Using Singular value decomposition for feature extraction from images

16 visualizaciones (últimos 30 días)
Ron Herman
Ron Herman el 17 de Abr. de 2020
Respondida: Yee Mon el 13 de Mzo. de 2024 a las 9:25
Suppose I have a single subject who has 10 images under different conditions. I need to apply singular value decomposition on all 10 images for feature extraction. Then storing the common extracted feature from 10 images.
The purpose is to save space. Rather than saving all 10 images I would like to save common extracted in one place and use the common extracted feature for future face recognition of the same subject (person).
The 10 images have been attached.
  2 comentarios
Christine Tobler
Christine Tobler el 17 de Abr. de 2020
Do you want to compute the singular value decomposition of each of these images and store those? Or compute a combined decomposition of all 10 images?
Ron Herman
Ron Herman el 17 de Abr. de 2020
Editada: Ron Herman el 17 de Abr. de 2020
SVD on each image for feature extraction.
I also want to know feature extraction using SVD

Iniciar sesión para comentar.

Respuestas (2)

Christine Tobler
Christine Tobler el 17 de Abr. de 2020
Image compression using SVD is a pretty common example (although not the most efficient way to compress an image), here are some relevant links:
For feature extraction, I don't know of a "standard" algorithm to do this using SVD. I found some papers searching for this topic, maybe one of these could be helpful:

Yee Mon
Yee Mon el 13 de Mzo. de 2024 a las 9:25
% Load the image
image1 = imread('yme.jpg');
% Convert the image to grayscale if it's RGB
if size(image, 3) == 3
image = rgb2gray(image);
end
% Perform Singular Value Decomposition (SVD) on the image matrix
[U, S, V] = svd(double(image));
% Choose the number of singular values to keep (feature extraction)
num_features = 50;
U = U(:, 1:num_features);
S = S(1:num_features, 1:num_features);
V = V(:, 1:num_features);
% Reconstruct the image using the selected features
reconstructed_image = unit8(U * S * V');
% Display the original and reconstructed images by SVD features
subplot(1, 2, 1);
imshow(image1);
title('Original Image');
subplot(1, 2, 2);
imshow(reconstructed_image);
title('Reconstructed Feature Image with SVD');

Categorías

Más información sobre Eigenvalues en Help Center y File Exchange.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by