CRE Sensors, Vision & Perception Flashcards
6 cards from real CRE practice questions. Tap to flip, then mark Knew It or Still Learning — missed cards come back until you master them.
Read the first 6 CRE Sensors, Vision & Perception flashcards as text
Camera calibration in robot vision primarily determines:
Answer: Intrinsic parameters such as focal length and lens distortion coefficients
Calibration estimates intrinsic parameters (focal length, principal point, distortion coefficients) that map pixel coordinates to accurate 3D measurements.
Stereo vision systems estimate scene depth by:
Answer: Computing the disparity between matching points in two horizontally offset cameras
Stereo vision triangulates 3D depth from the horizontal pixel offset (disparity) of corresponding features viewed by two cameras with a known baseline.
In robot vision, a homography matrix describes:
Answer: A projective transformation relating corresponding points on two planes
A homography is a 3×3 projective transformation that maps every point on one planar surface to its corresponding location in another camera view.
The RANSAC algorithm is applied in computer vision to:
Answer: Robustly estimate model parameters in the presence of data outliers
RANSAC iteratively fits models to minimal random subsets of data, identifying inliers and discarding outlier measurements to find a robust solution.
Point cloud data from 3D depth sensors is typically processed using:
Answer: The Point Cloud Library (PCL)
PCL provides specialized algorithms for filtering, segmentation, surface reconstruction, and feature extraction tailored to 3D point cloud data.
Convolutional Neural Networks (CNNs) are primarily applied in robot vision for:
Answer: Image classification and object detection tasks
CNNs learn hierarchical feature representations from images and are widely used to classify scenes and detect or localize objects in robotic perception pipelines.