1/24
Concise CV: 3.4
Name | Mastery | Learn | Test | Matching | Spaced | Call with Kai |
|---|
No analytics yet
Send a link to your students to track their progress
What is the first step in line detection with the Hough transform?
Use an edge detector to convert the image into a binary edge map.
Why must a line detector be robust?
Because real edge maps contain noise, so the line detector must handle imperfect data.
What is the common equation for a line in an image?
y=ax+b
What is the main idea of the original Hough transform for lines?
Map each image point to a line in parameter space, so collinear image points produce intersecting lines there
What does an intersection in Hough parameter space represent?
A line in the image.
Why does the original ababab parameter space not work well?
Because aaa and bbb are unbounded, so the parameter space is infinite.
What parameterization did Duda and Hart propose for lines?

Why is the dα parameter space better?
It is bounded, so it can be discretized into a finite accumulator array.
What range does the angular parameter α have?
[0,2π)
What is the maximum possible distance dmax in the standard Hough transform?

What does a point in the image become in dαd\alphadα Hough space?
A sinusoidal curve.
What are peaks in the Hough accumulator used for?
Detecting lines in the image.
What does the height of a counter in the accumulator represent?
How many sin/cos curves pass through that cell.
What is “cell accuracy” in the Hough transform?
The parameter precision determined by the size of the discretized Hough-space cells.
How can subcell accuracy be improved?
By using smaller bins or estimating the centroid in a neighborhood around a peak.
What shape do peak neighborhoods in Hough space often resemble?
A butterfly shape.
Why move the coordinate origin to the image center?
It halves the needed dmaxd, reduces accumulator size, and improves peak shape
What is the centered line equation in Hough space?

What happens after a line peak is detected when searching for multiple lines?
Its contribution is removed from the accumulator so the next peak can be found more easily.
How can line segment endpoints be found?
By tracing the detected line in the image and analyzing nearby edge pixels, or by working in Hough space.
How is circle detection generalized in the Hough transform?
By using a 3D parameter space with center (xc,yc) and radius r.
Why is the Hough space for circles bounded?
The image size gives an upper bound, and the radius is usually restricted to a known range.
What does a pixel become in circle Hough space?
A cone surface in 3D parameter space.
What defines a unique circle in the plane?
Three non-collinear points.
What is the key advantage of Hough transforms in general?
They detect geometric objects by accumulating evidence in a bounded parameter space and finding peaks.