How do I create a matrix of the surf features for about 100 images?

Basically, I'm creating a Neural Network.. I have all of my images and I have the code to detect the SURF Features. Now, I need help to place the images with detected features into a matrix please.

This is the SURF code with the image it display with SURF points. I need to do this for 300 images made into a matrix to start training my NN.

clear;
close all;
I = imread('D:\test\chair\0001.png');
figure; imshow(I);
I = rgb2gray(I);
figure; imshow(I);
points = detectSURFFeatures(I);
figure;
imshow(I); hold on;
plot(points.selectStrongest(10));

[features, valid_points] = extractFeatures(I, points);
plot(valid_points.selectStrongest(5),'showOrientation',true);

[KALYAN ACHARJYA]

You have done for one image, now you want to do same for 300 images right?

Now, I need help to place the images with detected features into a matrix please

Which matrix, variable name please?

[Lauren Pitts]

@KALYAN ACHARJYA Right, I need to do it on all of my testing images. Do you mean what do I want to name the matrix?

[KALYAN ACHARJYA]

@Prabhan is already answerd, hope the problem is resolved, if not let me know?

Answer 1

[Prabhan Purwar]

Hello,

I am assuming that you are making use of a Neural Network for classification purpose.

Following code makes a row-wise matrix for 300 images using SURF features as descriptors.

clear; 
close all 
clc; 
for i=1:300 
 st=strcat('D:\test\chair\000',num2str(i,'%02d'),'.png'); 
    I = imread(st); 
    I = rgb2gray(I); 
    points = detectSURFFeatures(I); 
    [features, valid_points] = extractFeatures(I, points); 
    [s1,s2]=size(features); 
    S=s1*s2; 
    ftr=reshape(features,1,S); 
    F(i,:)=ftr; 
end 

The following link describes basic image classification using a simple convolution Neural Network:

• https://www.mathworks.com/help/deeplearning/gs/create-simple-deep-learning-classification-network.html
• https://www.mathworks.com/videos/what-is-deep-learning-toolbox--1535667599631.html

I would like to suggest you make use of "Image category classification using a bag of features approach" which uses SURF features and SVM classifier. This gives you a single vector that you can use at the input to a neural network or any other classifier of your choice.

Please refer to the following link: https://www.mathworks.com/help/vision/examples/image-category-classification-using-bag-of-features.html

[Zamokuhle Ngubane]

Hey @Prabhan, The code you provided above does not work, it gives me an error: "The following error occurred converting from binaryFeatures to double:

Conversion to double from binaryFeatures is not possible."

[Andrea Daou]

Hello, I am trying to use your code but for each image the number of SURF features will change so the final matrix F cannot be generated because at each loop iteration F(k,:) changes size.

Do you have any idea how can I fix this problem?

Thank you in advance!

[Pinaki Saha]

Store the features in cell instead

F{i}=ftr;