I'm looking for the Fourier transformation of the (constant) uniform B-Spline $$N_0(x) = \begin{cases}1 & 0 \leqslant x < 1 \\ 0 & otherwise \end{cases}$$

If $N_0(x)$ would also attain value $1$ when $x=1$ (i.e. the $<$ would be $\leqslant$), it would just be a shifted version of the rectangular function R(x), which has (if I'm not mistaken) the following Fourier transform ($T=1$ and shifted $1/2$ to the right):

$$\hat{R}(\omega) = e^{-\tfrac{i\omega}{2}}\, \frac{\sin(\omega/2)}{\omega/2}$$

However, I'm not sure if this is helpful in finding $\hat{N_0}(\omega)$, the Fourier transform of $N_0(x)$...

Once I know $\hat{N_0}(\omega)$, it is easy to find the Fourier transforms of higher order uniform B-Splines, since they are (or can be) defined using convolution. Of course, the Fourier transform of a convolution is just a multiplication of the two Fourier transformed functions, therefore:

$$\hat{N_k}(\omega) = \left( \hat{N_0}(\omega) \right)^{k+1}$$


Since the Fourier transform is defined through an integral, the condition $f=g$ almost everywhere implies $\hat{f} = \hat{g}$. So in this case the Fourier transform of $N_0$ is indeed the same as that of $R$, and your formula is correct, assuming you use the convention $\hat{f}(\omega) = \int_{-\infty}^\infty f(x) e^{-i\omega x} \, dx$ for the Fourier transform.

  • $\begingroup$ Thanks. Somewhere I found stated that $\hat{N_0}(\omega)=\dfrac{1-e^{-i \omega}}{i \omega}$. I'm not quite sure whether this is the same as the function I found -- I'll take a more detailed look at a later time and update my post / add a comment. $\endgroup$ – Ailurus Oct 21 '12 at 20:11
  • $\begingroup$ Yes, that is the same, using the identity $\sin z = \frac{e^{iz} - e^{-iz}}{2i}$ with $z = \omega/2$. $\endgroup$ – Lukas Geyer Oct 21 '12 at 20:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.