How to find the valley areas of the image?

How to calculate the depth of the letters in the given image, so that i can find the valley areas ? Please help me.

Answer 1

[Image Analyst]

In your duplicate post I referred you to Vision Bob where there are dozens of papers on this

11.3.3 Specific 3-D Models, Rock Art, Petroglyphs, Rock Structures, Caves

22.4.11 Cultural Heritage Sites, Modeling, Analysis, Large Scale Models

19.6.3.12 Geological Analysis, Rocks

You could also try Google:

https://www.google.com/search?biw=1599&bih=1036&ei=mgUeXvOML--v_QbQ97vQCQ&q=rock+stone+carving+etchings+engraving++image+analysis&oq=rock+stone+carving+etchings+engraving++image+analysis&gs_l=psy-ab.3...15799.17653..18205...0.2..0.102.951.10j1......0....1..gws-wiz.......0i71j33i10.Ez8aKnBh6m0&ved=0ahUKEwjzwZPI2YPnAhXvV98KHdD7Dpo4ChDh1QMICg&uact=5

Answer 2

[KALYAN ACHARJYA]

The segmention approach for this image is already answerd, for localizing valley region please see the watershed algorithm. It would be great if you follow the Gonzalez Book (Digital Image Processing using MATLAB), where detail explanation is discussed with matlab code. Please note that here you can get the area of valley or basin, not the depth.

rgb=imread('rgb.jpg');
BW1=im2bw(rgb2gray(rgb));
BW2=bwareaopen(~BW1,60);
%% This Part @Image Analyst
windowSize=5;
kernel=ones(windowSize)/windowSize^2;
blurryImage=conv2(single(BW2),kernel,'same');
BW3=blurryImage>0.5;
%% Morpho Operation 
se=strel('disk',2);
BW3=imclose(BW3,se);
rgb(~BW3)=0; % Mask Apply
imshow(rgb);

[Sk]

Thanks sir. But the above code is not segmenting the letters. If i execute it for the given image, the entire image is in blue colour.

[KALYAN ACHARJYA]

See the BW3 Mask, and do needful

[Sk]

Sir, after making the needed changes, the output is like this:

[KALYAN ACHARJYA]

My result is different, the code has been already provided. What are you represengments by valley areas?? Is those letter areas, s shown in segmented image (Previously Attached)?

[Sk]

@Kalyan Acharjya, For the image I have given in the first, I need to get the exact shape of the letters alone separately without the background. So I'm representing the letters as valley areas

[KALYAN ACHARJYA]

Have you seen the mask (or ~mask), which is the letters only?

[Image Analyst]

Your second image does not have unform exposure so a global threshold cannot be used. Try adapthisteq() before binarizing.

[Sk]

@ Kalyan Acharjya, Yes sir i have tried in the mask part. And that is the output i posted recently.

[Sk]

@Image Analyst, when i use adapthisteq() and also i tried adaptthresh() , the output is a full black image.

[Image Analyst]

adapthisteq() will not give a black image. You're either using it incorrectly or displaying it incorrectlly.

[Sk]

@Image Analyst, The output after applying adapthisteq() and binarization step is black. And my final output is given below:

[KALYAN ACHARJYA]

May be it can slightly improved by preprocessing & appropriate morpho operation.

[Sk]

Fine sir. I will try.