Use this URL to cite or link to this record in EThOS:
Title: Landmine detection algorithm design based on data fusion technology
Author: Jing, Hongyuan
ISNI:       0000 0004 7657 9223
Awarding Body: University of Leicester
Current Institution: University of Leicester
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
This research has focused on close-in landmine detection, which aims to identify landmines in a particular landmine area. Close-range landmine detection requires both sub-surface sensors, such as metal detectors and ground penetrating radar (GPR), and surface sensors, such as optical cameras. A new multi-focus image fusion algorithm is proposed which outperforms the existing intensity-hue-saturation (IHS) and principle components analysis (PCA) algorithms on both visual and fusion parameter analysis. In addition, the proposed algorithm can save 30.9% running time than the IHS algorithm, which is the same level as the existing PCA algorithm. A novel single GPR sensor landmine detection algorithm entropy-based region selecting algorithm is proposed which uses the entropy value of the region as the feature and continuous layers instead of a hard threshold. Two A-scan based statistics algorithms and a GPR signal oscillation feature based detection algorithm are also proposed. The results show that the proposed entropy-based algorithm outperforms the existing region selection algorithm on both detection accuracy and running time. The proposed statistics algorithms and GPR feature-based algorithm outperform the edge histogram descriptor and edge energy algorithms on both detection accuracy range, running time and memory usage. In addition, the GPR feature-based algorithm can reduce the false alarm rate (FAR) by 22% for all targets at 90% probability of detection. With regards to data fusion system design, this research overcomes the limitations of the existing Bayesian fusion approach. A new Kalman-Bayes based fusion system is developed which reduces the system uncertainty and improves the fusion process. The experimental results have shown that the proposed Kalman-Bayes fusion system and enhanced fuzzy fusion system can reach 7.8% FAR at 91.1% detection rate and 6.30% FAR at 92.4% detection rate, correspondingly, outperforming the existing Bayes and fuzzy fusion systems in terms of detection ability.
Supervisor: Vladimirova, Tanya Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available