3D construction by 2D images from single camera from Structure from Motion

1 次查看（过去 30 天）

Vinitha Patil 2020-4-7

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/515965-3d-construction-by-2d-images-from-single-camera-from-structure-from-motion

评论： Vinitha Patil 2020-4-7

I am construction 3D sturucre of object from single camera images. I used redmi note pro 5 mobile camera ,its image size is 3000*4000. Its rgb. I took several photos of chess board pattern calibrated using calibration code of single camera from Matlab. I got Interensic parameters and and Exentrisic parameters of camera. I used camera parameters for 3D costruction. I used 3D costruction by Stucture From Motion algorithm from 2 Images used in matlab suport. When I use my images using

I1 = undistortImage(I1, cameraParams);
I2 = undistortImage(I2, cameraParams);

I am getting very much more bent images

last I am getting only few points in 3D map.

please let me know what is problem. Is there any solution for this

2 个评论
显示无隐藏无

Sai Bhargav Avula 2020-4-7

can you attach your output image?

Vinitha Patil 2020-4-7

在 MATLAB Online 中打开

I think I loaded camera parameter from vision toolbox directory.

I used

load upToScaleReconstructionCameraParameters.mat

from calibration code I got camera parameters in parmas. but I don't know how to load it in 3D costruction code.If

I used

load parmas

it is giving error

please hep me out for solving this problem

i attached calibration code

images = imageDatastore('C:\Users\HP\Documents\MATLAB\vinihtha\calibra');
imageFileNames = images.Files;
[imagePoints, boardSize] = detectCheckerboardPoints(imageFileNames);
squareSize = 39; % millimeters
worldPoints = generateCheckerboardPoints(boardSize, squareSize);
I = readimage(images, 1); 
imageSize = [size(I, 1), size(I, 2)];
[params, ~, estimationErrors] = estimateCameraParameters(imagePoints, worldPoints, ...
                                     'ImageSize', imageSize);figure; 
showExtrinsics(params, 'CameraCentric');
figure; 
showExtrinsics(params, 'PatternCentric');
figure; 
showReprojectionErrors(params);displayErrors(estimationErrors, params);