Aiming at the automatic recognition of motorized vehicles in cluttered infrared images, this paper presents an approach for grouping multiple supervised neural networks into an efficient classifier, taking into account the variety of targets and background scenarios available in the imagery provided for the EUCLID project RTP 8.2. The proposed neural network architecture consists of a modular combination of several small multi-layer-perceptron neural networks. To take into account the false targets generated by the detection stage, along with the networks used for the discrimination between a target class and all the other classes of targets, auxiliary neural networks aim to separate targets from non-targets. For ambiguous situations it is also introduced an additional level of neural networks trained to discriminate sub-groups of classes that present similar features. Training and testing was performed using five classes of targets within cluttered environments: tanks, trucks, cars, airplanes and helicopters. Most of the data used was from real infrared imagery, although complementary synthetic target models were also introduced to test the validity of the presented approach in a wide variety of situations.

Grouping multiple neural networks for automatic target recognition in infrared imagery

Multitarget tracking in cluttered environments has to deal with noisy measurements from ambiguous origins, which leads to a process of hypotheses generation and evaluation in addition to the traditional estimation problem. The multiple hypothesis tracking (MHT) algorithm is a statistical data association that provides a high level of support for false alarms and temporary occlusion at the cost of high processing requirements. This paper describes a methodology developed for the detection of moving and stationary point targets that feeds the MHT based tracking algorithm implemented. The time filtering approach developed for target detection is based on a combined spatio-temporal analysis approach optimized for clutter driven situations. The analysis of the local variance behavior within a time window is followed by a spatial processing stage based on motion segmentation by region growing. Moving regions presenting similar properties are thus clustered into single entities. Furthermore, a background estimation algorithm is also used, aiming to increase the robustness of the detection method. Multitarget tracking is then achieved through a MHT methodology based on the spatial features of the segmented motion regions. Special attention is given to the evaluation of performance of the method in real IR image sequences.

Detection and tracking of multiple targets in infrared images

Currently there is no algorithm which can be adapted to all of the imaging conditions. So, it is necessary for us to
find a method to evaluate the existing ATR (automatic target recognition) algorithm. We do some researches on ATR
algorithm performance evaluation based on test methodology. The basic idea of the algorithm performance evaluation is
to establish the relationship model between the image quality characteristics and the algorithm’s performance. In this
paper, the algorithm performance evaluation’s techniques are studied, which include the algorithm performance
assessment framework, the universal test image database’s creating, and the research of the image quality evaluation
model. Firstly, under the guidance of the orthogonal experimental design method, we construct a universal test image
database which includes the simulation image and the outfield flight data. And then this paper propose a method to
establish the relation model between image quality characteristic and ATR algorithm based on SVM classifier. Finally
we use the model to evaluate algorithm’s performance. We conduct experiments on the matching algorithm’s
performance evaluation. The experimental results show that the proposed evaluation framework is efficient and the
evaluation model is well.

ATR algorithm performance evaluation based on the simulation image and real image

We have been studying the Chemical Oxygen-Iodine Laser (COIL) Thermal Image Marker System to the far field objects. This system can mark the distinguishable thermal image on the far field objects with the laser beam of the COIL to guide the Imaging Infrared homing air vehicle to the object marked thermal image with pinpoint accuracy. For the development of this system the study of the COIL resonator is the main task to meet the generation of the required high quality laser beam.
Therefore, first we made two kinds of the experiments. One is to generate the distinguishable thermal image mark (TIM) on the object with stable resonator of the 13 kW output COIL system in the near field. Another is to improve the laser beam quality with the unstable resonators of the COIL system in the low gain condition. Then we studied the high power unstable resonator design for this system with the numerical simulation based on its experimental data and the two-dimensional Fresnel-Kirchhoff integration method with partially coherent scalar electric field. Finally we made the numerical far field TIM generation to verify the TIM generation with the laser beam of the studied high power unstable resonator. The result of simulation shows the fine TIM generation.
The result of the experiment and the resonator design study shows that it is possible to realize the good thermal image mark, the good quality laser beam and the promising unstable resonator for the COIL Thermal Image Marker System.

An experimental and computational study of the COIL resonators for the development of the COIL thermal image marker system to the far-field objects

The infrared target processor (IRTP) is a real-time processor capable of automatic and operator-aided detection, lock-on and tracking of multiple targets in infrared imagery. The IRTP has been developed under a feeder project to the advanced land fire control system (ALFCS) program, a project to develop accurate fire on the move capability for Canadian tanks. The IRTP is built around the datacube pipelined architecture using commercially available image processing and general purpose components. Presently, the IRTP is capable of real-time detection and tracking of up to five targets at video rates (30 frames/sec), and its operation is controlled via a single monitor that contains graphical control panels and a real- time video display. To augment the operation of the target detection/tracking and the gun fire control operations of the ALFCS a position orientation system (POS) can be used to provide accurate vertical reference measurements of the turret. The POS is built around an inertial measurement unit (IMU) and specialized real-time software implemented on a general purpose CPU card. Although the IRTP is currently being developed for ALFCS applications, the testbed architecture and algorithms are flexible and the system has been used to test and develop algorithms for general target processing applications.

Real-time infrared target processor for advanced land fire control applications

As part of the development of a real-time JR target processor test bed, a number of image processing algorithms were developed, simulated in software, and evaluated for implementation. Algorithms performing image pre-processing, target localization, segmentation and target/clutter discrimination were evaluated using an IR image data base. The algorithms selected are being implemented on the test bed using commercially available board-level components, and are capable of pmcessing imagery at real-time rates (30 frames/see).© (1990) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Erik Lithopoulos

Papers

Algorithm evaluation for an infrared target-detection test bed