Does feature level fusion needs to have data normalized only before it can be made as a large feature vector ? What are the steps needed to perform feature level fusion ?
According to my knowledge; yes normalized data are required. You may also profit from the following code:
https://ch.mathworks.com/matlabcentral/fileexchange/55405-feature-fusion-using-discriminant-correlation-analysis-dca
If there is no correlation between the different feature vectors, they can be normalized before the fusion.
Data fusion has many aspects. Based on the scenario fusion can be done in sensor level, extracted feature level or classifier level (late fusion). In this study fusion approaches have been used : Article Emotion Based Music Recommendation System Using Wearable Phy...
- Badges
- Science topic