How to use a particular feature vector to encode another feature vector as a training set using bag of features(Multiply Distorted Images)

Hello Experts in the house,

I need your help on this subject matter.

Method 1: I have two dataset of images(distorted) using bag of features, I have 100 images and supplementary images of 50. I used the 100 images to create codebook using the 50 images. All are distorted images, it is just like multiply distorted images. When I applied SVM after the 100 features vector were extracted. The performance was very poor.

Method 2: I extracted the 100 images using LBP and 50 images using LBP(Local Binary Pattern) to see if there could be improvement but I have issues with the coding.

The dataset of images are different images that is 100 and 50 images but the same distortion type.

................................................................................................................................................................................

clc, clear, close all

load 980LBPfffff % the 100 images extracted using LBP

load featureLBP1200 % the 50 images extracted using LBP

feature1 = (feature); %extracted 100 images

feature2 = (feature1); % extracted 50 images

bag = bagOfFeatures(feature1,'StrongestFeatures',1,'GridStep',[32 32],'BlockWidth',[32 64 96 128],'VocabularySize',500);

featureVector = encode(bag, feature2);

% Plot the histogram of visual word occurrences

figure

bar(featureVector)

title('Visual word occurrences')

xlabel('Visual word index')

ylabel('Frequency of occurrence')

.........................................................................................................................

Error:

Error using bagOfFeatures/parseInputs (line 1159)

The value of 'imds' is invalid. Expected imds to be one of these types:

imageSet, matlab.io.datastore.ImageDatastore

Instead its type was double.

Error in bagOfFeatures (line 183)

[imgSets, params] = bagOfFeatures.parseInputs(varargin{:});

Error in twofeatLBPcobk (line 10)

bag = bagOfFeatures(feature1,'StrongestFeatures',1,'GridStep',[32 32],'BlockWidth',[32 64 96 128],'VocabularySize',500);

.........................................................................................................................................................................................................

I believe am wrong somewhere but I appreciate if you can help with better code using bag of features for the codebook of the extracted feature.

Thank you all.

[Fatma Bouhlel]

did you found a solution for your problem