#Remove noise from grayscale images using image processing.

1 messages · Page 1 of 1 (latest)

leaden sedge
#

Hi Guys,   I hope you are all doing well.
I need help, please; I am so stuck 😦
I want to remove noise from grayscale images using image processing.
I have 2D elevation roof plans that I need to extract the following from: (outer line of the roof, inner lines of the roof).
I used the "Sobel operator" to do so because I found it the most genarizable way within different types of patterns.

my problem: 
I am facing a lot of noise that I need to remove. "Anything regarding the main outer (red) and inner lines (blue) has to be removed."
What is the best approach to take that would also generalize all of the patterns of the roofs?

timber dagger
#

I'm kinda also working on the same problem of defining building edge detection but its from aerial view like satellites or UAVs. What researchers did to have the noise removed while keeping the edges preserved there were either image pre-processing or using computationally heavy models Attention models. I found this paper right now though which surveys the different pre-processing used... https://vciba.springeropen.com/articles/10.1186/s42492-019-0016-7
^ I'll recommend going to Fig. (2,3,4) and directly seeing results of their survey.
And if any suggestions are there for this problem I'd like to hear them as well

leaden sedge
#

@timber dagger
I have checked the figures you recommended, but this is making the images more clear, as I understand.
I need to remove anything except the blue and red lines here.

craggy tide
leaden sedge
valid cedar