Sampling a Pattern Without Aliasing
Fine threads in the woven sample turn into wide false bands.
Sampling a Pattern Without Aliasing explained visually
Fine threads in the woven sample turn into wide false bands. English captions are optional and start off. The complete transcript follows the written lesson.
What you will learn
Fine threads in the woven sample turn into wide false bands.
A practical question
The fabric looks striped in person, but the camera image shows broad bands that aren't there. This can happen when the weave repeats faster than our pixel grid can represent. The camera samples a continuous pattern at separate locations. When several different fine patterns give similar samples, the recorded image can show a false, slower pattern. We call that aliasing.
The useful idea
Adding pixels can help if the lens and sensor still resolve the real threads. But simply saving a larger file from the same samples won't change what was measured. When reducing an existing image, the reliable order is to suppress details that the smaller grid cannot hold and then take the smaller sample grid. A low-pass filter is often part of that preparation.
Work through the example
Try a dense stripe photograph. Shrink one copy directly, and filter another copy before shrinking. Compare the resulting bands with the original. The filtered copy may look softer, yet it can be more faithful because it avoids inventing a large false stripe. The filter has a cost: genuinely fine detail may also disappear.
Limits and next step
For our fabric inspection, decide what thread size the task must distinguish before choosing the capture and resize settings. Once fine information is lost at sampling, interpolation can't measure it again. That's what we'll test next.
Worked example
Filter a dense stripe pattern before reducing the sampled image to half width.
Quick check: test your understanding
Try it yourself: In the situation described above, explain the method and one reason its output might be misleading.
Check the reasoning
Sampling records a continuous pattern on a discrete grid; detail above the grid's usable frequency can alias. A false band cannot be fixed just by making the saved image larger afterward.
What is the main idea behind sampling a pattern without aliasing?
Sampling records a continuous pattern on a discrete grid; detail above the grid's usable frequency can alias.
How is it used in the example?
Filter a dense stripe pattern before reducing the sampled image to half width.
What limitation should be checked?
A false band cannot be fixed just by making the saved image larger afterward.
Further reading
Digital Image Processing, 4th edition — Rafael C. Gonzalez and Richard E. Woods.
Original ToolsInVision teaching examples. Editorial approach · Suggest a correction
Accessible referenceLesson 18 transcript212 words · Video transcript · Matches the published narration and captionsOpen transcript
The fabric looks striped in person, but the camera image shows broad bands that aren't there. This can happen when the weave repeats faster than our pixel grid can represent. The camera samples a continuous pattern at separate locations. When several different fine patterns give similar samples, the recorded image can show a false, slower pattern. We call that aliasing.
Adding pixels can help if the lens and sensor still resolve the real threads. But simply saving a larger file from the same samples won't change what was measured. When reducing an existing image, the reliable order is to suppress details that the smaller grid cannot hold and then take the smaller sample grid. A low-pass filter is often part of that preparation.
Try a dense stripe photograph. Shrink one copy directly, and filter another copy before shrinking. Compare the resulting bands with the original. The filtered copy may look softer, yet it can be more faithful because it avoids inventing a large false stripe. The filter has a cost: genuinely fine detail may also disappear.
For our fabric inspection, decide what thread size the task must distinguish before choosing the capture and resize settings. Once fine information is lost at sampling, interpolation can't measure it again. That's what we'll test next.