Feature Extraction for Radar Data

I have Matlab data file containing information for 27 different targets which have been scanned using radar (I cannot share this data). I load the data into matlab using

load('data')

[a, b] = size(data4); %the data is called data4 once loaded, I don't know why

which produces a structure with 27 fields (27 targets) with 6 columns holding information for each target, such as RCS (radar cross section), target array values, name, start frequency etc. The columns which produce values to be plotted are RCS and Target_array which are both stored as complex doubles (both 501x81) within the struct. I can then produce graphs for each target using the following code:

for row=1:b
    [a, b]=size(data4(row).RCS);
    freq=linspace(16.5,26.5,(length(data4(row).RCS)));
    figure;mesh((0:b-1),freq,abs(data4(row).RCS));
    axis tight
    xlabel('Angle');
    ylabel('Freq GHz');
    zlabel('RCS m^2');
    s=data4(row).target;
    title(s);view(-5,60)
    set(gca,'FontSize',20,'fontWeight','bold')
    set(findall(gcf,'type','text'),'FontSize',20,'fontWeight','bold')
    colormap jet;colorbar;
end

My question is, what can I do now to extract features from this data? I can't find any help online with regards to toolboxes or code to extract features and my knowledge on this subject is very limited.

[Nicholas]

Dear Dean,

I am working on a similar problem with a similar situation presumably. Have you made any headway?

Best regards,

Nick

[Dean Kennedy]

Hi Nicholas,

I have tried a few different methods, though my programming ability means I'm not quite up to the task.

There are a few different methods I have used:

• Measuring symmetry
• Measuring peaks
• Principle component analysis

I've been suggested these by a few independent sources, so I assume these methods are most likely to work but I haven't been able to produce any significant results.

Apologies for not helping more

Dean