En el ppt de la lección 4, hay un diagrama de bloques del procesamiento de SAR. El código matlab de SAR básicamente coincide con esto.
El código de matlab se divide en dos archivos, SBAND_RMA_opendata es para leer los datos y llamar a SBAND_RMA_IFP para procesar los datos después de leerlos.
Una cosa a tener en cuenta es que el do_all_plots predeterminado es 0, y muchos códigos para trazar datos intermedios realmente no se utilizan.
A continuación se muestra SBAND_RMA_opendata.m
%-------------------------------------------%
%Process raw data here
clear all;
close all;
%read the raw data .wave file here
[Y,FS,NBITS] = wavread('towardswarehouse.wav');
%constants
c = 3E8; %(m/s) speed of light
%radar parameters
Tp = 20E-3; %(s) pulse time
Trp = 0.25; %(s) min range profile time duration
N = Tp*FS; %# of samples per pulse
fstart = 2260E6; %(Hz) LFM start frequency
fstop = 2590E6; %(Hz) LFM stop frequency
%fstart = 2402E6; %(Hz) LFM start frequency for ISM band
%fstop = 2495E6; %(Hz) LFM stop frequency for ISM band
BW = fstop-fstart; %(Hz) transmti bandwidth
f = linspace(fstart, fstop, N/2); %instantaneous transmit frequency
%the input appears to be inverted
trig = -1*Y(:,1);
s = -1*Y(:,2);
clear Y;
%parse data here by position (silence between recorded data)
rpstart = abs(trig)>mean(abs(trig));
count = 0;
Nrp = Trp*FS; %min # samples between range profiles
for ii = Nrp+1:size(rpstart,1)-Nrp
if rpstart(ii) == 1 & sum(rpstart(ii-Nrp:ii-1)) == 0
count = count + 1;
RP(count,:) = s(ii:ii+Nrp-1);
RPtrig(count,:) = trig(ii:ii+Nrp-1);
end
end
%parse data by pulse
count = 0;
thresh = 0.08;
clear ii;
for jj = 1:size(RP,1)
%clear SIF;
SIF = zeros(N,1);
start = (RPtrig(jj,:)> thresh);
count = 0;
jj
for ii = 12:(size(start,2)-2*N)
[Y I] = max(RPtrig(jj,ii:ii+2*N));
if mean(start(ii-10:ii-2)) == 0 & I == 1
count = count + 1;
SIF = RP(jj,ii:ii+N-1)' + SIF;
end
end
%hilbert transform
q = ifft(SIF/count);
sif(jj,:) = fft(q(size(q,1)/2+1:size(q,1)));
end
sif(find(isnan(sif))) = 1E-30; %set all Nan values to 0
%SAR data should be ready here
clear s;
s = sif;
save routsidewarehouse2 s; %for image data
%-------------------------------------------%
%load additional varaibles and setup constants for radar here
clear all;
c = 3E8; %(m/s) speed of light
%load IQ converted data here
load routsidewarehouse2 s; %load variable sif %for image data
for ii = 1:size(s,1)
s(ii,:) = s(ii,:) - mean(s,1);
end
%sif = s-sif_sub; %perform coherent background subtraction
%sif = sif_sub; %image just the background
sif = s; %image without background subtraction
clear s;
clear sif_sub;
%***********************************************************************
%radar parameters
fc = (2590E6 - 2260E6)/2 + 2260E6; %(Hz) center radar frequency
B = (2590E6 - 2260E6); %(hz) bandwidth
cr = B/20E-3; %(Hz/sec) chirp rate
Tp = 20E-3; %(sec) pulse width
%VERY IMPORTANT, change Rs to distance to cal target
%Rs = (12+9/12)*.3048; %(m) y coordinate to scene center (down range), make this value equal to distance to cal target
Rs = 0;
Xa = 0; %(m) beginning of new aperture length
delta_x = 2*(1/12)*0.3048; %(m) 2 inch antenna spacing
L = delta_x*(size(sif,1)); %(m) aperture length
Xa = linspace(-L/2, L/2, (L/delta_x)); %(m) cross range position of radar on aperture L
Za = 0;
Ya = Rs; %THIS IS VERY IMPORTANT, SEE GEOMETRY FIGURE 10.6
t = linspace(0, Tp, size(sif,2)); %(s) fast time, CHECK SAMPLE RATE
Kr = linspace(((4*pi/c)*(fc - B/2)), ((4*pi/c)*(fc + B/2)), (size(t,2)));
%Save background subtracted and callibrated data
save sif sif delta_x Rs Kr Xa;
%clear all;
%run IFP
SBAND_RMA_IFP;
Lo que hace es similar a la medición de distancia anterior, leyendo los datos (pulso) cuando se enciende cada onda cuadrada.
for ii = Nrp+1:size(rpstart,1)-Nrp
if rpstart(ii) == 1 & sum(rpstart(ii-Nrp:ii-1)) == 0
count = count + 1;
RP(count,:) = s(ii:ii+Nrp-1);
RPtrig(count,:) = trig(ii:ii+Nrp-1);
end
endDespués de leer, sume estos datos de pulso
for ii = 12:(size(start,2)-2*N)
[Y I] = max(RPtrig(jj,ii:ii+2*N));
if mean(start(ii-10:ii-2)) == 0 & I == 1
count = count + 1;
SIF = RP(jj,ii:ii+N-1)' + SIF;
end
endEntonces Hilbert se transforma
%hilbert transform
q = ifft(SIF/count);
sif(jj,:) = fft(q(size(q,1)/2+1:size(q,1)));Todos los algoritmos posteriores se completan en SBAND_RMA_IFP.m.
%***********************************************************************
%a note on formatting, our convention is sif(Xa,t)
clear all;
load sif;
%apply hanning window to data first
N = size(sif,2);
for ii = 1:N
H(ii) = 0.5 + 0.5*cos(2*pi*(ii-N/2)/N);
end
for ii = 1:size(sif,1)
sif_h(ii,:) = sif(ii,:).*H;
end
sif = sif_h;
figcount = 1;
close_as_you_go = 0;
do_all_plots = 0;
set(0,'defaultaxesfontsize',13); %set font size on plots so we can see it in the dissertation
% NOTE: the function 'dbv.m' is just dataout = 20*log10(abs(datain));
%***********************************************************************
if do_all_plots == 1,
figure(figcount);
S_image = angle(sif);
imagesc(Kr, Xa, S_image);
colormap(gray);
title('Phase Before Along Track FFT');
xlabel('K_r (rad/m)');
ylabel('Synthetic Aperture Position, Xa (m)');
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'radians','fontsize',13);
print(gcf, '-djpeg100', 'phase_before_along_track_fft.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;
end
%along track FFT (in the slow time domain)
%first, symetrically cross range zero pad so that the radar can squint
zpad = 2048; %cross range symetrical zero pad
szeros = zeros(zpad, size(sif,2));
for ii = 1:size(sif,2)
index = round((zpad - size(sif,1))/2);
szeros(index+1:(index + size(sif,1)),ii) = sif(:,ii); %symetrical zero pad
end
sif = szeros;
clear ii index szeros;
S = fftshift(fft(sif, [], 1), 1);
clear sif;
Kx = linspace((-pi/delta_x), (pi/delta_x), (size(S,1)));
if do_all_plots == 1,
figure(figcount);
S_image = dbv(S);
imagesc(Kr, Kx, S_image, [max(max(S_image))-40, max(max(S_image))]);
colormap(gray);
title('Magnitude After Along Track FFT');
xlabel('K_r (rad/m)');
ylabel('K_x (rad/m)');
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'dB','fontsize',13);
print(gcf, '-djpeg100', 'mag_after_along_track_fft.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;
end
if do_all_plots == 1,
figure(figcount);
S_image = angle(S);
imagesc(Kr, Kx, S_image);
colormap(gray);
title('Phase After Along Track FFT');
xlabel('K_r (rad/m)');
ylabel('K_x (rad/m)');
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'radians','fontsize',13);
print(gcf, '-djpeg100', 'phase_after_along_track_fft.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;
end
if do_all_plots == 1,
figure(figcount);
S_image = dbv(fftshift(fft(S, [], 2), 2));
imagesc(linspace(-0.5, 0.5, size(S, 2)), Kx, S_image, [max(max(S_image))-40, max(max(S_image))]);
colormap(gray);
title('Magnitude of 2-D FFT of Input Data');
xlabel('R_{relative} (dimensionless)');
ylabel('K_x (rad/m)');
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'dB','fontsize',13);
print(gcf, '-djpeg100', 'mag_after_2D_fft.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;
end
%**********************************************************************
%matched filter
%create the matched filter eq 10.8
for ii = 1:size(S,2) %step thru each time step row to find phi_if
for jj = 1:size(S,1) %step through each cross range in the current time step row
%phi_mf(jj,ii) = -Rs*Kr(ii) + Rs*sqrt((Kr(ii))^2 - (Kx(jj))^2);
phi_mf(jj,ii) = Rs*sqrt((Kr(ii))^2 - (Kx(jj))^2);
Krr(jj,ii) = Kr(ii); %generate 2d Kr for plotting purposes
Kxx(jj,ii) = Kx(jj); %generate 2d Kx for plotting purposes
end
end
smf = exp(j*phi_mf); %%%%%%%%%%%%
%note, we are in the Kx and Kr domain, thus our convention is S_mf(Kx,Kr)
%appsly matched filter to S
S_mf = S.*smf;
clear smf phi_mf;
if do_all_plots == 1,
figure(figcount);
S_image = angle(S);
imagesc(Kr, Kx, S_image);
colormap(gray);
title('Phase After Matched Filter');
xlabel('K_r (rad/m)');
ylabel('K_x (rad/m)');
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'radians','fontsize',13);
print(gcf, '-djpeg100', 'phase_after_matched_filter.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;
end
clear S;
if do_all_plots == 1,
figure(figcount);
S_image = dbv(fftshift(fft(S_mf, [], 2), 2));
imagesc(linspace(-0.5, 0.5, size(S_mf, 2)), Kx, S_image, [max(max(S_image))-40, max(max(S_image))]);
colormap(gray);
title('Magnitude of 2-D FFT of Matched Filtered Data');
xlabel('R_{relative} (dimensionless)');
ylabel('K_x (rad/m)');
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'dB','fontsize',13);
print(gcf, '-djpeg100', 'mag_after_downrange_fft_of_matched_filtered_data.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;
end
%**********************************************************************
%perform the Stolt interpolation
%FOR DATA ANALYSIS
kstart =73;
kstop = 108.5;
%kstart = 95;
%kstop = 102;
Ky_even = linspace(kstart, kstop, 1024); %create evenly spaced Ky for interp for real data
clear Ky S_St;
%for ii = 1:size(Kx,2)
count = 0;
for ii = 1:zpad;
%for ii = round(.2*zpad):round((1-.2)*zpad)
count = count + 1;
Ky(count,:) = sqrt(Kr.^2 - Kx(ii)^2);
%S_st(ii,:) = (interp1(Ky(ii,:), S_mf(ii,:), Ky_even)).*H;
S_st(count,:) = (interp1(Ky(count,:), S_mf(ii,:), Ky_even));
end
S_st(find(isnan(S_st))) = 1E-30; %set all Nan values to 0
clear S_mf ii Ky;
if do_all_plots == 1,
figure(figcount);
S_image = angle(S_st);
imagesc(Ky_even, Kx, S_image);
%imagesc(S_image);
colormap(gray);
title('Phase After Stolt Interpolation');
xlabel('K_y (rad/m)');
ylabel('K_x (rad/m)');
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'radians','fontsize',13);
print(gcf, '-djpeg100', 'phase_after_stolt_interpolation.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;
end
%apply hanning window to data, cleans up data ALOT
N = size(Ky_even,2);
for ii = 1:N
H(ii) = 0.5 + 0.5*cos(2*pi*(ii-N/2)/N);
end
for ii = 1:size(S_st,1)
S_sth(ii,:) = S_st(ii,:).*H;
end
%S_st = S_sth;
%*********************************************************************
%perform the inverse FFT's
%new notation: v(x,y), where x is crossrange
%first in the range dimmension
clear v Kr Krr Kxx Ky_even;
v = ifft2(S_st,(size(S_st,1)*4),(size(S_st,2)*4));
%bw = (3E8/(4*pi))*(max(xx)-min(xx));
bw = 3E8*(kstop-kstart)/(4*pi);
max_range = (3E8*size(S_st,2)/(2*bw))*1/.3048;
figure(figcount);
S_image = v; %edited to scale range to d^3/2
S_image = fliplr(rot90(S_image));
cr1 = -80; %(ft)
cr2 = 80; %(ft)
dr1 = 1 + Rs/.3048; %(ft)
dr2 = 350 + Rs/.3048; %(ft)
%data truncation
dr_index1 = round((dr1/max_range)*size(S_image,1));
dr_index2 = round((dr2/max_range)*size(S_image,1));
cr_index1 = round(( (cr1+zpad*delta_x/(2*.3048)) /(zpad*delta_x/.3048))*size(S_image,2));
cr_index2 = round(( (cr2+zpad*delta_x/(2*.3048))/(zpad*delta_x/.3048))*size(S_image,2));
trunc_image = S_image(dr_index1:dr_index2,cr_index1:cr_index2);
downrange = linspace(-1*dr1,-1*dr2, size(trunc_image,1)) + Rs/.3048;
crossrange = linspace(cr1, cr2, size(trunc_image, 2));
%scale down range columns by range^(3/2), delete to make like
%dissertation again
clear ii;
for ii = 1:size(trunc_image,2)
trunc_image(:,ii) = (trunc_image(:,ii)').*(abs(downrange*.3048)).^(3/2);
end
trunc_image = dbv(trunc_image); %added to scale to d^3/2
imagesc(crossrange, downrange, trunc_image, [max(max(trunc_image))-40, max(max(trunc_image))-0]);
colormap('default');
title('Final Image');
ylabel('Downrange (ft)');
xlabel('Crossrange (ft)');
axis equal;
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'dB','fontsize',13);
print(gcf, '-djpeg100', 'final_image.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;Primero hice fft, no sé si es la matriz de calibración mencionada en el ppt.
%along track FFT (in the slow time domain)
%first, symetrically cross range zero pad so that the radar can squint
zpad = 2048; %cross range symetrical zero pad
szeros = zeros(zpad, size(sif,2));
for ii = 1:size(sif,2)
index = round((zpad - size(sif,1))/2);
szeros(index+1:(index + size(sif,1)),ii) = sif(:,ii); %symetrical zero pad
end
sif = szeros;
clear ii index szeros;
S = fftshift(fft(sif, [], 1), 1);
clear sif;
Kx = linspace((-pi/delta_x), (pi/delta_x), (size(S,1)));Luego está el filtro combinado
%matched filter
%create the matched filter eq 10.8
for ii = 1:size(S,2) %step thru each time step row to find phi_if
for jj = 1:size(S,1) %step through each cross range in the current time step row
%phi_mf(jj,ii) = -Rs*Kr(ii) + Rs*sqrt((Kr(ii))^2 - (Kx(jj))^2);
phi_mf(jj,ii) = Rs*sqrt((Kr(ii))^2 - (Kx(jj))^2);
Krr(jj,ii) = Kr(ii); %generate 2d Kr for plotting purposes
Kxx(jj,ii) = Kx(jj); %generate 2d Kx for plotting purposes
end
end
smf = exp(j*phi_mf); %%%%%%%%%%%%
%note, we are in the Kx and Kr domain, thus our convention is S_mf(Kx,Kr)
%appsly matched filter to S
S_mf = S.*smf;
clear smf phi_mf;Luego está la interpolación de stolt, que es la transformación de stolt en ppt
%perform the Stolt interpolation
%FOR DATA ANALYSIS
kstart =73;
kstop = 108.5;
%kstart = 95;
%kstop = 102;
Ky_even = linspace(kstart, kstop, 1024); %create evenly spaced Ky for interp for real data
clear Ky S_St;
%for ii = 1:size(Kx,2)
count = 0;
for ii = 1:zpad;
%for ii = round(.2*zpad):round((1-.2)*zpad)
count = count + 1;
Ky(count,:) = sqrt(Kr.^2 - Kx(ii)^2);
%S_st(ii,:) = (interp1(Ky(ii,:), S_mf(ii,:), Ky_even)).*H;
S_st(count,:) = (interp1(Ky(count,:), S_mf(ii,:), Ky_even));
end
S_st(find(isnan(S_st))) = 1E-30; %set all Nan values to 0
clear S_mf ii Ky;El último es IFFT, que es IDFT 2D en ppt
%perform the inverse FFT's
%new notation: v(x,y), where x is crossrange
%first in the range dimmension
clear v Kr Krr Kxx Ky_even;
v = ifft2(S_st,(size(S_st,1)*4),(size(S_st,2)*4));
Entonces puedes dibujar
bw = 3E8*(kstop-kstart)/(4*pi);
max_range = (3E8*size(S_st,2)/(2*bw))*1/.3048;
figure(figcount);
S_image = v; %edited to scale range to d^3/2
S_image = fliplr(rot90(S_image));
cr1 = -80; %(ft)
cr2 = 80; %(ft)
dr1 = 1 + Rs/.3048; %(ft)
dr2 = 350 + Rs/.3048; %(ft)
%data truncation
dr_index1 = round((dr1/max_range)*size(S_image,1));
dr_index2 = round((dr2/max_range)*size(S_image,1));
cr_index1 = round(( (cr1+zpad*delta_x/(2*.3048)) /(zpad*delta_x/.3048))*size(S_image,2));
cr_index2 = round(( (cr2+zpad*delta_x/(2*.3048))/(zpad*delta_x/.3048))*size(S_image,2));
trunc_image = S_image(dr_index1:dr_index2,cr_index1:cr_index2);
downrange = linspace(-1*dr1,-1*dr2, size(trunc_image,1)) + Rs/.3048;
crossrange = linspace(cr1, cr2, size(trunc_image, 2));
%scale down range columns by range^(3/2), delete to make like
%dissertation again
clear ii;
for ii = 1:size(trunc_image,2)
trunc_image(:,ii) = (trunc_image(:,ii)').*(abs(downrange*.3048)).^(3/2);
end
trunc_image = dbv(trunc_image); %added to scale to d^3/2
imagesc(crossrange, downrange, trunc_image, [max(max(trunc_image))-40, max(max(trunc_image))-0]);
colormap('default');
title('Final Image');
ylabel('Downrange (ft)');
xlabel('Crossrange (ft)');
axis equal;
cbar = colorbar;
set(get(cbar, 'Title'), 'String', 'dB','fontsize',13);
print(gcf, '-djpeg100', 'final_image.jpg');
if close_as_you_go == 1,
close(figcount);
end
figcount = figcount + 1;
Además de la matriz de calibración, se pueden combinar básicamente los códigos ppt y matlab.
En cuanto a por qué la imagen sar se puede obtener de acuerdo con los módulos en el ppt, y cómo el código específico de matlab implementa los algoritmos de esos módulos, debe leer el artículo detenidamente para saberlo.
Hay un artículo correspondiente a la traducción china de ppt
http://www.360doc.com/content/14/1125/00/12979957_427822702.shtml
