How to find the peaks in the signal with descending order only?

The following code find the peaks in my signal perfectly.

[pks,locs] = findpeaks(Power,depth,'MinPeakDistance',100,'MinPeakProminence',4);

But I am more interested in the descending order peaks and the code should exclude the values not following the desceding order pattern.

For instance at locs = [1,2,3,4,5,6,7], pks are [100,80,60,70,50,80].

Required: [pks1,locs1] = function_descending(pks,locs);

I only need pks and corresponding locs which are only in descending order like the result should be locs1=[1,2,3,5] and pks1=[100,80,60,50]. It should exclude the values which contradicts the descending phenomena. The values of 70 at position 4 and 80 at position 6 are excluding because they are higher then all preceeding numbers.

[Syed Abdul Salam]

Example 2:

pks = [100,80,60,70,50,80,90,100,51];

locs = [1,2,3,4,5,6,7,8,9];

result should be

pks1 = [100,80,60,50];

locs1 = [1,2,3,5]

Answer 1

[Star Strider]

This works, with the example you provided:

pks = [100,80,60,70,50,80]';
locs = [1,2,3,4,5,6]';

dpks = diff([pks(1); pks]);                     % Non-Descending Will Be Positive
Lv = dpks <= 0;                                 % Logical Vector Mask
pks1 = pks(Lv);                                 % New Peaks
locs1 = locs(Lv);                               % New Locs

figure
plot(locs, pks, '^')
hold on
plot(locs1, pks1, 's')
hold off

It does not do any rigorous checking, and just looks ad adjacent peaks.

You can easily wrap it in a function:

function [pks1,locs1] = descending_peaks(pks,locs)
dpks = diff([pks(1); pks]);                     % Non-Descending Will Be Positive
Lv = dpks <= 0;                                 % Logical Vector Mask
pks1 = pks(Lv);                                 % New Peaks
locs1 = locs(Lv);                               % New Locs
end

Experiment to get the result you want.

[Syed Abdul Salam]

Thanks for the answer, I think it works with the provided example but checking only the next element. For instance, its failing with the following example pks = [100,80,60,70,50,80,90,100,51]';

locs = [1,2,3,4,5,6,7,8,9]';

dpks = diff([pks(1); pks]); % Non-Descending Will Be Positive

Lv = dpks <= 0; % Logical Vector Mask

pks1 = pks(Lv) % New Peaks

locs1 = locs(Lv) % New Locs

It results in pks1= [100,80,60,50,51].

I need it for a long sequence of numbers which can have higher value at any later position and it should only result descending sequence.

[Syed Abdul Salam]

Thanks I have done it with the help of your code, just repeating it many times.

function [pks1,locs1] = descending_peaks(pks,locs)

for i = 1:length(pks)-1

dpks = diff([pks(1); pks]); % Non-Descending Will Be Positive

Lv = dpks <= 0; % Logical Vector Mask

pks = pks(Lv); % New Peaks

locs = locs(Lv); % New Locs

end

pks1 = pks;

locs1 = locs;

end

[Star Strider]

As always, my pleasure!

I was considering that option as well, although I was considering a one-pass solution. If I develop one, I will post back here.