Hello everyone, I would like to ask that how to modify the following code. It includes bwlabel function and regionprops. I don't know where the code is wrong.

BW21=imread('m1.jpg');

%Step 1: Label each object using the following code.

ppp = imbinarize(BW21); % be sure your image is binary

L= bwlabel(ppp); % label each object

%Step 2: see the label of each object

s = regionprops(L,'Centroid');

imshow(ppp)

hold on

Code Analyzer has found a call of regionprops with a first argument that is a call to bwlabel. In many cases, the argument of bwlabel is a logical matrix, or you can convert it to one. Beginning in MATLAB Version 7.8 (release R2009a), you can pass this logical matrix directly to regionprops without the call to bwlabel. This is faster and uses less space than the previous method.

Suggested Action

If bw is a logical matrix, make the indicated change. If bw is not a logical matrix, convert it to a logical matrix, and then pass the converted matrix as the first argument to regionprops.

For example, if bw is a logical matrix, replace:

regionprops(bwlabel(bw), ...)

with:

regionprops(bw, ....)

Answer 1

[Image Analyst]

I wouldn't worry about it. You can pass in (the poorly-named) ppp if you want. But you still may want the labeled image because often you filter the blobs based on some criteria and to extract them, you can conveniently do it with ismember() if you have the labeled image. The time difference between using ppp or L will be unnoticeable in most cases.

[Tamara Ahmed]

Hello, I have a project that goes beyond image processing in matlab. I need the special code to extract shape properties from images using Hog and regionprops, noting that the image contains a rectangle and a circle. I will attach an image as an example. The images whose properties are to be extracted are the words encoded by Morse code

[Tamara Ahmed]

And this one too

[Image Analyst]

@Tamara Ahmed OK. Looks like fun. Good luck. If you need help with it, post your questions and images in a new question (not here in Wesley's question) and study my Image Segmentation Tutorial.

Image Analyst's File Exchange

You'll see how I differentiate coins types based on area. You'll do the same thing with your blobs to determine if it's a dot or dash. You'll want to threshold for dark things though, like

binaryImage = grayImage < someThreshold;

or use imbinarize() and invert it if you need to, to get the blobs to be white and background dark.