UNDISTORT STEREO Images before RECTIFICATION ?

2024 8 22
2 个回答
回答已采纳
更新时间：2024 8 23
16 次查看（30 天）

0 个投票

This is a query reg. Stereo Rectification using Matlab inbuilt functions
Does one need to :
UNDISTORT the STEREO Images before RECTIFICATION ?
My pair of views are near-perfect, but has large disparity.
Using stereo-Calibration, I get the distortion parameters as (given below for only one of the cameras):
RadialDistortion:
[-0.163 0.0958]
TangentialDistortion:
~ [ 0, 0];
Skew: -0.600.
The 3D back-projected results appear OK, but I am looking for high accuracy. Hence the query.
Any useful advise will be beneficial (thanx in advance, for that)

请先登录，再回答此问题。

 采纳的回答

Sukhendu
Sukhendu 2024-8-23

0 个投票

Thanx a lot.
Your assumptions of functions (stereoCameraCalibrator, followed by rectifyStereoImage) being used are correct.
The disparity is so high that I cannot use disparitySGM or disparityBM functions.
Wrote my own stuff, using ORB/SIFT based correspondences.
I am truely not interested in full scene 3D reconstruction from stereo now, rather estimating 3D Coordinates of few landmark points in the images, with extremely high accuracy.

更多回答（1 个）

Abhinav Aravindan
Abhinav Aravindan 2024-8-23

0 个投票

Hi @Sukhendu
I assume you are using the “rectifyStereoImages” function for Stereo Image Rectification. In this case, it is not required to undistort your image in prior as the function automatically handles both undistortion and rectification using the stereo calibration parameters you have obtained.
https://www.mathworks.com/help/vision/ref/rectifystereoimages.html#:~:text=rectifyStereoImages(I1%2CI2%2CstereoParams)%20undistorts%20and%20rectifies%20versions%20of%20I1%20and%20I2%20input%20images%20using%20the%20stereo%20parameters%20of%20a%20stereo%20camera%20system%20stored%20in%20the%20stereoParams%20object
To improve your 3D screen reconstruction results, you may try out the following suggestions present in the below link:
https://www.mathworks.com/matlabcentral/answers/153348
I hope this answers your query!

类别

在 帮助中心 和 File Exchange 中查找有关 Process Point Clouds 的更多信息

产品

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by