How to find particular points from a waveform data

[cbutcher]

---

@Henrik_Volkers wrote:

Here is my try 😉

remove mean, detect zero crossings , slice the wfrm, fit a polynom (migth be better ones) , get the endpoints

---

It's great that this worked, but in GerdW's defense I feel like removing the mean and looking for zero crossings is the same as using a threshold detection with the mean value, no?

[Henrik_Volkers]

---

@cbutcher wrote:

---

@Henrik_Volkers wrote:

Here is my try 😉

remove mean, detect zero crossings , slice the wfrm, fit a polynom (migth be better ones) , get the endpoints

---

It's great that this worked, but in GerdW's defense I feel like removing the mean and looking for zero crossings is the same as using a threshold detection with the mean value, no?

---

Yep, that`s equal, .. 

Gio, if you post some critical data, maybe someone comes up with a better idea.

[EamonnFPMcKernan]

As ico82 intimated, the second derivative rapidly changes sign at point 2 before going to zero again, so one does not require a threshold to detect this change. . .

[Henrik_Volkers]

Maybe:

Detect the falling slope (mean ...)

use peakdetect with 3 points to get the next valley

do a lin fit for next n (100) noisy points

for the rising edge just reverse the data ....

as always: look at the data, analyse how you detekt the point, draw it in dataflow 😄

[EamonnFPMcKernan]

I have not run the code I am just looking at the graph. Based upon the plot it appears that point 2 occurs unambiguously when

a) the second derivative changes sign from negative to positive

b) the first derivative is positive before the sign change