views:

53

answers:

3

Please suggest me some methods on thick horizontal line detection in images. Any papers, libraries etc please.

A: 

OpenCV is a good library. You may use a Hough transform to detect lines.

Note that the answer may depend on the quality of your images, some methods can be more effective that other ones.

Chubas
A: 

If the line images are really easy to detect (say you can find the line from ten yards away), just scale the image from (width, height) to (1, height) and find the y-coordinate of the darkest pixel. Use CV_INTER_AREA when resizing.

rwong
The idea is really good, to resize image so that the line will not be thick any more, but you are talking a little bit about different thing.Why resizing is made until (1, height) ?There can be many thick lines in image.
erjik
Resizing to (1, height) is equivalent to computing a single slice of the Hough transform at zero degree for the detection of strictly horizontal lines. Hough transform should be used with a carefully chosen set/range of angles for slightly-off horizontal lines.
rwong
Seeing erjik's new comment, I realize my answer is totally irrelevant. Please disregard. I posetd a new answer.
rwong
+1  A: 

Binary threshold the image, then use a combination of morphological operations to detect "borders that are thicker then N1 but thinner than N2" to produce several images and combine (logically AND) them to selectively detect borders of a certain expected thickness.

rwong
do you mean opening and closing operations?
erjik