引導式測驗 · 中等 等級
RGB-D 相機與深度感測器測驗
查看 RGB-D 對齊、立體和主動深度感測、相機內在特性、無效像素、同步和 3D 映射檢查。
問題 1 的 8
A color image and depth image come from separate sensors. What links their pixels correctly?
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。
- A color image and depth image come from separate sensors. What links their pixels correctly?
- Matching image resolutions alone
- A shared filename and frame index alone
- Calibrated intrinsics and an extrinsic transform between the sensors
- Only a color white-balance correction
- How does a stereo depth camera primarily estimate distance?
- From emitted-light travel time
- From disparity between calibrated views
- From distortion of a projected pattern
- From one RGB view without paired geometry
- Why can far-away stereo depth be sensitive to small matching errors?
- Disparity is small at distance, so an error has a larger depth effect
- Depth error must shrink when disparity becomes small
- A small disparity error changes only image color
- Stereo depth is independent of baseline
- A depth pixel is flagged invalid on shiny packaging. Which action best preserves measurement honesty?
- Set missing values to zero and call them measured
- Interpolate every invalid pixel and label it direct sensor output
- Discard the validity mask after color alignment
- Keep it marked missing or uncertain instead of treating it as measured depth
- What does the term RGB-D tell you about a camera system?
- It provides color plus depth data, without specifying one sensing mechanism
- It necessarily uses time-of-flight hardware
- It necessarily uses two stereo imagers
- It derives true range from RGB color alone
- Why can unsynchronized color and depth streams misalign a moving hand?
- Their physical baseline changes with hand motion
- Synchronization is irrelevant once intrinsics are known
- All depth pixels become invalid whenever a hand moves
- They may describe different poses of the hand at different capture times
- What did KinectFusion demonstrate beyond one depth snapshot?
- That one depth image is already a complete 3D map
- Tracking and fusing moving depth-camera frames into an indoor surface map
- That RGB alone supplies metric depth without calibration
- That moving-camera tracking can be omitted during fusion
- Which check is most useful before using an RGB-D cloud for physical measurement?
- Inspect only whether colored rendering looks smooth
- Fill every depth hole and call the values observed
- Compare dimensions and invalid-depth coverage against trusted references
- Compare file size rather than physical dimensions