引導式測驗 · 中等 等級
對極幾何和基本矩陣測驗
了解極線、基本矩陣的約束、魯棒匹配如何使用它,以及為什麼它不單獨提供深度。
問題 1 的 8
What does epipolar geometry constrain for a point seen by two cameras?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- What does epipolar geometry constrain for a point seen by two cameras?
- It must share the same pixel coordinates in both images
- It lies at the epipole in the second image
- It is the nearest descriptor match anywhere in the second image
- Its matching point in the other image lies along a corresponding epipolar line
- What does applying the fundamental matrix to an image point produce?
- An epipolar line in the other image
- A 3D point in world coordinates
- A camera intrinsic matrix
- A disparity value for the point
- Which relation should corresponding homogeneous points satisfy?
- x′ = Fx, as if F were a point-to-point transform
- xᵀx′ = 0 for every pair of image points
- x′ᵀFx = 0
- Fx = 0 for any corresponding point
- Does the fundamental matrix require known camera intrinsics?
- Yes; F is defined only after metric depth is known
- No; it relates uncalibrated image coordinates
- Yes; F requires camera intrinsics and translation in metric units
- No; F directly gives each point’s 3D position
- During robust fundamental-matrix estimation, how can RANSAC help?
- Find the most visually similar descriptors without rejecting outliers
- Estimate focal length from image brightness
- Smooth the images before feature detection
- It can fit the relation while rejecting inconsistent match outliers
- What does a fundamental matrix alone provide?
- Exact depth for every pixel
- A complete textured 3D model
- A geometric correspondence constraint, not depth by itself
- The camera’s absolute global position
- Why can a nearly planar scene cause difficulty for two-view geometry estimation?
- Its correspondences may not constrain general 3D motion robustly
- It makes lens distortion disappear without calibration
- A plane prevents feature detection entirely
- All camera intrinsics become identical for a plane
- After estimating a fundamental matrix, what does stereo rectification aim to do?
- Equalize image brightness before stereo matching
- Aligning epipolar lines to simplify correspondence search along scanlines
- Remove lens distortion without camera calibration
- Guarantee metric depth in textureless regions