High-Pass Filters and Noise
A high-pass view makes a pale stripe obvious but also magnifies grain.
Read the full lesson and open its transcript below. A narrated video will be added after production.
What you will learn
A high-pass view makes a pale stripe obvious but also magnifies grain.
A practical question
Switch to a high-pass view of that microscope image. The faint interference stripe becomes striking. Unfortunately, so does the grain in an area that should be quiet. A high-pass filter reduces slow variation and emphasizes rapid spatial change. Noise can be rapid too.
The useful idea
Place a confirmed defect and a smooth background side by side in the original image. After filtering, check both. If their responses rise together, the dramatic display may not make the defect easier to distinguish. Also watch for negative intermediate values or clipped display values that hide how the filter actually behaved.
Work through the example
We may only need the high-pass result as one analysis channel, with the original kept for interpretation. If a simpler change in illumination or capture gives a clearer comparison, use that evidence as well. A bright response in a filtered picture is a measurement of change, not proof of a physical object. High-pass filtering can make the defect stand out because it emphasizes quick changes. The same operation can brighten sensor noise, dust, and compression edges. Check an empty part of the scan before celebrating a brighter scratch. If the supposedly empty region turns busy as well, the filter has raised background variation, not only useful evidence. Compare the processed crop with the untouched one and decide whether a human or a downstream detector can separate the defect from the extra texture. Use this case to test your understanding: Inspect both the bright defect and a quiet background after filtering. Here is the limitation to watch for: An emphasized response is not proof of a defect; noise often responds too. This supports a careful conclusion: High-pass filtering reduces slowly changing regions and emphasizes rapid spatial change.
Limits and next step
The interference here is unusually regular. Instead of suppressing every rapid change, we can target its particular repeat frequencies. That's where we'll go next.
Worked example
Inspect both the bright defect and a quiet background after filtering.
Quick check: test your understanding
Try it yourself: Consider this case: Inspect both the bright defect and a quiet background after filtering. Explain how you would check the result. In your answer, address this limitation: An emphasized response is not proof of a defect; noise often responds too.
Check the reasoning
High-pass filtering reduces slowly changing regions and emphasizes rapid spatial change. An emphasized response is not proof of a defect; noise often responds too.
What is the main idea behind high-pass filters and noise?
High-pass filtering reduces slowly changing regions and emphasizes rapid spatial change.
How is it used in the example?
Inspect both the bright defect and a quiet background after filtering.
What limitation should be checked?
An emphasized response is not proof of a defect; noise often responds too.
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 42 transcript305 words · Narration ready for the upcoming videoOpen transcript
Switch to a high-pass view of that microscope image. The faint interference stripe becomes striking. Unfortunately, so does the grain in an area that should be quiet. A high-pass filter reduces slow variation and emphasizes rapid spatial change. Noise can be rapid too.
Place a confirmed defect and a smooth background side by side in the original image. After filtering, check both. If their responses rise together, the dramatic display may not make the defect easier to distinguish. Also watch for negative intermediate values or clipped display values that hide how the filter actually behaved.
We may only need the high-pass result as one analysis channel, with the original kept for interpretation. If a simpler change in illumination or capture gives a clearer comparison, use that evidence as well. A bright response in a filtered picture is a measurement of change, not proof of a physical object. High-pass filtering can make the defect stand out because it emphasizes quick changes. The same operation can brighten sensor noise, dust, and compression edges. Check an empty part of the scan before celebrating a brighter scratch. If the supposedly empty region turns busy as well, the filter has raised background variation, not only useful evidence. Compare the processed crop with the untouched one and decide whether a human or a downstream detector can separate the defect from the extra texture. Use this case to test your understanding: Inspect both the bright defect and a quiet background after filtering. Here is the limitation to watch for: An emphasized response is not proof of a defect; noise often responds too. This supports a careful conclusion: High-pass filtering reduces slowly changing regions and emphasizes rapid spatial change.
The interference here is unusually regular. Instead of suppressing every rapid change, we can target its particular repeat frequencies. That's where we'll go next.