Plotting a contour plot on top of an image?

Hi, I want to plot a contour plot on top of an image, but I cannot manage to do so. I want to overlay following images:

My code is as:

summation_front3=flipud(summation_front2);                              
figure(3);
imshow(avg_background_front);
hold on
fig = contourf(summation_front3);
set(gca,'YTick',[],'XTick',[])
colormap(jet);
hold off

The plot I receive looks like this:

I want the 'blue area' in the contour plot, which is zeros, to be transparent. Is someone able to help me, thanks?

Answer 1

[Abel Babu]

Hi,

As of R2014b there is no direct way of doing the same. In case you are using any version prior do refer to this post: https://in.mathworks.com/matlabcentral/answers/60106-how-to-make-one-contour-transparent-in-contourf

For versions after R2014b, this post discusses a workaround:

https://in.mathworks.com/matlabcentral/answers/250279-how-to-make-contour-plots-transparent-in-matlab-r2015a#answer_211204

This workaround involves manually defining the area that is required to be transparent. In your use case, to define the area, you can maybe check for location in the contour plot which is zero and then use the fill function with the corresponding x and y values.

Abel

Answer 2

[Olivier Haas]

This shows different ways to overlay a contour on an image

%%  plotting contour from edges detected
% obtain a binary image 
I = imread('rice.png');
BW = imbinarize(I);
% find the edges - easy for a binary image
Iedge = edge(BW,'Roberts');
% the edge may be different depending on the algorithm and may also be
% different to that found using bwboundaries (latter is perfect), BUT it is the edge contour
% found by the edge detector you may want to evaluate.

% plot the contours of the edges
figure
imshow(I)
hold on
[X,Y]=find(Iedge==1);
plot(Y,X,'c.') % note Y and X swapped
contour(Iedge,1,'k--'); %you get 2 contours on either side of the edge
contour(BW,1,'m-.'); % you get 1 contour outside the white blob
% Using  mathematical morphology - from other post
[B, L] = bwboundaries(BW,'noholes');
for k = 1:length(B)
    boundary = B{k};plot(boundary(:,2), boundary(:,1), 'r:', 'LineWidth', 2)
end
legend('edge > find > plot','edge > contour','BW > contour','bwboundaries')
title('Overlayed images')

Answer 3

[Image Analyst]

Look into labeloverlay and visboundaries

help labeloverlay

 LABELOVERLAY Overlay label matrix regions on a 2-D image.
    B = LABELOVERLAY(A,L) fills the input image with a different solid
    color for each label in the label matrix L. L must be a valid MxN label matrix
    that agrees with the size of A.
 
    B = LABELOVERLAY(A,BW) fills the input image with a solid color where
    BW is true. BW must be a valid mask that agrees with the size of A.
 
    B = LABELOVERLAY(A,C) fills in the input image with a different solid
    color for each label specified by the categorical matrix C.
 
    B = LABELOVERLAY(___,NAME,VALUE) computes the fused overlay image B
    using NAME/VALUE parameters to control aspects of the computation.
 
    Parameters include:
 
    'Colormap'              Mx3 colormap where M is the number of labels in
                            the label matrix L or binary mask BW. RGB
                            triplets in each row of the colormap must be
                            normalized to the range [0,1]. A string or
                            character vector corresponding to one of the
                            valid inputs to the MATLAB colormap function is
                            also accepted, in which case a permuted form of
                            the specified colormap suitable for labeled
                            region visualization will be used.
 
                            Default: 'jet'
 
    'IncludedLabels'        Scalar or vector of integer values in the range
                            [0,max(L(:))] that specify the labels that will
                            be falsecolored and blended with the input
                            image. When a categorical, C, is provided as
                            the specification of the labeled regions,
                            'IncludedLabels' can also be a vector of
                            strings corresponding to labels in C.
 
                            Default: 1:length(L(:))
 
    'Transparency'          Scalar numeric value in the range [0,1] that
                            controls the blending of the label matrix with
                            the original input image A. A value of 1.0
                            makes the label matrix coloring completely
                            transparent. A value of 0.0 makes the label
                            matrix coloring completely opaque.
 
                            Default: 0.5
 
    Class Support
    -------------
    The input image A is of type uint8, uint16, single, double, or int16.
    The input label matrix L is a numeric matrix. B is an RGB image of type
    uint8.
 
    Example 1 - Visualize over-segmentation of RGB data
    ---------
     A = imread('kobi.png');
     [L,N] = superpixels(A,20);
     figure
     imshow(labeloverlay(A,L));
 
    Example 2 - Visualize binary-segmentation of greyscale image
    ---------
    A = imread('coins.png');
    t = graythresh(A);
    BW = imbinarize(A,t);
    figure
    imshow(labeloverlay(A,BW))
 
    Example 3 - Visualize segmentation specified as categorical array
    ---------
    A = imread('coins.png');
    t = graythresh(A);
    BW = imbinarize(A,t);
    stringArray = repmat("table",size(BW));
    stringArray(BW) = "coin";
    categoricalSegmentation = categorical(stringArray);
    figure
    imshow(labeloverlay(A,categoricalSegmentation,'IncludedLabels',"coin"));
 
    See also superpixels, imoverlay

    Documentation for labeloverlay
       doc labeloverlay

help visboundaries

 VISBOUNDARIES Plot region boundaries.
    VISBOUNDARIES(BW) draws boundaries of regions in the binary image BW on
    the current axes. BW is a 2D binary image where pixels that are logical
    true belong to the foreground region and pixels that are logical false
    constitute the background. VISBOUNDARIES uses BWBOUNDARIES to find the
    boundary pixel locations in the image.
 
    VISBOUNDARIES(B) draws region boundaries specified by B, where B is a
    cell array containing the boundary pixel locations of the regions,
    similar in structure to the first output from BWBOUNDARIES (see
    function help for BWBOUNDARIES). Each cell contains a Q-by-2 matrix,
    where Q is the number of boundary pixels for the corresponding region.
    Each row of these Q-by-2 matrices contains the row and column
    coordinates of a boundary pixel.
 
    VISBOUNDARIES(AX, ___)  draws region boundaries on the axes specified
    by AX.
    
    H = VISBOUNDARIES(___) returns a handle to an hggroup object for the
    boundaries. The hggroup object, H, is the child of the axes object, AX.
 
    H = VISBOUNDARIES(___,NAME,VALUE,...) passes the name-value pair
    arguments to specify additional properties of the boundaries. Parameter
    names can be abbreviated.
 
    'Color'                
        Specifies the color of the boundary using a 1x3 RGB triplet which 
        must be normalized to the range [0,1]. A string or character vector 
        corresponding to one of the valid inputs to the MATLAB colormap 
        function is also accepted. For example, valid specifications for a 
        red fill color are: 'red','r', and [1 0 0]. Default color is red.
 
    'LineStyle'
        Specifies the line style for the boundary. The line style can be
        any of the following.
        {-} | -- | : | -. | none
        
    'LineWidth'
        Size in points
        Width of the boundary. Specify this value in points. 1 point = 1/72
        inch. The default value is 2 points.
  
    'EnhanceVisibility'
        Specifies whether or not to augment the drawn boundary with
        contrasting features to improve visibility on a varying background.
        Setting the value to true draws an augmented boundary and setting
        it to false does not. Default value is true.
              
    Example 1
    ---------
    % This example computes the boundaries of a binary images and plots it on
    % the image.
 
     BW = imread('blobs.png');
     B = bwboundaries(BW);
     imshow(BW)
     hold on
     visboundaries(B)
       
    Example 2
    ---------
    % This example shows how to visualize the result of segmentation of an
    % image using VISBOUNDARIES.
 
     I = imread('toyobjects.png');
     imshow(I)
     hold on    
  
     % Segment the image using active contour
     % First, specify initial contour location close to the object that is     
     % to be segmented.    
     mask = false(size(I));
     mask(50:150,40:170) = true;
  
     % Display the initial contour on the original image in blue.
     visboundaries(mask,'Color','b');
  
     % Segment the image using the 'edge' method and 200 iterations
     bw = activecontour(I, mask, 200, 'edge');
     
     % Display the final contour on the original image in red.
     visboundaries(bw,'Color','r');
     title('Red - Initial Contour, Blue - Final Contour');
  
  See also BWBOUNDARIES, BWPERIM, BWTRACEBOUNDARY, VISCIRCLES.

    Documentation for visboundaries
       doc visboundaries

Answer 4

[DGM]

Assuming that the overlay is an actual contour() object:

From @Abel Babu's links, we find this answer from Will Grant, which should work in modern versions.

Here are two examples, one employing Will's technique of manipulating undocumented properties, and another using a reverse stacking technique. The first should work here.

As of R2022b, contour()/contourf() objects now support FaceAlpha and EdgeAlpha properties.

Assuming that the overlay is an image, then this example might be of use, but that's mainly aimed at handling unfilled contours.

If I were to assume that the task is to combine the two given images, and that their extents should correspond to each other, then:

% read the images
BG = imread('thing.png');
FG = imread('contour.png');

% figure capture should be avoided whenever possible
% it will tend to add unpredictable extraneous padding
FG = FG(1:end-1,:,:);
% and it will not easily preserve aspect ratio or size
FG = imresize(FG,[size(BG,1) size(BG,2)],'nearest');

% alpha as a scalar
% imfuse(...,'blend') can also do this, but only for 50%
alpha = 0.5;
outpict = im2uint8(alpha.*im2double(FG) + (1-alpha).*im2double(BG));

imshow(outpict)

% alpha as a map
alpha = abs(double(FG) - permute([1 1 143],[1 3 2])) > 3;
outpict = im2uint8(alpha.*im2double(FG) + (1-alpha).*im2double(BG));

imshow(outpict)

I imagine that this latter composition is what's intended, though the concepts may be combined.