How to use vision.BlobAnalysis to represent regionprops

Hi,

I am trying to convert a program from MATLAB to C using the MATLAB Coder App. I have a lot of the code in the Image Processing Toolbox, one which will not convert using the Coder. I am trying to convert functions to the Computer Vision System Toolbox in order to convert it.

My current issue is with the function regionprops. I have a line of code as follows:

---------------------------------------------------------

% Filter the red pixels with the intensity constraints
imgbw_red = imgbw_int & imgbw_red;

% Remove small objects by erosion
se = strel('square',3);
imgbw_red = imclose(imgbw_red,se);
imgbw_red = imfill(imgbw_red,'holes');

% Find objects in the binary image
redBlobs = bwconncomp(imgbw_red);

% Compute blob properties
S = regionprops(redBlobs,'Centroid','PixelList','BoundingBox');

-----------------------------------------------------------------

The original variables are images in binary form. I believe the bwconncomp function will be unnecessary if using the function vision.BlobAnalysis, but I am unsure of how to use this function to obtain a variable equivalent to the PixelList from regionprops. This output is a p-by-q matrix that specifies the locations of pixels in the region, in coordinates. If you could give me an example of how to make this change, I would appreciate it.

Jon

Answer 1

[Walter Roberson]

http://www.mathworks.com/help/vision/ref/vision.blobanalysis-class.html

H = vision.BlobAnalysis('AreaOutputPort', false, 'CentroidOutputPort', true, 'BoundingBoxOutputPort', true, 'LabelMatrixOutputPort');

Then

[Centroids_R, BB_R, Labels_R] = step(H, imgbw_red);

Each of the outputs will be an array. The Centroids and BB will have one row per blob. The Labels will be the same size as the input. When

numblob = size(Centroids_R,1);

then

for K = 1 : numblob
  [row, col] = find(Labels_R==K);
end

to get the pixel indices.

[JVins]

Awesome, that looks good. I will try it out and see how it works! Thanks!