Arie N. de Jong

Band selection from a hyperspectral data-cube for a real-time multispectral 3CCD camera

Given a specific task, like detection of hidden objects (i.e. vehicles and landmines) in a natural background, hyperspectral data gives a significant advantage over RGB- color or gray-value images. It introduces however, a trade- off between cost, speed, signal-to-noise ratio, spectral resolution, and spatial resolution. Our research concentrates on making an optimal choice for spectral bands in an imaging system with a high frame rate and spatial resolution. This can be done using a real-time multispectral 3CCD camera, which records a scene with three detectors, each accurately set to a wavelength by selected optical filters. This leads to the subject of this paper: how to select three optimal bands from hyperspectral data to perform a certain task. The choice of these bands includes two aspects, the center wavelength, and the spectral width. A band-selection and band-broadening procedure has been developed, based on statistical pattern recognition techniques. We will demonstrate our proposed band selection algorithm, and present its classification results compared to red- green-blue and red-green-near-infrared data for a military vehicle in a natural background and for surface laid landmines in vegetation.

Infrared polarisation measurements of targets and backgrounds in a marine environment

The infrared (IR) radiation emitted or reflected in an off- normal direction from a smooth surface is partially polarized. This principle can be used for enhanced discrimination of targets from backgrounds in a marine environment. It has been shown that (man-made) targets do not demonstrate a pronounced polarization effect when observed from near normal direction whereas the sea background radiation has a certain degree of polarization in slant observation path. A measurement setup has been constructed for collecting polarized IR imagery. This setup contains a rotating polarization filter that rotates synchronously with the frame sync of the camera. Either a long wave IR (LWIR) or a mid wave IR (MWIR) camera can be mounted behind the rotating polarization filter. The synchronization allows a sequence of images to be taken with a predefined constant angle of rotation between the images. Out of this image sequence three independent Stokes images are constructed, containing the normal intensity part, the vertical/horizontal polarization and the diagonal polarization. Up to 20 full linearly polarized images can be acquired per second. Measurements are taken at the North Sea coast with this setup. The recorded images are analyzed to determine the influence of polarization on the detection of small targets in such an environment. Furthermore differences between polarization contrasts in MWIR are analyzed.

Detection of surface-laid and buried mines with IR and CCD cameras: an evaluation based on measurements

We analyzed a time series of high resolution 8 - 13.4 micrometers scanner images of a sandbox with buried (3 - 10 cm depth) and unburied, metal and plastic AP and AT mines, surrogates and other targets. With a high resolution DUDA scanner operating in 8 - 13.4 micrometers all the surface laid targets were visible during the whole diurnal cycle. The buried targets were only visible during sunrise and sunset. The emissivity of the targets and sand could not be derived from the measured apparent temperatures and contact temperatures.

IR panoramic alerting sensor concepts and applications

During the last decade, protection of military and civilian operational platforms against weapons like guns, grenades, missiles, Unmanned Combat Aerial (and surface) Vehicles (UCAVs) and mines, has been an issue of increased importance due to the improved kill-probability of these threats. The standard countermeasure package of armour, guns, decoys, jammers, camouflage nets and smokes is inadequate when not accompanied by a suitable sensor package, primarily consisting of an alerting device, triggering consecutive steps in the countermeasure-chain. In this process of alert four different detection techniques are considered: pre-alert, giving the directions of possible attack, detection of an action of attack, identification of the threat and finally the precise localization (3-D). The design of the alerting device is greatly depending on the platform, on which it will be used, the associated and affordable cost and the nature of the threat. A number of sensor packages, considered, developed and evaluated at TNO-FEL is presented for simple, medium size and large and expensive platforms. In recent years the requirements for these sensors have become more and more strigent due to the growing number of scenarios. The attack can practically be from any direction, implying the need for a large Field of Regard (FOR), the attack range can vary considerably and the type of threat can be very diverse, implying great flexibility and dynamic range and rapid response of the sensor. Especially the localization at short ranges is a challenging issue. Various configurations including advantages and drawbacks are discussed.

Preliminary results of the FATMOSE atmospheric propagation trials in the False Bay, South Africa, November 2009-July 2010

The FATMOSE trial (FAlse-bay ATMOSpheric Experiment) running over a period from November 2009 to July 2010, was a continuation of the cooperation between TNO and IMT on atmospheric propagation and point target detection and identification in a maritime environment. Instruments were installed for measuring scintillation, blurring- and refraction effects over a 15.7 km path over sea. Simultaneously, a set of instruments was installed on a mid-path lighthouse for collecting local meteorological data, including scintillation, sea surface temperature and visibility. The measurements covered summer and winter conditions with a prevailing high wind speed from the South East, bringing in maritime air masses. The weather conditions included variations in the Air-Sea Temperature Difference (ASTD), that may affect the vertical temperature gradient in the atmospheric boundary layer, causing refraction effects in the lightpath. This was measured with a theodolite camera, providing absolute Angles of Arrival (AOA). Blur data were collected with a high resolution camera system with 10 bits dynamic range. Specially designed image analysis software allows determination of the atmospheric blur, while simultaneously providing information on the Scintillation Index (S.I.). This S.I. was also measured by using the Multiband Spectral Radiometer Transmissometer (MSRT). The ratio of the transmission levels of this instrument contains information on the size distribution of the aerosols along the path. In the paper, experimental details on the set-up and the instrumentation are given as well as the methods of analysis. Preliminary results are shown, including a comparison of measured blur and scintillation data with Cn 2 data from the scintillometer, correlation between AOA and ASTD and comparison of transmission data with data from the visibility meter. Blur and scintillation data are compared with predictions from standard turbulence model predictions, using Cn 2. In future studies the data will be used for validation of propagation models such as EOSTAR.

Optical Characterisation of small surface targets

Present-day naval operations take place in coastal environments as well as narrow straits all over the world. Coastal environments around the world are exhibiting a number of threats to naval forces. In particular a large number of asymmetric threats can be present in environments with cluttered backgrounds as well as rapidly varying atmospheric conditions. In these conditions the threat contrast may be low and varying, and the amount of background clutter can be severe. These conditions require the electro-optical means of detection and classification to be optimized in order to have more time to act against threats. In particular the assessment of classification means is an important issue. Beside short-range coastal paths, long-range horizontal paths with variable atmospheric conditions are of interest. The small differences between types of vessel can help us determine the classification of the vessel type. Different payloads and people on-board can be clues to the classification of the vessel. Operations in warmer environments, limiting the atmospheric transmission due to water vapour absorption, are challenging. Understanding of the impact of the different environments on the optical characteristics of threats is of great importance. For this purpose a trial was planned to assess the optical characteristics of different types of small surface vessels in a coastal environment. During this trial a number of small targets were used during different parts of the day and night. Furthermore positional as well as atmospheric characterisation was performed as ground truth information. From this data a first analysis was performed showing strong intensity fluctuation in target as well as background signal levels. At longer ranges and in coastal environments these target signals may well be hidden within the background clutter. This data is essential to feed models for the assessment of sensor performance in coastal environment.

Optical characteristics of small surface targets, measured in the False Bay, South Africa; June 2007

During the False Bay trial (June 2007), the performance of a set of three optical sensors was tested against several small surface targets in a coastal area: a hyperspectral camera, a camera with a rotating polarization filter and a high resolution camera. One objective was the validation of a contrast and clutter model for the visual spectral band in this type of scenarios. Another issue was to test the benefit of using a polarisation filter and a hyperspectral unit in front of standard TV cameras. Finally the loss in identification capability of a high resolution camera at long range due to atmospheric blur was investigated. Data were collected of targets in near-sun direction at ranges up to seven kilometers, in all cases for down looking angles (targets below the horizon). Environmental parameters such as solar irradiance and windspeed were measured as input for the contrast and clutter models. Spatial, spectral and temporal effects of the target contrast and of the background clutter behaviour in the visual spectral band were determined as function of range and compared with model predictions. Samples of data and predictions are presented in this paper. The spatial and temporal target characteristics are of key importance for the development of algorithms for target detection, classification and tracking. Finally, rules of thumb, based on the measurements and model predictions, on the detection and identification range performances of specific optical sensors against small surface targets in a maritime environment are presented.

Aerosol size distributions, retrieved from multi-band transmissometer data in the Southern Baltic Sea during the VAMPIRA trials

In an earlier paper [1], data from our Multi-Band Radiometer Transmissometer (MSRT) were used to compare the ratio of extinction coefficients in different spectral bands during periods of changing visibility conditions. This ratio is an indication of the characteristics and origin (eg rural or maritime) of the haze- or fog particles, present in the measurement path. In this paper we will analyze the VAMPIRA transmission data in more detail by separating the contributions due to molecular extinction, scattering and (potentially) refraction. In our analysis we take the contribution due to scattering in order to obtain the characteristics of the Particle Size Distribution (PSD). For this purpose we take the average value and the slope of the measured transmission level in two neighboring spectral bands. Via a special simulation tool, developed for Junge-type PSDs, the slope of the PSD (defined: Junge exponent) and its value at a particle diameter of 1 μm (Junge coefficient) can be determined via a set of retrieval steps. Reference is made to a similar approach [2] where in stead of a Junge distribution, three contiguous lognormal distributions are taken. The associated procedure for the Junge-type PSD is explained in detail in this paper and applied to the VAMPIRA transmission data. The versatility of the new retrieval method is demonstrated, especially when wavelengths around lμm are chosen (a somewhat higher number than the diameter of the majority of the particles, so that most of the scattering is in the so-called Rayleigh regime). It is obvious, that the method fails in conditions of dense fog, when the transmission levels (average value and slope) over the 8.6 km path approach zero. The results are compared with in-situ PSD measurements, carried out simultaneously with a PMS (Particle Measurement System) probe at the pier near the Suerendorf shore station. In many conditions different results appear due to the fact that the MSRT system delivers path integrated data, while the PMS probe measures locally in a small volume. The MSRT data, collected over an overseas path, are more relevant to be used in the data analysis of the shore based sensor systems [3], measuring simultaneously signal values of distant point targets. The MSRT system has a higher signal to noise ratio and due to the shorter time constant, rapid fluctuations in particle characteristics are observed, not measured by the PMS probe. The availability of reliable aerosol characteristics (i.e. Junge exponent and -coefficient) allows a more precise interpretation of the data from the surveillance systems.

Refraction effects of atmospheric inhomogeneities along the path

A critical moment in the detection process of incoming targets at sea occurs, when a target just appears above the horizon. The corresponding light rays cross the atmospheric boundary layer, in which the presence of temperature gradients may result in optical distortion effects. This geometric distortion implies propagation effects such as the variation of the angle of arrival of the horizon line, a change in the shape of an extended target, the possible presence of mirages and enhanced or decreased atmospheric transmission or apparent radiant intensity of a point target. This situation is becoming even more complex, when the temperature profile is not constant over the path length, which is likely to be the case in coastal areas with tidal currents. Another effect causing complexity is the presence of surface waves, introducing vertical motion in the marine boundary layer. All these effects may have an impact on the detection performance of optical and infrared sensors for detection and identification of targets near the horizon. Presently available propagation models are unable to predict accurately the effect of the phenomena on the propagation of surface grazing light beams. A new and accurate ray-tracing model has been developed, allowing quantitative predictions of the various propagation effects for a given profile of the temperature. This model, capable to take into account the dimensions of the target and the receiver aperture, is described in the paper. The ray tracing in the model is based upon the Huygens-Fresnel principle, in contrary to other models, where a layered atmosphere is used. Examples are given of the effect of different temperature profiles and comparisons of predictions are made with data from field measurements.A critical moment in the detection process of incoming targets at sea occurs, when a target just appears above the horizon. The corresponding light rays cross the atmospheric boundary layer, in which the presence of temperature gradients may result in optical distortion effects. This geometric distortion implies propagation effects such as the variation of the angle of arrival of the horizon line, a change in the shape of an extended target, the possible presence of mirages and enhanced or decreased atmospheric transmission or apparent radiant intensity of a point target. This situation is becoming even more complex, when the temperature profile is not constant over the path length, which is likely to be the case in coastal areas with tidal currents. Another effect causing complexity is the presence of surface waves, introducing vertical motion in the marine boundary layer. All these effects may have an impact on the detection performance of optical and infrared sensors for detection and identification of targets near the horizon. Presently available propagation models are unable to predict accurately the effect of the phenomena on the propagation of surface grazing light beams. A new and accurate ray-tracing model has been developed, allowing quantitative predictions of the various propagation effects for a given profile of the temperature. This model, capable to take into account the dimensions of the target and the receiver aperture, is described in the paper. The ray tracing in the model is based upon the Huygens-Fresnel principle, in contrary to other models, where a layered atmosphere is used. Examples are given of the effect of different temperature profiles and comparisons of predictions are made with data from field measurements.

Validation of atmospheric propagation models in littoral waters

Abstract. Various atmospheric propagation effects are limiting the long-range performance of electro-optical imaging systems. These effects include absorption and scattering by molecules and aerosols, refraction due to vertical temperature gradients and scintillation and blurring due to turbulence. In maritime and coastal areas, ranges up to 25 km are relevant for detection and classification tasks on small targets (missiles, pirates). From November 2009 to October 2010 a measurement campaign was set-up over a range of more than 15 km in the False Bay in South Africa, where all of the propagation effects could be investigated quantitatively. The results have been used to provide statistical information on basic parameters as visibility, air-sea temperature difference, absolute humidity and wind speed. In addition various propagation models on aerosol particle size distribution, temperature profile, blur and scintillation under strong turbulence conditions could be validated. Examples of collected data and associated results are presented in this paper.