Canny Thresholds in Different Scenes
One threshold misses weak scratch edges, while another admits grain.
Canny Thresholds in Different Scenes explained visually
One threshold misses weak scratch edges, while another admits grain. English captions are optional and start off. The complete transcript follows the written lesson.
What you will learn
One threshold misses weak scratch edges, while another admits grain.
A practical question
On one part, the scratch fades into a shadow. Set an edge threshold high enough to reject surface grain, and the shadowed half vanishes. Set it low enough to keep the faint half, and isolated specks appear everywhere. Canny edge detection offers a useful way to handle that trade-off.
The useful idea
First smooth the image, measure the gradients, and thin their responses. Then use two thresholds. Strong responses above the high threshold start candidate edges. Weaker responses between the two thresholds can remain when they connect to a strong edge. Isolated weak responses usually drop out. That connectivity can keep a faded continuation without admitting every weak spot.
Work through the example
Lay the original scratch and edge result side by side. Does the faint segment actually join the stronger one? Could a noisy bridge create a connection that looks persuasive but is false? Try a clean part under the same lighting too. These questions matter more than finding the most dramatic-looking edge map.
Limits and next step
Save the smoothing scale and both thresholds. A setting selected on one exposure may fail on the next. Before tuning further, we'll inspect a different operation that can make boundaries more visible while sometimes creating false outlines.
Worked example
Retain a weak edge connected to a strong edge while rejecting an isolated weak speck.
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
Canny detection smooths, measures gradients, thins responses, and uses two linked thresholds. Thresholds that work on one lighting condition may fail on another.
What is the main idea behind canny thresholds in different scenes?
Canny detection smooths, measures gradients, thins responses, and uses two linked thresholds.
How is it used in the example?
Retain a weak edge connected to a strong edge while rejecting an isolated weak speck.
What limitation should be checked?
Thresholds that work on one lighting condition may fail on another.
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 35 transcript195 words · Video transcript · Matches the published narration and captionsOpen transcript
On one part, the scratch fades into a shadow. Set an edge threshold high enough to reject surface grain, and the shadowed half vanishes. Set it low enough to keep the faint half, and isolated specks appear everywhere. Canny edge detection offers a useful way to handle that trade-off.
First smooth the image, measure the gradients, and thin their responses. Then use two thresholds. Strong responses above the high threshold start candidate edges. Weaker responses between the two thresholds can remain when they connect to a strong edge. Isolated weak responses usually drop out. That connectivity can keep a faded continuation without admitting every weak spot.
Lay the original scratch and edge result side by side. Does the faint segment actually join the stronger one? Could a noisy bridge create a connection that looks persuasive but is false? Try a clean part under the same lighting too. These questions matter more than finding the most dramatic-looking edge map.
Save the smoothing scale and both thresholds. A setting selected on one exposure may fail on the next. Before tuning further, we'll inspect a different operation that can make boundaries more visible while sometimes creating false outlines.