Comparison of Drone and Helicopter Image Classification Accuracy Using Naïve Bayes Based on Mean Red-Green-Blue (RGB) Values and First-Order Statistics

Abstract

The increasing use of Unmanned Aerial Vehicles (UAVs) such as drones and helicopters across various sectors presents a challenge in distinguishing between them due to their similar appearances in aerial imagery. This similarity necessitates the development of an accurate image classification system to differentiate the two types of flying objects. This study proposes a classification method using the Naïve Bayes algorithm combined with two feature extraction approaches: (1) the mean intensity values of Red, Green, and Blue (RGB) image channels, and (2) first-order statistical features including mean, standard deviation, skewness, and kurtosis of pixel intensities. A dataset of 60 images, consisting of 30 drone and 30 helicopter images, was used. Feature extraction was conducted using Python in the Google Collab environment, while classification was performed with WEKA software. The results show that the RGB mean features yielded a classification accuracy of 91.67% with an area under the ROC curve (AUC) of 1.0, outperforming the statistical features, which achieved 75% accuracy and an AUC of 0.938. These findings demonstrate that colour-based RGB features are more effective in distinguishing drones from helicopters compared to statistical texture features.