How to crop image with nonlinear cropping shape?

Dear all,

I have following image:

And would like to crop it as follows:

Here is the code I'm using right now is:

    clc;close all;clear all;
    corn=imread('After_phi_o-phi-f.jpg');
    mm_corn=imerode(corn,strel('disk',1));
    bw_corn=im2bw(mm_corn, graythresh(mm_corn));

    cc_corn = bwconncomp(bw_corn);                                                       
    aba_corn = [cellfun(@numel,cc_corn.PixelIdxList)];                                                                          
    [mv_corn,ind] = sort(aba_corn,'descend');
    L_corn=labelmatrix(cc_corn);                                                                                                   
    ki_corn = find(aba_corn >= mv_corn(2));       
    mbi_corn = ismember(L_corn, ki_corn);                                                                                           
    bw_corn(~mbi_corn) = 0;                                                                                                      
    Ibw = imfill(bw_corn,'holes');
    Ilabel = bwlabel(Ibw);
    stat = regionprops(Ilabel,'centroid');
    imshow(bwconvhull(im2bw(corn, graythresh(corn)))); hold on;

    plot([stat(1).Centroid(1),stat(2).Centroid(1)], [stat(1).Centroid(2),stat(2).Centroid(2)], 'r');1)

And result is:

So I want cut information from the left until the red line, how I can do it?

Thanks for any help.

[Adam]

Why can't you use imcrop?

[Ivan Shorokhov]

@Adam, because imcrop will crop it with standard rectangular shape, but I want a little bit nonlinear cropping shape, for example as such:

                         ______
                        /______|

[Image Analyst]

Images MUST be rectangular. Period.

[Ivan Shorokhov]

I don't want to change shape of the image, I want to create a mask with such shape.

Answer 1

[Image Analyst]

Get the horizontal and vertical profiles using sum(), and get the top and bottom row and left and right column using find()

horizontalProfile = sum(binaryImage, 1);
verticalProfile = sum(binaryImage, 2);
topRow = find(verticalProfile, 1, 'first');
bottomRow = find(verticalProfile, 1, 'last');
leftColumn= find(horizontalProfile, 1, 'first');
rightColumn = find(horizontalProfile, 1, 'last');
croppedImage = binaryImage(topRow:bottomRow, leftColumn:rightColumn);

[Ivan Shorokhov]

@Image Analyst, you showed an alternative way of cropping image, BUT I have asked my question incorrectly. So my actual question is: how to make nonlinear cropping shape, for example as such:

          ______
         /______|

So I will get result will looks like this:

Original:

Cropped:

[Image Analyst]

Is the line always red? And does it never extend to the boundaries of the blob? If so I'd suggest you extract the red line, then find the endpoints of it with bwmorph(), then draw a black line in the image using imline(), demo attached. Then extract the biggest blob, demo attached.