Computer Vision Foundations · free preview

Temporal Filters and Causality

Causal, anti-causal and non-causal temporal kernels, the recursive one-pole filter, and the price of streaming.

1. Time has an arrow

Time has an arrow

2. Truncation costs delay

Truncation costs delay

3. The recursive escape

The recursive escape

4. Stability is one condition

Stability is one condition

5. Classifying kernels

Which of these temporal impulse responses could run live on a streaming camera feed, processing frame tt using only frames up to tt?

6. Running the recursion

The filter ℓout[t]=ℓin[t]+α ℓout[t−1]\ell_{out}[t] = \ell_{in}[t] + \alpha\,\ell_{out}[t-1] starts from rest with α=0.5\alpha = 0.5 and receives a unit impulse at t=0t = 0 only. What are the first three output values?

7. Third echo of the impulse

For the recursion ℓout[t]=ℓin[t]+α ℓout[t−1]\ell_{out}[t] = \ell_{in}[t] + \alpha\,\ell_{out}[t-1] with α=0.5\alpha = 0.5 starting from rest and a unit impulse at t=0t = 0, compute the output value at t=2t = 2.

Practice this interactively

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

Start this course for free