Gaussian Smoothing and Scale
A small scratch disappears after the image is smoothed.
Gaussian Smoothing and Scale explained visually
A small scratch disappears after the image is smoothed. English captions are optional and start off. The complete transcript follows the written lesson.
What you will learn
A small scratch disappears after the image is smoothed.
A practical question
We smooth the inspection image to reduce grain, and the narrow scratch disappears. That is not automatically a success. Gaussian smoothing gives more weight to nearby pixels and less to distant ones. Its spread controls the scale over which values are blended. A wider spread removes more rapid variation, including noise and fine defects.
The useful idea
Imagine two copies of the same metal part. Apply a mild blur to one and a wider blur to the other. Compare the scratch with the original at the same size. If the mild version quiets grain while keeping the mark visible, it may help this inspection. If the wider version erases the mark, a cleaner-looking background has cost us the very evidence we needed.
Work through the example
The right spread depends on the size of the scratch in pixels and on the noise. We should not choose it using only an easy part or a smooth background. Test small real defects and clean parts separately, and note whether subsequent edge or threshold steps change too.
Limits and next step
Save the original and the smoothing settings so another inspector can repeat the comparison. Next, we'll look at a different neighborhood filter that handles isolated bright specks without using a weighted average.
Worked example
Compare a narrow blur that retains the scratch with a wider blur that erases it.
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
Gaussian smoothing averages neighbors with distance-dependent weights, with scale set by its spread. Removing grain also removes some real fine detail.
What is the main idea behind gaussian smoothing and scale?
Gaussian smoothing averages neighbors with distance-dependent weights, with scale set by its spread.
How is it used in the example?
Compare a narrow blur that retains the scratch with a wider blur that erases it.
What limitation should be checked?
Removing grain also removes some real fine detail.
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 31 transcript199 words · Video transcript · Matches the published narration and captionsOpen transcript
We smooth the inspection image to reduce grain, and the narrow scratch disappears. That is not automatically a success. Gaussian smoothing gives more weight to nearby pixels and less to distant ones. Its spread controls the scale over which values are blended. A wider spread removes more rapid variation, including noise and fine defects.
Imagine two copies of the same metal part. Apply a mild blur to one and a wider blur to the other. Compare the scratch with the original at the same size. If the mild version quiets grain while keeping the mark visible, it may help this inspection. If the wider version erases the mark, a cleaner-looking background has cost us the very evidence we needed.
The right spread depends on the size of the scratch in pixels and on the noise. We should not choose it using only an easy part or a smooth background. Test small real defects and clean parts separately, and note whether subsequent edge or threshold steps change too.
Save the original and the smoothing settings so another inspector can repeat the comparison. Next, we'll look at a different neighborhood filter that handles isolated bright specks without using a weighted average.