publisher Site  Google Scholar  Zentralblatt MATH  MathSciNet
 Y. Chen, Y. Li, W. Yu, L. Luo, W. Chen, and C. Toumoulin, “Jointmap tomographic reconstruction with patch similarity based mixture prior model,” Multiscale Modeling & Simulation, vol. 9, no. 4, pp. 1399–1419, 2011. View at: Publisher Site  Google Scholar  Zentralblatt MATH  MathSciNet
 Y. Chen, X. Yin, L. Shi et al., “Improving abdomen tumor lowdose CT images using a fast dictionary learning based processing,” Physics in Medicine and Biology, vol. 58, pp. 5803–5820, 2013. View at: Google Scholar
 Z. Bian, J. Ma, J. Huang et al., “SRNLM: a sinogram restoration induced nonlocal means image filtering for lowdose computed tomography,” Computerized Medical Imaging and Graphics, vol. 37, no. 4, pp. 293–303, 2013. View at: Google Scholar
 J. Ma, H. Zhang, Y. Gao et al., “Iterative image reconstruction for cerebral perfusion CT using a precontrast scan induced edgepreserving prior,” Physics and Medical Biology, vol. 57, pp. 7519–7542, 2012. View at: Google Scholar
 Z. Liao, “Lowdosed xray computed tomography imaging by regularized fully spatial fractionalorder PeronaMalik diffusion,” Advances in Mathematical Physics, vol. 2013, Article ID 371868, 9 pages, 2013. View at: Publisher Site  Google Scholar  MathSciNet
 S. Hu, Z. Liao, and W. Chen, “Sinogram restoration for lowdosed Xray computed tomography using fractionalorder PeronaMalik diffusion,” Mathematical Problems in Engineering, vol. 2012, Article ID 391050, 13 pages, 2012. View at: Google Scholar  Zentralblatt MATH  MathSciNet
 Z. Jun and W. Zhihui, “A class of fractionalorder multiscale variational models and alternating projection algorithm for image denoising,” Applied Mathematical Modelling. Simulation and Computation for Engineering and Environmental Systems, vol. 35, no. 5, pp. 2516–2528, 2011. View at: Publisher Site  Google Scholar  Zentralblatt MATH  MathSciNet
 J. Bai and X.C. Feng, “Fractionalorder anisotropic diffusion for image denoising,” IEEE Transactions on Image Processing, vol. 16, no. 10, pp. 2492–2502, 2007. View at: Publisher Site  Google Scholar  MathSciNet
 M. Janev, S. Pilipović, T. Atanacković, R. Obradović, and N. Ralević, “Fully fractional anisotropic diffusion for image denoising,” Mathematical and Computer Modelling, vol. 54, no. 12, pp. 729–741, 2011. View at: Publisher Site  Google Scholar  Zentralblatt MATH  MathSciNet
 C. Wang, L. Lan, and S. Zhou, “GrunwaldLetnikov based adaptive fractional differential algorithm on image texture enhancing,” Journal of Computational Information Systems, vol. 9, no. 2, pp. 445–454, 2013. View at: Google Scholar
 H. A. Jalab and R. W. Ibrahim, “Texture enhancement based on the SavitzkyGolay fractional differential operator,” Mathematical Problems in Engineering, vol. 2013, Article ID 149289, 8 pages, 2013. View at: Publisher Site  Google Scholar  MathSciNet
 S. Hu, “External fractionalorder gradient vector PeronaMalik diffusion for sinogram restoration of lowdosed xray computed tomography,” Advances in Mathematical Physics, vol. 2013, Article ID 516919, 10 pages, 2013. View at: Publisher Site  Google Scholar  MathSciNet
 R. Hilfer, Applications of Fractional Calculus in Physics, World Scientific Publishing, River Edge, NJ, USA, 2000. View at: Publisher Site  MathSciNet
 S. Hu and P. Liang, “Theory analysis of lefthanded GrnwaldLetnikov formula with $0<\alpha <1$ to detect and locate singularities,” Abstract and Applied Analysis. In press. View at: Google Scholar
 W.S. Chen, C. Zhang, and S. Y. Chen, “Geometric distribution weight information modeled using radial basis function with fractional order for linear discriminant analysis method,” Advances in Mathematical Physics, vol. 2013, Article ID 825861, 9 pages, 2013. View at: Publisher Site  Google Scholar
 W.S. Chen, P. C. Yuen, and X. Xie, “Kernel machinebased ranklifting regularized discriminant analysis method for face recognition,” Neurocomputing, vol. 74, no. 17, pp. 2953–2960, 2011. View at: Publisher Site  Google Scholar
 M. Li, “A class of negatively fractal dimensional Gaussian random functions,” Mathematical Problems in Engineering, vol. 2011, Article ID 291028, 18 pages, 2011. View at: Publisher Site  Google Scholar  Zentralblatt MATH  MathSciNet
 M. Li and W. Zhao, “On bandlimitedness and laglimitedness of fractional Gaussian noise,” Physica A: Statistical Mechanics and its Applications, vol. 392, no. 9, pp. 1955–1961, 2013. View at: Publisher Site  Google Scholar  MathSciNet
 M. Li, S. C. Lim, C. Cattani, and M. Scalia, “Characteristic roots of a class of fractional oscillators,” Advances in High Energy Physics, vol. 2013, Article ID 853925, 7 pages, 2013. View at: Publisher Site  Google Scholar  MathSciNet
 I. Podlubny, Fractional Differential Equations, vol. 198 of Mathematics in Science and Engineering, Academic Press, San Diego, Calif, USA, 1999. View at: MathSciNet
 S. G. Samko, A. A. Kilbas, and O. I. Marichev, Fractional Integrals and Derivatives: Theory and Applications, Gordon and Breach Science Publishers, London, UK, 1993. View at: MathSciNet
 E. Sousa, “How to approximate the fractional derivative of order $1<\alpha \le 2$,” International Journal of Bifurcation and Chaos in Applied Sciences and Engineering, vol. 22, no. 4, 13 pages, 2012. View at: Publisher Site  Google Scholar  Zentralblatt MATH  MathSciNet
Copyright
Copyright © 2014 Zhiwu Liao. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.