Computer Vision Lecture Notes

These flashcards cover key concepts in camera models, calibration methods, homography, and distortion corrections from the Computer Vision lecture.

What is the main purpose of camera calibration?

To estimate intrinsic and extrinsic parameters for accurate 3D reconstruction.

What are weak-perspective projections typically used under?

When the object is not too close to the camera and the change in depth is small compared to the camera-object distance.

What is a homography in the context of camera models?

A projective transformation between two planes that preserves lines.

What is the difference between perspective and weak-perspective projection?

Perspective projection considers depth variations while weak-perspective projection assumes a uniform depth.

What is required to estimate a projection matrix P in camera calibration?

At least six correspondences between 3D scene points and image points.

What type of distortion correction is essential for accurate 3D reconstruction?

Radial distortion correction.

What does the projection matrix P consist of?

P = KR[I|−t], where K is the intrinsic matrix, R is the rotation matrix, and t is the translation vector.

How do you compute intrinsic parameters after obtaining the homography matrix?

By using linear equations that relate the elements of the calibration matrix to the homography.

What are the benefits of using chessboard patterns for camera calibration?

They allow for easy corner detection and can handle non-perspective distortion.

What is the primary challenge when dealing with noisy data in camera calibration?

Identifying and correcting outliers in the correspondences.

What is the main purpose of camera calibration?

To estimate intrinsic and extrinsic parameters for accurate 3D reconstruction.

What are weak-perspective projections typically used under?

When the object is not too close to the camera and the change in depth is small compared to the camera-object distance.

What is a homography in the context of camera models?

A projective transformation between two planes that preserves lines.

What is the difference between perspective and weak-perspective projection?

Perspective projection considers depth variations while weak-perspective projection assumes a uniform depth.

What is required to estimate a projection matrix P in camera calibration?

At least six correspondences between 3D scene points and image points.

What type of distortion correction is essential for accurate 3D reconstruction?

Radial distortion correction.

What does the projection matrix P consist of?

P = KR[I\|−t], where K is the intrinsic matrix, R is the rotation matrix, and t is the translation vector.

How do you compute intrinsic parameters after obtaining the homography matrix?

By using linear equations that relate the elements of the calibration matrix to the homography.

What are the benefits of using chessboard patterns for camera calibration?

They allow for easy corner detection and can handle non-perspective distortion.

What is the primary challenge when dealing with noisy data in camera calibration?

Identifying and correcting outliers in the correspondences.

What do intrinsic parameters describe?

The internal characteristics of the camera, such as focal length, optical center, and skew.

What do extrinsic parameters describe?

The camera's position and orientation (rotation and translation) relative to a world coordinate system.

What are the components of the intrinsic matrix K?

Focal lengths (fx, fy), principal point (cx, cy), and optionally a skew coefficient \gamma.

Why is efficient and accurate corner detection a key benefit of chessboard patterns?

Their distinct alternating dark and light squares provide high-contrast points that are easy to locate precisely in an image.

What is the role of [I\|−t] in the projection matrix P = KR[I\|−t]?

It represents the transformation from world coordinates to camera coordinates, before scaling and rotation by K and R, respectively.