How to segment cells(round) and bacterias(longitudinal) and label them ?

Question

A Feng 2022-2-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1640885-how-to-segment-cells-round-and-bacterias-longitudinal-and-label-them

回答： Shreeya 2023-11-17

I would like to segment the cells(round) and bacterias(thin lines) in the image and label them respectively in a robust manner without using any machine learning methods. I can now draw the boundaries of the cells that intersect with the border by padding the borders and thus enclosing the incomplete cells. However, I don't know how to proceed from here.

Below is my code so far:

% Read image
I2 = imread("images like the one i uploaded");
% Convert RGB to Grayscale
I2_gray = rgb2gray(I2);
% Enhance image
I2_enhanced = imadjust(I2_gray);
% Binarise Image
I2_binarised = imbinarize(I2_enhanced);
% Set up threshold-based filter to remove noise
I2_binarised = bwpropfilt(I2_binarised, 'Area', [120, 15856]); %How can i do this by algorithm over trial and error?
% Edge detection
I2_edge = edge(I2_binarised);
% Mask 4 borders 
% - pad top + left then remove
I2_padTL = padarray(I2_edge,[1 1],1,'pre');
I2_padTLfilled = imfill(I2_padTL,'holes');
I2_padTLfilled = I2_padTLfilled(2:end,2:end);
% - pad top + right then remove
I2_padTR = padarray(padarray(I2_edge,[1 0],1,'pre'),[0 1],1,'post');
I2_padTRfilled = imfill(I2_padTR,'holes');
I2_padTRfilled = I2_padTRfilled(2:end,1:end-1);
% - pad bottom + left then remove
I2_padBL = padarray(padarray(I2_edge,[1 0],1,'post'),[0 1],1,'pre');
I2_padBLfilled = imfill(I2_padBL,'holes');
I2_padBLfilled = I2_padBLfilled(1:end-1,2:end);
% - pad bottom + right then remove
I2_padBR = padarray(I2_edge,[1 1],1,'post');
I2_padBRfilled = imfill(I2_padBR,'holes');
I2_padBRfilled = I2_padBRfilled(1:end-1,1:end-1);
% - put together sub-images 
I2_filled = I2_padTLfilled | I2_padTRfilled | I2_padBLfilled | I2_padBRfilled;
% display segemented image
imshow(I2_filled),hold on;
% label shapes with color
[B, L, N]= bwboundaries(I2_filled);
for k=1:length(B)
   boundary = B{k};
   if k == 3 || k == 6
     fill(boundary(:,2), boundary(:,1), 'b','LineWidth', 2);
   else
     fill(boundary(:,2), boundary(:,1), 'r','LineWidth',2);
   end
end

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Shreeya 2023-11-17

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1640885-how-to-segment-cells-round-and-bacterias-longitudinal-and-label-them#answer_1354537

Hello,

I understand that you want to classify the circular and longitudinal shapes in the images without the use of any machine/deep learning methods. You can follow the below mentioned approach to perform the task using image processing and computer vision methods:

Binarize the image using the “imbinarize” function. The document linked below talks about the usage of this function. https://www.mathworks.com/help/images/ref/imbinarize.html
Perform a flood fill operation such that the entire cell is highlighted, rather than only the borders. This is required for creating connected components for further processing. MATLAB’s “imfill” function can be used for this purpose. Refer to the documentation for more details: https://www.mathworks.com/help/images/ref/imfill.html
The next step would require noise removal to remove any unwanted pixel clusters in the image. There are several approaches for this. Refer to the documentation below to understand the usage and choose a filter which best suits the requirements. https://www.mathworks.com/help/images/noise-removal.html
After all the processing, you can apply the “regionprops” method on the image. This function returns the properties of connected components such as centroid, major axis length, minor axis length and eccentricity. Find the documentation below: https://www.mathworks.com/help/images/ref/regionprops.html
Having acquired the eccentricity, e, of the shape, it can be classified as a circle if e is tending to zero and a longitudinal shape if e is tending to one.