Computer Vision Foundations · free preview

Interpolation in Practice

Nearest, bilinear, bicubic and Lanczos interpolation compared on real magnification, and the limits of upsampling as prediction.

1. Nearest neighbor and its staircase

Nearest neighbor and its staircase

2. Bilinear, bicubic and Lanczos

Bilinear, bicubic and Lanczos

3. Upsampling adds pixels, not information

Upsampling adds pixels, not information

4. Choosing a kernel

Choosing a kernel

5. Kernel anatomy

Which statement correctly distinguishes nearest-neighbor and bilinear interpolation when magnifying an image by a factor of two?

6. What magnification provides

A 32×3232 \times 32 image is upsampled to 256×256256 \times 256 with a high-quality Lanczos kernel. What has been added to the image's information content?

7. Bilinear midpoint

A one-dimensional image has sample values 2 and 6 at adjacent integer positions. Bilinear interpolation evaluates the value exactly halfway between them. Compute that midpoint value.

Practice this interactively

Free account · instant grading · spaced review that schedules itself.

Start this course for free