System Information

C++ user
OpenCV versions, 4.6.0 and 4.11.0
Operating System: both Ubuntu 22.04 and iOS 18.4.

Detailed description

Hi,

I think there might be a regression with findFundamentalMat and using USAC_MAGSAC. I noticed a difference in the resulting Fundamental matrix, and worse reconstruction results when using 4.11.0.

Is there a difference in settings? Here is my command:

	cv::Mat mask;
	cv::Mat F_cv = cv::findFundamentalMat(
		points1,
		points2,
		cv::USAC_MAGSAC, //USAC_MAGSAC //USAC_FM_8PTS
		3.0,
		0.99,
		mask);

I've attached my points to test (pt1.x pt1.y pt2.x pt2.y).

In 4.11, my F is:

F_cv: [-4.319846175372044e-08, 7.319864684475541e-07, 0.001450549004465776;
-4.082099655921459e-07, -1.071111124551967e-07, -0.0003954902886679395;
-0.001770586681786347, 0.0003432840369838192, -0.09341279152094734]

In 4.6, my F is:

F_cv: [1.904422812968894e-08, 1.048170617221457e-06, 0.0006680824655863789;
-7.927984975206173e-07, -1.368874670413949e-07, -0.0002958165791529331;
-0.00108963192812404, 0.0003136612755694536, -0.08914227831028487]

points.txt

Thanks in advance!

Steps to reproduce

Run findFundamentalMat using points here:

points.txt

Issue submission checklist

• I report the issue, it's not a question
• I checked the problem with documentation, FAQ, open issues, forum.opencv.org, Stack Overflow, etc and have not found any solution
• I updated to the latest OpenCV version and the issue is still there
• There is reproducer code and related data files (videos, images, onnx, etc)