Not the problem with the data, but in MATLAB replace
for k=1:length(filtrado)
if filtrado(k)<0
filtrado(k)=0;
end
end
with
filtrado(filtrado<0)=0;
The problem in your code is that you didn't write code to handle the possibility a given row in the image may, indeed, not have a discernible peak.
We have no way to know anything about whether there's some way to be able to help findpeaks locate minimal excursions or not without the data to observe; generally it is pretty adept--but, it's designed for typical time series one obtains from things like measurements of acceleration or the like; depending upon what your image looks like, it might not be the best tool.
Or, there just may not be any discernible peaks in the data you have...
