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。

相關引導路徑對極幾何基本矩陣
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。

  1. 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
  2. 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
  3. 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
  4. 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
  5. 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
  6. 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
  7. 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
  8. 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