Matlab Source Code: Unsupervised Change Detection from SAR Images via Non-Local Mean Filter and Hyperbolic Tangent Sigmoid Function

Hts_f.m: Main code for change detection goruntu_kalite.m , imageQualityIndex.m and kappaindex.m codes are for accuarcy asessment In Matlab: %Suggested Windows size (sws in code) for non local means filter (11 for Bern dataset, 55 for Ottawa dataset, 73 for Yellow River dataset, and 127 for Farmland dataset);

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

[change,All_Errors]=hts_f(img1,img2,gt,sws)

----Inputs: “img1” and “img2” are SAR images, “gt” is ground truth image, “sws” is search window size for non-local means filtering of image ---Outputs: “change” is change map and “All_Errors” is error values computed using ground truth

---

For Ottawa Application In Matlab

[1] load('Ottawa_Dataset.mat')

[2] [change,All_Errors]=hts_f(img1,img2,gt,55)

---

For Bern Application In Matlab

[1] load('Bern_Dataset.mat')

[2] [change,All_Errors]=hts_f(img1,img2,gt,11)

---

For Yellow River Application

[1] load('Yellow_River_Dataset.mat')

[2] [change,All_Errors]=hts_f(img1,img2,gt,73)

---

For Farmland Application (Not included in our article)

[1] load(FarmC.mat')

[2] [change,All_Errors]=hts_f(img1,img2,gt,127)