Matlab - How to fill part of an image using imfill?

Question

I have an image like this:

img

and I wanted to fill the bottom part using a mask like this as the location of seed points:

mask

This is what I tried but didn't work:

img = imread('img.jpg'); maskImg = imread('mask.png');
[Gmag, Gdir] = imgradient(img,'sobel');
mask = imbinarize(maskImg);
[rows,columns] = find(mask);
bw = imfill(Gmag, [rows(:) columns(:)]);
figure, imshow(bw);

The error message is:

Function IMFILL expected input number 2, CONN, to be a valid connectivity specifier. A nonscalar connectivity specifier must be 3-by-3-by- ... -by-3.

How can I fill the road part of the image above?

Answer 1

Not great and not highly reproducible for a variety of images but maybe something interesting can be pulled from this mess of rudimentary image processing techniques.

Image = imread("img.jpg");
Mask = imread("mask.png");
Contrast_Stretched_Image = imadjust(Image,stretchlim(Image),[]);
Max_Filtered_Image = ordfilt2(Contrast_Stretched_Image,1,ones(9,9));
Filtered_Image = medfilt2(Max_Filtered_Image, [15 15]);
Filtered_Image = Filtered_Image > 200;
[Boundaries,Boundary_Image] = bwboundaries(Filtered_Image, 'noholes');
Largest_Boundary = Boundaries(1);
Largest_Boundary = Largest_Boundary{1,1};

imshow(Boundary_Image)
[Image_Height,Image_Width] = size(Image);
Binary_Image = zeros(Image_Height,Image_Width);
for Boundary_Index = 1: length(Largest_Boundary)

    X = Largest_Boundary(Boundary_Index,1);
    Y = Largest_Boundary(Boundary_Index,2);
    Binary_Image(X,Y) = 255;

end

Filled_Binary_Image = logical(imfill(Binary_Image,'holes'));
imshow(Filled_Binary_Image);
Image_Fill = uint8(~Filled_Binary_Image).*Image;
imshow(Image_Fill);
Mask_Fill = uint8(Filled_Binary_Image).*Mask;
imshow(Image_Fill+Mask_Fill);

The colour of the black unmasked portions can be changed prior to or afterwards. Complementary statements can be used to change the white and black regions.

Four Types of Combinations

Image = imread("img.jpg");
Mask = imread("Mask.png");
Mask = (Mask ~= 0);
Image(Mask) = 255;
imshow(Image);

White Portion Filled

Image = imread("img.jpg");
Mask = imread("Mask.png");
Mask = uint8(Mask > 0);
Masked_Image = Image.*Mask;
imshow(Masked_Image);

Black Portion Filled

Image = imread("img.jpg");
Mask = imread("Mask.png");
Mask = uint8(Mask == 0);
Masked_Image_2 = Image.*Mask;
imshow(Masked_Image_2);

Image = imread("img.jpg");
Mask = imread("Mask.png");
Mask = (Mask == 0);
Image(Mask) = 255;
imshow(Image);

Using MATLAB version: R2019b