Håkan Larsson

Laser imaging of small surface vessels and people at sea

The development of new asymmetric threats to civilian and naval ships has been a relatively recent occurrence. The bombing of the USS Cole is one example and the pirate activities outside Somalia another. There is a need to recognize targets at long ranges and possibly also their intentions to prepare for counteractions. Eye safe laser imaging at 1.5 μm offers target recognition at long ranges during day and night. The 1.5 μm wavelength is suitable for observing small targets at the sea surface such as boats and swimmers due to the low reflectivity of water compared to potential targets. Turbulence and haze limits the sensor performance and their influence is estimated for some cases of operational interest. For comparison, passive EO images have been recorded with the same camera to investigate the difference between sun illuminated and laser illuminated images. Examples of laser images will be given for a variety of targets and external conditions.Image segmentation for future automated recognition development is described and examplified. Examples of relevant 1.5 μm laser reflectivities of small naval targets are also presented. Finally a discussion of system aspects is made.

Characterizing targets and backgrounds for 3D laser radars

Exciting development is taking place in 3 D sensing laser radars. Scanning systems are well established for mapping from airborne and ground sensors. 3 D sensing focal plane arrays (FPAs) enable a full range and intensity image can be captured in one laser shot. Gated viewing systems also produces 3 D target information. Many applications for 3 D laser radars are found in robotics, rapid terrain visualization, augmented vision, reconnaissance and target recognition, weapon guidance including aim point selection and others. The net centric warfare will demand high resolution geo-data for a common description of the environment. At FOI we have a measurement program to collect data relevant for 3 D laser radars using airborne and tripod mounted equipment for data collection. Data collection spans from single pixel waveform collection (1 D) over 2 D using range gated imaging to full 3 D imaging using scanning systems. This paper will describe 3 D laser data from different campaigns with emphasis on range distribution and reflections properties for targets and background during different seasonal conditions. Example of the use of the data for system modeling, performance prediction and algorithm development will be given. Different metrics to characterize the data set will also be discussed.

Performance of 3D laser radar through vegetation and camouflage

One of the more exciting capabilities foreseen for future 3-D imaging laser radars is to see through vegetation and camouflage nettings. We have used ground based and airborne scanning laser radars to collect data of various types of terrain and vegetation. On some occasions reference targets were used to collect data on reflectivity and to evaluate penetration. The data contains reflectivity and range distributions and were collected at 1.5 and 1.06 μm wavelength with range accuracies in the 1-10 cm range. The seasonal variations for different types of vegetation have been studied. A preliminary evaluation of part of the data set was recently presented at another SPIE conference. Since then the data have been analyzed in more detail with emphasis on testing algorithms and future system performance by simulation of different sensors and scenarios. Evaluation methods will be discussed and some examples of data sets will be presented.

Spatial filtering for detection of partly occluded targets

A Bayesian approach for data reduction based on spatial filtering is proposed that enables detection of targets partly occluded by natural forest. The framework aims at creating a synergy between terrain mapping and target detection. It is demonstrates how spatial features can be extracted and combined in order to detect target samples in cluttered environments. In particular, it is illustrated how a priori scene information and assumptions about targets can be translated into algorithms for feature extraction. We also analyze the coupling between features and assumptions because it gives knowledge about which features are general enough to be useful in other environments and which are tailored for a specific situation. Two types of features are identified, nontarget indicators and target indicators. The filtering approach is based on a combination of several features. A theoretical framework for combining the features into a maximum likelihood classification scheme is presented. The approach is evaluated using data collected with a laser-based 3-D sensor in various forest environments with vehicles as targets. Over 70% of the target points are detected at a false-alarm rate of <1%. We also demonstrate how selecting different feature subsets influence the results.

Continuously scanning time-correlated single-photon-counting single-pixel 3-D lidar

Abstract. Time-correlated single-photon-counting (TCSPC) lidar provides very high resolution range measurements. This makes the technology interesting for three-dimensional imaging of complex scenes with targets behind foliage or other obscurations. TCSPC is a statistical method that demands integration of multiple measurements toward the same area to resolve objects at different distances within the instantaneous field-of-view. Point-by-point scanning will demand significant overhead for the movement, increasing the measurement time. Here, the effect of continuously scanning the scene row-by-row is investigated and signal processing methods to transform this into low-noise point clouds are described. The methods are illustrated using measurements of a characterization target and an oak and hazel copse. Steps between different surfaces of less than 5 cm in range are resolved as two surfaces.

Lidar on small UAV for 3D mapping

Small UAV:s (Unmanned Aerial Vehicles) are currently in an explosive technical development phase. The performance of UAV-system components such as inertial navigation sensors, propulsion, control processors and algorithms are gradually improving. Simultaneously, lidar technologies are continuously developing in terms of reliability, accuracy, as well as speed of data collection, storage and processing. The lidar development towards miniature systems with high data rates has, together with recent UAV development, a great potential for new three dimensional (3D) mapping capabilities. Compared to lidar mapping from manned full-size aircraft a small unmanned aircraft can be cost efficient over small areas and more flexible for deployment. An advantage with high resolution lidar compared to 3D mapping from passive (multi angle) photogrammetry is the ability to penetrate through vegetation and detect partially obscured targets. Another advantage is the ability to obtain 3D data over the whole survey area, without the limited performance of passive photogrammetry in low contrast areas. The purpose of our work is to demonstrate 3D lidar mapping capability from a small multirotor UAV. We present the first experimental results and the mechanical and electrical integration of the Velodyne HDL-32E lidar on a six-rotor aircraft with a total weight of 7 kg. The rotating lidar is mounted at an angle of 20 degrees from the horizontal plane giving a vertical field-of-view of 10-50 degrees below the horizon in the aircraft forward directions. For absolute positioning of the 3D data, accurate positioning and orientation of the lidar sensor is of high importance. We evaluate the lidar data position accuracy both based on inertial navigation system (INS) data, and on INS data combined with lidar data. The INS sensors consist of accelerometers, gyroscopes, GPS, magnetometers, and a pressure sensor for altimetry. The lidar range resolution and accuracy is documented as well as the capability for target surface reflectivity estimation based on measurements on calibration standards. Initial results of the general mapping capability including the detection through partly obscured environments is demonstrated through field data collection and analysis.

Data collection and simulation of high range resolution laser radar for surface mine detection

Rapid and efficient detection of surface mines, IEDs (Improvised Explosive Devices) and UXO (Unexploded Ordnance) is of high priority in military conflicts. High range resolution laser radars combined with passive hyper/multispectral sensors offer an interesting concept to help solving this problem. This paper reports on laser radar data collection of various surface mines in different types of terrain. In order to evaluate the capability of 3D imaging for detecting and classifying the objects of interest a scanning laser radar was used to scan mines and surrounding terrain with high angular and range resolution. These data were then fed into a laser radar model capable of generating range waveforms for a variety of system parameters and combinations of different targets and backgrounds. We can thus simulate a potential system by down sampling to relevant pixel sizes and laser/receiver characteristics. Data, simulations and examples will be presented.

Registration and change detection techniques using 3D laser radar data from natural environments

In this paper, we present techniques related to registration and change detection using 3D laser radar data. First, an experimental evaluation of a number of registration techniques based on the Iterative Closest Point algorithm is presented. As an extension, an approach for removing noisy points prior to the registration process by keypoint detection is also proposed. Since the success of accurate registration is typically dependent on a satisfactorily accurate starting estimate, coarse registration is an important functionality. We address this problem by proposing an approach for coarse 2D registration, which is based on detecting vertical structures (e.g. trees) in the point sets and then finding the transformation that gives the best alignment. Furthermore, a change detection approach based on voxelization of the registered data sets is presented. The 3D space is partitioned into a cell grid and a number of features for each cell are computed. Cells for which features have changed significantly (statistical outliers) then correspond to significant changes.

Geometric calibration of thermal cameras

There exist several tools and methods for camera resectioning, i.e. geometric calibration for the purpose of estimating intrinsic and extrinsic parameters. The intrinsic parameters represent the internal properties of the camera such as focal length, principal point and distortion coefficients. The extrinsic parameters relate the cameras position to the world, i.e. how is the camera positioned and oriented in the world. With both sets of parameters known it is possible to relate a pixel in one camera to the world or to another camera. This is important in many applications, for example in stereo vision. The existing methods work well for standard visual cameras in most situations. Intrinsic parameters are usually estimated by imaging a well-defined pattern from different angles and distances. Checkerboard patterns are very often used for calibration since it is a well-defined pattern with easily detectable features. The intersections between the black and white squares form high contrast points which can be estimated with sub pixel accuracy. Knowing the precise dimension and structure of the pattern makes enables calculation of the intrinsic parameters. Extrinsic calibration can be performed in a similar manner if the exact position and orientation of the pattern is known. A common method is to distribute markers in the scene and to measure their exact locations. The key to good calibration is well-defined points and accurate measurements. Thermal cameras are a subset of infrared cameras that work with long wavelengths, usually between 9 and 14 microns. At these wavelengths all objects above absolute zero temperature emit radiation making it ideal for passive imaging in complete darkness and widely used in military applications. The issue that arises when trying to perform a geometric calibration of a thermal camera is that the checkerboard emits more or less the same amount of radiation in the black squares as in the white. In other words, the calibration board that is optimal for calibration of visual cameras might be completely useless for thermal cameras. A calibration board for thermal cameras should ideally be a checkerboard with high contrast in thermal wavelengths. (It is of course possible to use other sorts of objects or patterns but since most tools and software expect a checkerboard pattern this is by far the most straightforward solution.) Depending on the application it should also be more or less portable and work booth in indoor and outdoor scenarios. In this paper we present several years of experience with calibration of thermal cameras in various scenarios. Checkerboards with high contrast both for indoor and outdoor scenarios are presented as well as different markers suitable for extrinsic calibration.

Waveform analysis of lidar data for targets in cluttered environments

In this paper we study the potential of using deconvolution techniques on full-waveform laser radar data for pulse detection in cluttered environments, e.g. when a land-mine is partly occluded by vegetation. A pulse width greater than the distance between the reflecting surfaces within the footprint results in a signal that is composed by overlapping reflections that may be very difficult to analyze successfully with standard pulse detection techniques. We demonstrate that deconvolution improves the chance of successful decomposition of waveform signals into the components corresponding to the reflecting objects in the path of the laser beam. Experimental data were analyzed in terms of pulse extraction capability and distance accuracy. It was found that deconvolution increases the pulse extraction performance, but that surfaces closer than about 40% of the laser pulse width are still very difficult to detect and that the number of spurious, erroneously extracted, points is the price to pay for increased pulse detection probability.