New approach using Bayesian Network to improve content based image classification systems

TL;DR

This paper introduces an augmented naive Bayes approach with Bayesian Network variants for content-based image classification, demonstrating improved accuracy over traditional methods.

Contribution

It presents a novel combination of image descriptor vectors with Bayesian Network variants for enhanced image classification.

Findings

FAN outperforms NB and TAN in accuracy

The approach improves classification results significantly

Bayesian Network variants effectively model image label dependencies

Abstract

This paper proposes a new approach based on augmented naive Bayes for image classification. Initially, each image is cutting in a whole of blocks. For each block, we compute a vector of descriptors. Then, we propose to carry out a classification of the vectors of descriptors to build a vector of labels for each image. Finally, we propose three variants of Bayesian Networks such as Naive Bayesian Network (NB), Tree Augmented Naive Bayes (TAN) and Forest Augmented Naive Bayes (FAN) to classify the image using the vector of labels. The results showed a marked improvement over the FAN, NB and TAN.