How to select one area from several in a logical array (mask)?

I have a 3D logical array where is several spatial areas marked as "true" or "1". 1st dimension is latitude, 2nd – longitude, 3d – depth. I need to choose only one area which contains a preassigned point in space. This point is always inside one of the areas. As a result, I should get the similar logical array, which has only one area with values "1".

I think the first problem is to find the border for each area. I tried to use "alphaShape" for it but I’m not sure that I actually need it. Besides, I had a problem with it because my areas stick together into one. I have added a data file in the hope that it will make my question clearer.

shp = alphaShape(lon3(mask), lat3(mask), z3(mask), 1);
plot(shp)

% lat3, lon3, z3 - 3d array of latitudes, longitudes and depths
% mask - logical 3d array with several spatial areas
% core_lat, core_lon, core_z - coordinates of preassigned point (red cross
% in picture) (in data)

An example of data is shown in the figure below. The figure above shows a vertical section along the latitude of 69.63 N, the figure below shows the distribution at the depth of 417 meters. Yellow is several areas, from which I need to select one. Red cross is preassigned point.

Thank you in advance!

[KSSV]

You want depth for the given point at every level is it?

[Elena Novoselova]

I use a fixed depth (about 417 m) because the area I need always exists at that depth. Latitude and longitude change over time.

[Elena Novoselova]

Perhaps I didn't understand your question. I need to leave only one yellow area (inside which there is a red cross).

Answer 1

[Adam Danz]

The demo uses an image from the Image Processing Toolbox. You can replace the mask variable with your binary matrix and define the target with your (x,y) target value.

Generate inputs

• mask is a 2D binary matrix (true|false values)
• target is the (x,y) coordinates of the target that lies within one of the objects in the image.

RGB = imread('pillsetc.png'); % image processing toolbox
mask = imbinarize(rgb2gray(RGB));
target = [430,170];

Find object closest to the target

1. The closest true pixel in the image to the target is identified using pdist2.
2. Each group of true values in the binary image is assigned a group number using bwlabel.
3. The group number that belongs to the group closest to the target is stored in targetGroup.
4. All other groups of true values in the binary matrix are removed.

[rownum, colnum] = find(mask); 
[~,I] = pdist2([colnum,rownum],target,'cityblock','smallest',1);

maskLbl = bwlabel(mask);
targetGroup = maskLbl(rownum(I),colnum(I));

maskClean = mask;
maskClean(maskLbl ~= targetGroup) = false;

Plot results

The original image and the cleaned image are plotted along with the target location (x) and the closest true pixel (o).

figure()
tiledlayout(1,2)
nexttile
imagesc(mask)
colormap(parula(2))
hold on
plot(target(1), target(2), 'rx', 'DisplayName', 'target')
plot(colnum(I),rownum(I), 'ro', 'DisplayName', 'Nearest true value')
title('full binary image')
axis tight
axis equal

nexttile
imagesc(maskClean)
colormap(parula(2))
hold on
plot(target(1), target(2), 'rx', 'DisplayName', 'target')
plot(colnum(I),rownum(I), 'ro', 'DisplayName', 'Nearest true value')
title('trimmed binary image')
axis tight
axis equal
lg = legend();
lg.Layout.Tile = 'south';

[Elena Novoselova]

Thank you so much! This works great!

I adopted it for 3D mask using bwlabeln instead of bwlabel.

[Adam Danz]

Glad I could help!