Toward an operational sensor-fusion system for antipersonnel land mine detection

Abstract

To acquire detection performance required for an operational system for the detection of anti-personnel Iandmines, it is necessary to use multiple sensors and sensor-fusion techniques. This paper describes five decision-level sensor-fusion techniques and their common optimisation method.The performance of the sensor-fusion techniques is evaluated by means of Receiver Operator Characteristics curves. These techniques are tested on an outdoor test facility. Three of four test lanes of this facility are used as training set and the fourth is used as evaluation set.The detection performance of naive Bayes, Dempster-Shafer, voting and linear discriminant are very similar on both the training and the evaluation set. This is probably caused by the flexibility of the sensor-fusion techniques resulting into similar optimal solutions independent of the fusion technique.The first LOTUS trial took place from May 17th through 21st 1999 at the TNO test site, situated in the dunes of the Hague, the Netherlands. This test site consists of 6 test lanes, each with a different soil type: sand, clay, peat, ferruginous, forest and road. Each test lane measures 3 by 1 0 m2 and has a controlled water table. The test lanes are free of vegetation and the soilof the first four test lanes is filtered to remove metal objects. In the test lanes, a mixture of anti-personal (AP) mines and anti-tank (AT) mines are buried or laid on the surface. The sensors are attached to a metal free X-Y table for taking stand-off measurements.In Table 1 an overview is given of the mine types in the six test lanes. In total there are 265mines consisting of 216 APs and 49 ATs. The mine types AP5 and AT3 are metal free. The others (73%) contain at least some metal, see Table 2. The first four test lanes contain roughly similar number of mines of each type. However, there are more mines present of mine type AP5 and AP7 on the first four test lanes.

A Systematic Approach of Data Fusion Technique in RFID Sensor Network using Neuro-Fuzzy Technique

“…Dempster-Shafer and Bayesian data fusion approach are describe in [5] [6]. In [7][8] [9][10] [11][12] the works focused on object identification using data fusion and image analysis technique. A multi-level approach in Sensor fusion is described in [13] [14][15] [16].…”

A Review of data fusion models and architectures: towards engineering guidelines

“…In 1986 the US Department of Defence established the Data Fusion Sub-Panel of the Joint Directors of Laboratories (JDL) to address some of the main issues in data fusion and chart the new field in an effort to unify the terminology and procedures. The present applications of data fusion span a wide range of areas: maintenance engineering [1], robotics [2], pattern recognition and radar tracking [3], mine detection [4] and other military applications [5], remote sensing [6], traffic control [7], [8], aerospace systems [9], law enforcement [10], medicine, finance, metrology [11], and geo-science.…”

Statistical Evaluation Method for Comparing Grid Map Based Sensor Fusion Algorithms

“…For landmine detection, Cremer et al (1999Cremer et al ( , 2001) compared several fusion methods (naive Bayes, Dempster-Shafer, fuzzy-based, rule-based, voting techniques) using the probability of detection P (d) (Lombardi and Zavidovique 2004) and of false alarm P(fa) (Lombardi and Zavidovique 2004), ROC, SCOOP, and prod-area algorithm (Schwering et al 2002). The algorithms were also compared, based on their expected costs.…”

“…The algorithms were also compared, based on their expected costs. All algorithms were evaluated on a specific experimental data set and compared to the baseline derived by the individual sensors ROC (Cremer et al 1999(Cremer et al , 2001). Influence of the data set on fusion performance is a well-known fact (Schwering et al 2002) and therefore its applicability is limited.…”