Discover how 3-D curves help machines recognize objects fast.
This book explains a practical approach to matching observed curves in three dimensions to model curves. It covers how to compute the best rotation and translation that align curves, and how to use multiple curves together to identify an object reliably.
- How single-curve and multi-curve matching work, including the role of arc-length parameterization and normalization.
- Techniques for handling occluded curves and unknown starting points on the model.
- Methods to speed up calculations using center-of-mass alignment and fast Fourier transforms.
- How rigidity constraints keep the problem tractable by linking multiple curves to a single object pose.
Ideal for readers interested in computer vision, robotics, and 3-D object recognition who want a clear, implementation-focused overview.