EFFECT OF MINMAX NORMALIZATION ON ORB DATA FOR IMPROVED ANN ACCURACY

Abstract

This study delves into the impact of MinMax normalization on Oriented FAST and Rotated BRIEF (ORB) data when utilized as input for an Artificial Neural Network (ANN). The primary objective is to compare the accuracy of an ANN model using two distinct types of input data: raw ORB data and MinMax-normalized ORB data. The results underscore the pivotal role played by MinMax normalization in significantly enhancing the performance of the ANN model. Through a series of comprehensive experiments, it becomes evident that MinMax-normalized ORB data consistently outperforms raw ORB data in terms of accuracy. Impressively, the highest accuracy attained through MinMax normalization reaches 76.6%, whereas the utilization of raw ORB data yields a maximum accuracy of merely 51.1%. This noteworthy improvement effectively validates the prowess of MinMax normalization in counteracting the adverse effects stemming from varied scales within raw data. As a result, the ANN benefits from improved pattern recognition capabilities and heightened predictive accuracy.