Insights Into Object Recognition: a Comparative Study of Feature Detection and Matching Algorithms Using Kinect Data
Krishnammal Narayanan, G.Muthu Lakshmi, M. Vasumathy · 2024
In today’s rapidly evolving technological landscape, object recognition is a critical component of computer vision, impacting numerous fields such as robotics, surveillance, and augmented reality. Object recognition involves identifying and classifying semantic objects within digital images and videos. A fundamental aspect of this process is feature detection, which is essential for accurate recognition and matching. This paper presents a comparative analysis of various image matching algorithms, categorized into template-based, texture feature-based, and intensity-based methods. Specifically, we examine the performance of Edge-Oriented Gradient (EOG) for template-based matching, Speeded-Up Robust Features (SURF) for texture-based matching, and entropy and mean for intensity-based matching. Experimental results demonstrate the efficacy of these algorithms in detecting and matching features under different transformations and deformations. Our study evaluates each algorithm’s matching accuracy and computational efficiency, identifying the optimal approach for specific object recognition scenarios.