how can i open .out extension file in matlab ?

Question

ba 2014-8-4

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/146193-how-can-i-open-out-extension-file-in-matlab

评论： dpb 2014-8-12

采纳的回答： dpb

I'm trying to open a bundle.out file in matlab. Does anyone has any idea of how to do so ?

6 个评论
显示 4更早的评论隐藏 4更早的评论

ba 2014-8-4

编辑：ba 2014-8-4

I'm currently working with bundler for bundle adjustment. I have the bundle.out file and points001.ply files as output. I need to open the bundle.out file in matlab and locate the position of the cameras and points in 3D. The format of .out file is :

269 31913

3.0107383211e+003 1.2457159171e-001 -3.2061110722e-001

6.3846423609e-001 -2.0702958325e-002 7.6937299585e-001

-3.4926013391e-001 8.8299277113e-001 3.1359388546e-001

-6.8584311478e-001 -4.6892979608e-001 5.5652858709e-001

8.8146166784e+001 1.0107379995e+001 2.3539905552e+001

4.7501507199e-001 2.3681082673e+001 -9.9221710766e+001

43 78 72

4 80 3252 1270.5100 90.1400 81 3258 1233.3401 64.4800 76 3183 1233.0701

85.0601 65 1896

which follows the following convention :

num_cameras num_points [two integers]

f k1 k2 [the focal length, followed by two radial distortion coeffs]

R [a 3x3 matrix representing the camera rotation]

t [a 3-vector describing the camera translation]

position [a 3-vector describing the 3D position of the point]

color [a 3-vector describing the RGB color of the point]

list [a list of views the point is visible in]

I need to produce a 3D reconstruction of cameras and a rough point-based scene geometry as output in matlab.

dpb 2014-8-9

编辑：dpb 2014-8-9

Looks like a link, acts like a link, is a link...excepting no target given. Don't do that in future, please... most confusing at first, then irritating when discover what poster has done.

You can use bold or italic (or both) to emphasize and/or identify items--I tend to use bold for Matlab functions and italics for variables in my postings but that's just my personal convention, no Forum protocol specifically.

ba 2014-8-11

Okay! Will take care of this from next time.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

dpb 2014-8-8

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/146193-how-can-i-open-out-extension-file-in-matlab#answer_147899

在 MATLAB Online 中打开

line no. 4 to 6 is a 3x3 matrix. how can i read it?

Create a dummy file w/ a 3x3 array...

>> dlmwrite('ba.txt',rand(3)*10,'precision','%12.8e','delimiter',' ')
>> type ba.txt
5.67821641e+00 5.30797553e+00 1.29906208e+00
7.58542896e-01 7.79167230e+00 5.68823661e+00
5.39501187e-01 9.34010684e+00 4.69390641e+00

Read it back...

>> fid=fopen('ba.txt');
>> x=fscanf(fid,'%f',[3,3]).'
x =
  5.6782    5.3080    1.2991
  0.7585    7.7917    5.6882
  0.5395    9.3401    4.6939
>> fid=fclose(fid);
>>

NB: the .' element transpose to order by row since Matlab fills in column order on reading the nine values. You can, of course, also simply read 9 values and reshape and transpose...

As outlined in initial answer, use a combination of textscan and/or fscanf and friends that operate on an open file handle with the rosetta stone of how the file is structured to return the various pieces in sequence using things like your magic number of 5*NumCameras and knowing it's a 3x3 array where appropriate.

It's not really terribly difficult, but is as (I believe it was IA?) mentioned, drudgery initially to work through and get going. Once done, however, if done properly, you should be able to parse any file of the given structure easily.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

ba 2014-8-11

Thanks!

请先登录，再进行评论。

Answer 2

dpb 2014-8-4

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/146193-how-can-i-open-out-extension-file-in-matlab#answer_147365

编辑：dpb 2014-8-9

在 MATLAB Online 中打开

num_cameras num_points [two integers]
f k1 k2 [the focal length, followed by two radial distortion coeffs]
R [a 3x3 matrix representing the camera rotation]
t [a 3-vector describing the camera translation]
position [a 3-vector describing the 3D position of the point]

Well, reading in the camera and position points doesn't appear too difficult; but there are some uncertainties left --

1) I presume perhaps there are num_cameras datasets in the file, maybe?

2) How does one determine the length of this list array?

fid=fopen('yourfile');
n=fscanf(fid,'%d %d');  % the two number values--camera, points-->[n(1) n(2)]
t=cell2mat(textscan(fid,'%f %f %f','headerlines',2));   % camera position???
p=cell2mat(textscan(fid,'%f %f %f'));                   % point position

Now the question is as above, what need to get to the next group. In general one builds a small snippet for each file section like the above and then place that in a loop for either a counted number of times or while(~feof(fid)) if length is unknown.

13 个评论
显示 11更早的评论隐藏 11更早的评论

ba 2014-8-5

编辑：ba 2014-8-5

在 MATLAB Online 中打开

function header = readOutHeader(~,F)

            token = fscanf(F,'%s',[1 1]);
            if strcmpi(token,'OUT')
                header.type = 'OUT';
            else
                WARNING('Unsupported filetype');
                header = NaN;
            end
        end
       function cameraHeader = readOutCameraHeader(~,F)
            cameraHeader.ncams = fscanf(F,'%d', [1 1]);
end
            function cameras = readOutCameras(~,F,header)
            cameras = cell(header.ncams,1);
            for i = 1:header.ncams
                camera.focal_length = fscanf(F,'%f', [1 1]);
                camera.radial_distortion1 = fscanf(F,'%f', [1 1]);
                camera.radial_distortion2 = fscanf(F,'%f', [1 1]);
                camera.rotation = fscanf(F,'%f\n', [1 9]);
                camera.translation = fscanf(F,'%f\n', [1 3]);
                fscanf(F,'%f', [1 1]);
                cameras{i} = camera;
            end
        end
        function [features,points ] = readOutPoints(~,F,descriptors)
            positions = fscanf(F,'%d', [1 3]);
            color = fscanf(F,'%d\n', [1 3]);
            for i = 1:positions
                points(i,:) = fscanf(F,'%f',[1 3]);
                [~] = fscanf(F,'%d',[1 3]); 
                nproj = fscanf(F,'%d',[1 1]);
                for j = 1:nproj
                    camera_idx = fscanf(F,'%d',[1 1]) + 1;
                    feature_idx = fscanf(F,'%d',[1 1]) + 1;
                    c_features =  descriptors{camera_idx};
                    c_feature = single(c_features(feature_idx,:));
                    features(i,:) = features(i,:)  +  c_feature;   
                    [~] = fscanf(F, '%f',[1 2]);
                end
                features(i,:) = features(i,:) / nproj;
            end
        end
        function model = readOutModel(obj,filename,datadir)
            model = {};
            if exist(filename,'file')
                F = fopen(filename,'r');
                header = obj.readOutHeader(F);
            end
        end
        function model = loadOutModel(obj,filename,datadir) 
            model ={};
            if exist(filename,'file')
                F = fopen(filename,'r');
                header = obj.readOutHeader(F);
                c = 0;
                while(true)
                    c = c+1;
                    cameraHeader = obj.readOutCameraHeader(F);
                    if cameraHeader.ncams>0
                        cameras = obj.readOutCameras(F,cameraHeader);
                        descriptors = obj.readOutDescriptors(datadir,cameras, cameraHeader);
                        [features,points ] = obj.readOutPoints(F,descriptors);
                        tmpmodel = Model();
                        tmpmodel.setCameras(cameras);
                        tmpmodel.setFeatures(features);
                        tmpmodel.setPoints(points);
                        model{numel(model)+1} = tmpmodel;
                    else
                        break;
                    end
                end
            else
               WARNING(['Invalid model file name : ' filename]);
            end
        end
        function saveMatModel(~,model,filename) %#ok<INUSL>
            save(filename,'model');
        end
        function model=loadMatModel(~,filename)
            load(filename);
        end
    end