David Bednarz

Fuzzy-logic-based sensor fusion of images

The fusion of visual and infrared sensor images of potential driving hazards in static infrared and visual scenes is computed using the Fuzzy Logic Approach (FLA). The FLA is presented as a new method for combining images from different sensors for achieving an image that displays more information than either image separately. Fuzzy logic is a modeling approach that encodes expert knowledge directly and easily using rules. With the help of membership functions designed for the data set under study, the FLA can model and interpolate to enhance the contrast of the imagery. The Mamdani model is used to combine the images. The fused sensor images are compared to metrics to measure the increased perception of a driving hazard in the sensor-fused image. The metrics are correlated to experimental ranking of the image quality. A data set containing IR and visual images of driving hazards under different types of atmospheric contrast conditions is fused using the Fuzzy Logic Approach (FLA). A holographic matched-filter method (HMFM) is used to scan some of the more difficult images for automated detection. The image rankings are obtained by presenting imagery in the TARDEC Visual Perception Lab (VPL) to subjects. Probability of detection of a driving hazard is computed using data obtained in observer tests. The matched-filter is implemented for driving hazard recognition with a spatial filter designed to emulate holographic methods. One of the possible automatic target recognition devices implements digital/optical cross-correlator that would process sensor-fused images of targets. Such a device may be useful for enhanced automotive vision or military signature recognition of camouflaged vehicles. A textured clutter metric is compared to experimental rankings.

Noise and contrast comparison of visual and infrared images of hazards as seen inside an automobile

The purpose of this experiment was to quantitatively measure driver performance for detecting potential road hazards in visual and infrared (IR) imagery of road scenes containing varying combinations of contrast and noise. This pilot test is a first step toward comparing various IR and visual sensors and displays for the purpose of an enhanced vision system to go inside the driver compartment. Visible and IR road imagery obtained was displayed on a large screen and on a PC monitor and subject response times were recorded. Based on the response time, detection probabilities were computed and compared to the known time of occurrence of a driving hazard. The goal was to see what combinations of sensor, contrast and noise enable subjects to have a higher detection probability of potential driving hazards.

Use of a photosimulation laboratory for estimating vehicle detection probability

Abstract : A method is described for using a photosimulation laboratory environment to compare detection metrics and evaluate the effectiveness of camouflage for military vehicles. There are distinct advantages to acquiring images at the field site and then bringing them back to a laboratory environment for observer testing versus taking the subjects out to the field for estimating detection probability. Laboratory testing using field acquired imagery provides a repeatable, secure, and relatively low-cost way to generate consistent data for the measurement of the effectiveness of camouflage relative to a baseline vehicle, and the calibration and validation of target acquisition models. A laboratory test procedure is described by the authors in which a baseline Light Armored Vehicle (LAV) is compared to a treated LAV in the TACOM Visual Perception Laboratory (VPL) using imagery collected from the field in the manner prescribed by an experimental design.

Statistical measures of sensor performance based on the correlation of background clutter

The authors report the statistical analysis of a digitized cluttered background scene containing a military ground vehicle. This is the first phase of a study to evaluate several sensors and scenes to generate a statistical measure of IR sensor performance based on the pixel by pixel correlation of the output imagery.

Mobility versus terrain: a game theoretic approach

Mobility and terrain are two sides of the same coin. You cannot describe mobility unless you describe the terrain. For example, if my world is trench warfare, the tank may be the ideal vehicle. If my world is urban warfare, clearing buildings and such, the tank may not be an ideal vehicle, perhaps an anthropomorphic robot would be better. We seek a general framework for mobility that captures the relative value of different mobility strategies. Game theory is positively the right way to analyze the interactions of rational players who behave strategically. In this paper, we will describe the interactions between a mobility player, who is trying to make it from point A to point B with one chance to refuel, and a terrain player who is trying to minimize that probability by placing an obstacle somewhere along the path from A to B. In previous work [1], we used Monte Carlo methods to analyze this mobility game, and found optimal strategies for a discretized version of the game. Here we show the relationship of this game to a classic game of timing [2], and use solution methods from that literature to solve for optimal strategies in a continuous version of this mobility game.

A game of timing with detection uncertainty

Mobility and terrain are two sides of the same coin. I cannot speak to my mobility unless I describe the terrains ability to thwart my maneuver. Game theory describes the interactions of rational players who behave strategically. In previous work we described the interactions between a mobility player, who is trying to maximize the chances that he makes it from point A to point B with one chance to refuel, and a terrain player who is trying to minimize that probability by placing an obstacle somewhere along the path from A to B. In this paper, we add the twist that the mobility player cannot use their resource until they detect the terrain player. This relates to the literature of games of incomplete information, and can be thought of as a more realistic model of this interaction. In this paper we generalize the game of timing studied in the previous paper to include the possibility that one of the players has imperfect ability to detect his adversary.

Comparison of 2D and 3D displays and sensor fusion for threat detection, surveillance and telepresence

Visible, infrared (IR) and sensor-fused imagery of scenes that contain occluded camouflaged threats are compared on a two dimensional (2D) display and a three dimensional (3D) display. A 3D display is compared alongside a 2D monitor for hit and miss differences in the probability of detection of objects. Response times are also measured. Image fusion is achieved using a Gaussian Laplacian pyramidal approach with wavelets for edge enhancement. Detecting potential threats that are camouflaged or difficult to see is important not only for military acquisition problems but, also for crowd surveillance as well as tactical use such as on border patrols. Imaging and display technologies that take advantage of 3D and sensor fusion will be discussed.

Computing the probability of target detection in dynamic visual scenes containing clutter using fuzzy logic approach

The probability of detection (Pd) of moving targets in visually cluttered scenes is computed using the fuzzy logic approach (FLA). The FLA is presented as a robust method for the computation and prediction of the Pd of targets in cluttered scenes with sparse data. A limited data set of visual imagery is used to model the relationships between several input parameters: the contrast, vehicle camouflage, range, aspect, width, and experimental Pd. The fuzzy and neurofuzzy models provide predicted Pd values that have 0.9 correlation with the experimental Pds. The results obtained indicate the robustness of the fuzzy-based modeling techniques and the potential applicability of the FLA to those types of problems that have to do with the modeling of aided or unaided detection of a signal (acoustic, electromagnetic) in any spectral regime.

Calculation of background clutter in infrared imagery: a semiempirical study

The authors report the statistical analysis of infrared scenes containing a military ground vehicle. The purpose is to attempt to determine the important variables in clutter as well as the robustness of the present definition of clutter through computer simulation. Both variance based and texture based clutter metrics are compared. The authors analyzed both cluttered and non-cluttered scenes.