How can i get the second image using opencv or image processing

asked 2016-08-02 06:07:27 -0500

shek gravatar image

image description from this image which operation should i do to get image description

edit retag flag offensive close merge delete


the 1st image looks like a color-mapped visualization of 3d data. you sure want to work with the original data, not the latter.

berak gravatar imageberak ( 2016-08-02 06:43:45 -0500 )edit

what should i do should i go with watershed algorithm

shek gravatar imageshek ( 2016-08-02 08:14:41 -0500 )edit

watershed is a good idea, you just need a "seed" point for each region

berak gravatar imageberak ( 2016-08-02 08:36:03 -0500 )edit

how can i do it can you point me some examples

shek gravatar imageshek ( 2016-08-02 08:56:13 -0500 )edit

can you show us, how the original data looked like ? not a false-color screenshot with a text overlay (which is pretty useless, imho.)

watershed sample

berak gravatar imageberak ( 2016-08-02 08:59:04 -0500 )edit
shek gravatar imageshek ( 2016-08-02 09:11:49 -0500 )edit

It looks like you should map it so that the colors are intensity values, then blur, and find local maximums. Use those as your watershed seed points.

Tetragramm gravatar imageTetragramm ( 2016-08-02 21:10:22 -0500 )edit

How about giving us the publication from which you are ripping the idea from? I am pretty convinced that it simply states how they done it ...

StevenPuttemans gravatar imageStevenPuttemans ( 2016-08-03 04:48:20 -0500 )edit

this is the link but its not in there

shek gravatar imageshek ( 2016-08-03 08:35:17 -0500 )edit

sorry, but after a quick glance at the paper, i have to downvote your question.

(did you even read it ?) they're deriving their segmentation from 3d laser scanner, not a 2d false-color img.

so, your question is quite hypothetical (it would not ever been done in the way you asked), and you're just wasting anyones (including your own) time here.

berak gravatar imageberak ( 2016-08-03 08:47:55 -0500 )edit