Research on precise identification and localization methods for static small targets based on multimodal data fusion

Abstract

This investigation introduces a multimodal fusion-based methodology for the precise detection and accurate localization of static small targets against uniform backgrounds. Small targets often present fewer pixels in imagery, which significantly reduces the detectable features and complicates accurate identification and localization. Our approach integrates visual data from CMOS cameras, distance readings from laser rangefinders, and orientation data from angle sensors. The image processing involves non-local means denoising and Canny edge detection to enhance small target detection. We map two-dimensional images to three-dimensional spaces, combining depth and orientation to precisely determine the target’s coordinates, aiding in three-dimensional localization. Empirical results confirm the method’s high accuracy, with a localization error under 2.5%. Such conclusive evidence highlights the approach’s efficacy, presenting a cost-efficient alternative for the nuanced detection and localization of small entities.