How to find a local maximum in surf (3D) plot ?

I would like to find a local maximum in 3D surf plot and to draw the identified maximum as a line To further understand what i want to do, i drew a black line by my hand as shown in attached figure,

How could i do this?

[sung-cheol kwon]

Here`s a example plot

which i want to obtain

[José-Luis]

To me this looks like a catchment delineation problem. You could start with a flow accumulation and follow all the cells that have an accumulation of one and are not on the border.

https://se.mathworks.com/matlabcentral/fileexchange/14504-flow-accumulation--upslope-area-

[Scout Patel]

[TF1,P] = islocalmax(Array,2,'MinProminence',4);%theshold at 4...change accordingly

P(TF1)

V1 = find(TF1, 3, 'first');%this finds the first 3...change accordingly

[y1,x1]=ind2sub(size(Array),V1)

% Fit line to data using polyfit

c = polyfit(x1,y1,1);

% Display evaluated equation y = m*x + b

disp(['Equation is y = ' num2str(c(1)) '*x + ' num2str(c(2))])

% Evaluate fit equation using polyval

y_est = polyval(c,x1);

% Add trend line to plot

hold on

plot(x1,y_est,'r--','LineWidth',0.5)

hold off

Answer 1

[Walter Roberson]

Possibly https://www.mathworks.com/help/images/ref/imgradient.html

You can also proceed by taking max() along a dimension of your data. There are a couple of problems you would need to work out with that:

A) The line you are interested in might not cross the entire array, but max() is going to find a maximum for every row or column anyhow

B) The line you are interested in might happen to travel sharply in the direction along which you are taking the max; you would only get one location per row or column, whereas if you had happened to ask along the other dimension you would have gotten a nice answer.

Answer 2

[Dean Culver]

This would be computationally inefficient, but I think it would get the job done:

• Determine the maximum value of the surface function z on the perimeter. Call it z_(p,max).
• Determine the row and column indices (i,j) such that z(i,j)=z_(p,max).
• Record x(i,j) and y(i,j)
• (begin loop) Sample the points immediately adjacent to the current (i,j).
• Choose the maximum of these which is not a previously recorded value.
• Record the new (i,j), x(i,j), y(i,j), and z(i,j).
• (end loop when you've reached a perimeter and can no longer proceed)
• plot3 to get the curve