this error and thanks please ? about sfm_3d_pipeline HOT 15 OPEN

@Mahran55, could you please give me more details? Mine is running fine

from sfm_3d_pipeline.

clear all
close all
addpath(genpath(fileparts(which(mfilename))))

%% Program arguments
dataset = 'bb'; % images initials
calib = 'calib_12';
cleanSparse = 1;% no clean = 0
dense = 1;

% select for dense cloud
no_sample = length(dir(['images/' dataset '*.png']));
no_choice = 12; % how many images for making dense reconstruction
drp = fix(1:(no_sample/no_choice):no_sample); % fix, round, floor

threads = 1;
%% Reading image dataset
disp(['Running pipeline for dataset "' dataset '"'])
imsDir = dir(['ims/' dataset '*']);
disp(['Dataset contains ' num2str(length(imsDir)) ' images'])
imsNames = {imsDir.name};
imsNo = length(imsNames);
Cim = cell(1,imsNo);
for i = 1:length(imsNames)
Cim{i} = imread(imsNames{i});
% [~,Cim{i}] = createMask_wm(Cim{i});%%%% createMask for ginseng
imshow(Cim{i}); pause(0.5)
end

load(calib) % camera parameters (K)
disp('Calibration matrix: '), disp(K)
d=[0,0];
if exist('d','var')
disp('Radial distortion parameters: '), disp(d)
Cim = UndistortImages(Cim,K,d);
end
if threads ~= 1 && isempty(gcp('nocreate')), parpool(threads); end

%% Feature extraction and matching
selec = [1 0 0 1 1];%change based on %[Surf, kaze, fast, brisk, sift] combinaton is okay
% example [1 1 0 1 0] ==> combination of surf, kaze, brisk features
% selec = [0 1 0 0 1]
% selec = [1 1 0 0 1]
% selec = [0 1 1 0 0]
CfeatPts = cell(1,imsNo);
for i = 1:imsNo
disp(['Feature points estimation: image ' num2str(i) ' of ' num2str(imsNo)])
% CfeatPts{i} = EstimateFeaturePoints(Cim{i},'KAZE'); % change
CfeatPts{i} = EstimateFeaturePoints_lalit(Cim{i},selec);
end

Ccorrs = cell(1,imsNo-1);
CfeatPtsPlusOne = CfeatPts(2:end);

parfor i = 1:imsNo-1
disp(['Feature matching: pair ' num2str(i) ' of ' num2str(imsNo-1)])
mtds = {'Exhaustive',...%mtds{1,1}
'Approximate'};%mtds{1,2}
tr = 0.45;% range 0--1
Ccorrs{i} = MatchFeaturePoints(CfeatPts{i}, CfeatPtsPlusOne{i},mtds{1,2},tr);
end

%% Correspondence filtering (outlier rejection from matched features)
CcorrsIn = cell(1,imsNo-1);
CF = cell(1,imsNo-1);
parfor i = 1:imsNo-1
disp(['Correspondence filtering: pair ' num2str(i) ' of ' num2str(imsNo-1)])
[CF{i}, Cinliers] = RANSAC(num2cell(Ccorrs{i},1),...
@EstimateFundamentalMatrix, 8, @SampsonDistance, 8);
CcorrsIn{i} = cell2mat(Cinliers);
end

%% Fundamental Matrix estimation (unnormalized to normalized image using SVD)
parfor i = 1:imsNo-1
disp(['Fundamental Matrix estimation: pair ' num2str(i) ' of ' num2str(imsNo-1)])
CF(i) = EstimateFundamentalMatrix(num2cell(CcorrsIn{i},1)); % CF is the fundamental matrix
end

%% Background Filtering
CcorrsInFil = cell(1,imsNo-1);
parfor i = 1:imsNo-1
disp(['Background Filtering: pair ' num2str(i) ' of ' num2str(imsNo-1)])
CcorrsInFil{i} = FilterBackgroundFromCorrs(KCANONICAL_POSE,...
KEstimateRealPose(K'*CF{i}*K, NormalizeCorrs(CcorrsIn{i},K)), CcorrsIn{i});
CF(i) = EstimateFundamentalMatrix(num2cell(CcorrsInFil{i},1));
end

%% Essential Matrix from Fundamental Matrix
CE = cell(1,imsNo-1); % CE is the essential matrix
CcorrsNormInFil = cell(1,imsNo-1);
for i = 1:imsNo-1
disp(['Essential Matrix from Fundamental Matrix: '...
num2str(i) ' of ' num2str(imsNo-1)])
CE{i} = K'*CF{i}*K/norm(K'*CF{i}*K)*sqrt(2)/2;
CcorrsNormInFil{i} = NormalizeCorrs(CcorrsInFil{i}, K);
end

%% Structure from Motion
disp('Pose estimation: default first pair')
CP = cell(1,imsNo);
CP{1} = CANONICAL_POSE;
CP{2} = EstimateRealPose(CE{1}, CcorrsNormInFil{1});
LOCALBA_OCCUR_PER1 = 8;
LOCALBA_OCCUR_PER2 = 10;

for i = 2:imsNo-1
disp(['Pose estimation: ' num2str(i+1) ' of ' num2str(imsNo)])
CP{i+1} = EstimateRealPose(CE{i}, CcorrsNormInFil{i}, CP{i});
TrackedCorrs = CascadeTrack({CcorrsNormInFil{i-1}, CcorrsNormInFil{i}});
TrackedCorrsIso = IsolateTransitiveCorrs(TrackedCorrs);
disp(['Transitivity: ' num2str(size(TrackedCorrsIso,2))])
CP{i+1} = OptimizeTranslationBaseline(CP{i-1}, CP{i}, CP{i+1}, TrackedCorrsIso);

if ~mod(i-1, LOCALBA_OCCUR_PER1-2)
    disp('Local Bundle Adjustment 1...')
    disp(['Refine ' num2str(i-(LOCALBA_OCCUR_PER1-2)) '->' num2str(i+1)])
    C = IsolateTransitiveCorrs(CascadeTrack(...
        CcorrsNormInFil(i-(LOCALBA_OCCUR_PER1-2) : i)), 'displaySize');
    if size(C,2) > 1
        CPBA = BundleAdjustment(CP(i-(LOCALBA_OCCUR_PER1-2) : i+1), C);
        CP(i-(LOCALBA_OCCUR_PER1-2) : i+1) = CPBA;
    end
end
if ~mod(i-1, LOCALBA_OCCUR_PER2-2)
    disp('Local Bundle Adjustment 2...')
    disp(['Refine ' num2str(i-(LOCALBA_OCCUR_PER2-2)) '->' num2str(i+1)])
    C = IsolateTransitiveCorrs(CascadeTrack(...
        CcorrsNormInFil(i-(LOCALBA_OCCUR_PER2-2) : i)), 'displaySize');
    if size(C,2) > 1
        CPBA = BundleAdjustment(CP(i-(LOCALBA_OCCUR_PER2-2) : i+1), C);
        CP(i-(LOCALBA_OCCUR_PER2-2) : i+1) = CPBA;
    end
end

end

if imsNo > LOCALBA_OCCUR_PER1
disp(['Refine ' num2str((imsNo-1)-(LOCALBA_OCCUR_PER1-2)) '->' num2str(imsNo)])
C = IsolateTransitiveCorrs(CascadeTrack(...
CcorrsNormInFil((imsNo-1)-(LOCALBA_OCCUR_PER1-2) : imsNo-1)), 'displaySize');
if size(C,2) > 1
CPBA = BundleAdjustment(CP((imsNo-1)-(LOCALBA_OCCUR_PER1-2) : imsNo), C);
CP((imsNo-1)-(LOCALBA_OCCUR_PER1-2) : imsNo) = CPBA;
end
end
if imsNo > LOCALBA_OCCUR_PER2
disp(['Refine ' num2str((imsNo-1)-(LOCALBA_OCCUR_PER2-2)) '->' num2str(imsNo)])
C = IsolateTransitiveCorrs(CascadeTrack(...
CcorrsNormInFil((imsNo-1)-(LOCALBA_OCCUR_PER2-2) : imsNo-1)), 'displaySize');
if size(C,2) > 1
CPBA = BundleAdjustment(CP((imsNo-1)-(LOCALBA_OCCUR_PER2-2) : imsNo), C);
CP((imsNo-1)-(LOCALBA_OCCUR_PER2-2) : imsNo) = CPBA;
end
end

Tracks = CascadeTrack(CcorrsNormInFil);
X = TriangulateCascade(CP,Tracks);
C = SampleColors(Tracks,Cim,K);

% denoise sparse ptcd
if cleanSparse ==1
ptcd_sparse = pointCloud(X','Color', C');
minDistance = 0.03;% 0.03
[labels,numClusters] = pcsegdist(ptcd_sparse,minDistance);%
cluster_No = 1;
den_ptcd_sparse = selectPointColud (ptcd_sparse,labels,numClusters,cluster_No);
X = den_ptcd_sparse.Location';
C = den_ptcd_sparse.Color';
end
%% write sparse
MakePly([dataset '-sparse.ply'], X, [], C)
PlotSparse(CP,X,C);

%% Dense Matching
CX = cell(1,length(drp));
CC = cell(1,length(drp));
CXSc = cell(1,length(drp));
CCSc = cell(1,length(drp));
% Parallel groundwork
CimDrp = Cim(drp);
CimDrpPlusOne = Cim(drp+1);
CcorrsNormInFilDrp = CcorrsNormInFil(drp);
CFDrp = CF(drp);
CPDrp = CP(drp);
CPDrpPlusOne = CP(drp+1);

zz = [-10.6,... %mu
4,... %window
7,... %zone gap
0.9,... %pm_tau
0.7]; %tau

for i = 1:length(drp)
disp(['Dense Reconstruction: pair ' num2str(i) ' of ' num2str(length(drp))])
zRange = ExtractZRange(CANONICAL_POSE, DehomogenizeMat(...
HomogenizeMat(CPDrpPlusOne{i})/HomogenizeMat(CPDrp{i})),...
CcorrsNormInFilDrp{i});
[CX{i}, CC{i}, CXSc{i}, CCSc{i}] = RectifyAndDenseTriangulate(zz,CropBackground(...
CimDrp{i},Unnormalize(CcorrsNormInFilDrp{i}(1:2,:),K),1.1),...
CropBackground(CimDrpPlusOne{i},Unnormalize(CcorrsNormInFilDrp{i}(3:4,:),...
K),1.1), CFDrp{i}, K, CPDrp{i}, CPDrpPlusOne{i}, zRange,'denoise'); %,'denoise'
end

MakePly([dataset '-dense.ply'], cell2mat(CX), [], cell2mat(CC))
PlotDense(cell2mat(CX),cell2mat(CC))
%
%
% %% Remeshing
CXFil = cell(1,length(drp));
CCFil = cell(1,length(drp));
CNFil = cell(1,length(drp));
for i = 1:length(drp)
disp(['Computing normals: set ' num2str(i) ' of ' num2str(length(drp))])
[CNFil{i}, filInd] = ComputeNormalsAndFilter(CXSc{i});
CXFil{i} = CX{i}(:,filInd);
CCFil{i} = CC{i}(:,filInd);
end
%
disp('Remeshing')
MakePly([dataset '-colored.ply'],...
cell2mat(CXFil), cell2mat(CNFil), cell2mat(CCFil))

ptcd = pcread([dataset,'-colored.ply']);
figure(2), pcshow(ptcd)
delete(['out/',dataset,'-colored.ply']);
%
%
minDistance = .2;% 0.03
[labels,numClusters] = pcsegdist(ptcd,minDistance);%
cluster_No = 3;
ptcd_selec = selectPointColud (ptcd,labels,numClusters,cluster_No);

pcwrite(ptcd_selec,['out/',dataset '-colored.ply'])

PoissonReconWrapper(['out/',dataset '-colored.ply'], ['out/', dataset '-cleaned.ply'], 8, feature('numcores'));%8

ptcd = pcread('ffa-cleaned.ply');%% preview
pcshow(ptcd)
%
%
% % fclose('all');
%
% % %% volume computation
% % ptcd = pcread('r-cleaned.ply');%% preview
% % pcshow(ptcd)
% %
% %
% % % dt = delaunayTriangulation(double(ptcd.Location));
% % dt = delaunayTriangulation(double(ptcd.Location));
% % [ch, Volume] = convexHull(dt);
% % X = ch(:,1);
% % Y = ch(:,2);
% % Z = ch(:,3);
% %
% % trisurf(ch, X, Y, Z);
% %
% % [normals,curvature] = findPointNormals(ptcd.Location);
% %
% % pc = ptcd;
% % pc.select('UniformSampling', 2);
% % pc.plot('Color', 'r', 'MarkerSize', 5);
% % pc.plotNormals('Color', 'y', 'Scale', 10);
% %
% % % trisurf(ch, dt.Points(:,1),dt.Points(:,2),dt.Points(:,3), 'FaceColor', 'cyan')

this is the main code

from sfm_3d_pipeline.

you need more informations ?

from sfm_3d_pipeline.

Yeah, this is the main code.
Did you do calibration of your camera?

from sfm_3d_pipeline.

yeah i did it so i do a new calibration ?

from sfm_3d_pipeline.

Okay great, could you take for me a screenshot of your current working directory in matlab?

from sfm_3d_pipeline.

Or you could send me your sample data so that I could try it myself - I want to understand the real problem

from sfm_3d_pipeline.

from sfm_3d_pipeline.

i send you the image in dataset and the 36 others images ?

from sfm_3d_pipeline.

if you like can i send you all the folder to your mail ?

from sfm_3d_pipeline.

Send me a google drive link or wetransfer link

from sfm_3d_pipeline.

ok just a few minute and i will send you . thanks .

from sfm_3d_pipeline.

files are large :(

from sfm_3d_pipeline.

Please try to use wetransfer.com, it can allow transfer of files up to 2GB

from sfm_3d_pipeline.

from sfm_3d_pipeline.