Edge detection

Question

0 个投票

I'm trying to use matlab operators for edge detection,could anyone suggest me the best suited operators(i.e sobel,roberts,prewitt etc) for the operation?

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Geoff 2012-3-23

It depends on what your images are like and what you'll do with the edge images after. Why don't you try them all and choose what you think is most relevant?

请先登录，再进行评论。

请先登录，再回答此问题。

Follow Question

Answer 1

Image Analyst 2012-3-26

0 个投票

The best one is the one you like best. There are thousands of edge operators, each with different parameters so not even a single one will give you the same looking images (it depends on the parameters). There really is no one single edge operator that is best for all images. If there were, then there would only be that one and everyone would use it, but there are thousands and new ones are being invented weekly with no end in sight. That's why, since there is no one best one, we can only say that the best one is the one that is best for you with your specific images.

You may not even need an edge detector. I've seen plenty of instances where someone wanted to measure something and then thought they needed an edge detector when that was an unnecessary operation or a totally wrong approach. Maybe you do, maybe you don't. That's why it's good to get a second opinion by showing (posting) your image to us. http://www.mathworks.com/matlabcentral/answers/7924-where-can-i-upload-images-and-files-for-use-on-matlab-answers

3 个评论
显示 1更早的评论隐藏 1更早的评论

Image Analyst 2012-3-27

Please supply more guidance on what you want to detect and what other images have. Do all images have hand-drawn chalk letters? Do all images have the bright/brown right edge? Should that right edge be ignored or detected? I would think that if you want the "T" or other bright white letters on the black chalkboard that you can simply do thresholding. Then, once you have the binary image of where the "T" pixels are, then what? Do you want to do OCR on them, calculate the area, calculate the brightness, or something else? You need to say what you want to get out of this image.