Skip to main content
. 2014 Mar 6;9(3):e90500. doi: 10.1371/journal.pone.0090500

Figure 2. Filtering exponentials convolved with a system response function.

Figure 2

(a) Exponential on the interval [−1, 1] with Poisson noise added. Amplitude, Inline graphic, time constant Inline graphic. (b) Legendre spectrum of x as resulting from eq. 5. (c) Mean Inline graphic of Inline graphic (continuous) and inverse fLT (dashed, eq. 6) of Inline graphic through Inline graphic of the spectrum shown in b. (d) Legendre spectrum of the mean Inline graphic, largely lacking higher noise components. (e) Noisy curve is the convolution of Inline graphic with Inline graphic and Inline graphic. Inline graphic was chosen such that the curve overlaps with Inline graphic for large Inline graphic. (f) Legendre spectrum (gray bars) of convoluted noisy exponential shown in e (continuous curve). The lowpass-filtered inverse transform is shown in e (continuous curve) and approximates the convoluted noisy exponential. In addition, f shows the Legendre spectrum of Inline graphic, obtained through eq. 9. The lowpass-filtered inverse transform of this spectrum is shown as the red dashed curve in e and approximates the original non-convoluted exponential, from which the noisy convoluted curve was generated.