Segment and crop photos to JPEG files.

I am a total newbie at this and am attempting to use Matlab R2019b Trial Version to segment images from JPEG files. Here is the situation: I have numerous JPEG "photo of photos" files - in other words, someone scanned a whole bunch of photo album pages that each contain 3 (or more) photos on each to JPEG files (one photo album page per file). What I want is an automated cropping procedure whereby the 3 (or more) photos on the photo album page are segmented out and ultimately stored as 3 separate JPEGs. The photo album page has a matte black background; most photos have a white border. One additional challenge: sometimes the 3 photos overlap each other a bit. I have gone through the segmentation tutorials and have successfully created masks but I don't know how apply the mask to the original photo (need to do??) and to then save each cropped photo to a separate JPEG file. I have attached a sample starting JPEG.

[KALYAN ACHARJYA]

Ohh it may possible, it will take time to test and then answered, let's see.

Answer 1

[Image Analyst]

In your loop over all images, call a function called ExtractImages(). In that function you need to do something like

function outputImageCellArray = ExtractImages(rgbImage)
% Convert to grayscale with rgb2gray()
grayImage = rgb2gray(rgbImage);
% Threshold at some brightness just above the gray background you are using
mask = grayImage > 128;
% Call bwareaopen() or bwareafilt() to keep only blobs bigger than a certain size.
mask = bwareafilt(mask, [10000, inf]);
% Fill any holes.
mask = imfill(mask, 'holes');
% Call regionprops() and get the bounding boxes.
props = regionprops(mask, 'BoundingBox');
% Loop over all bounding boxes extracting the images.
for k = 1 : length(props)
    thisBB = props(k).BoundingBox;
    outputImageCellArray{k} = imcrop(rgbImage, thisBB);
end

[Glenn Schmidt]

Thank you ImageAnalyst! This got me on my way!