Lipchen Alex Chan

Experimental Evaluation of FLIR ATR Approaches–A Comparative Study

This paper presents an empirical evaluation of a number of recently developed Automatic Target Recognition algorithms for Forward-Looking Infrared (FLIR) imagery using a large database of real FLIR images. The algorithms evaluated are based on convolutional neural networks (CNN), principal component analysis (PCA), linear discriminant analysis (LDA), learning vector quantization (LVQ), modular neural networks (MNN), and two model-based algorithms, using Hausdorff metric-based matching and geometric hashing. The evaluation results show that among the neural approaches, the LVQ- and MNN-based algorithms perform the best; the classical LDA and the PCA methods and our implementation of the geometric hashing method ended up in the bottom three, with the CNN- and Hausdorff metric-based methods in the middle. Analyses show that the less-than-desirable performance of the approaches is mainly due to the lack of a good training set.

Dualband FLIR fusion for automatic target recognition

Abstract We investigate the potential benefits of fusing two bands of forward-looking infrared (FLIR) data for target detection and clutter rejection. We propose a similar set of neural-based clutter rejecters and target detectors, each of which consists of an eigenspace transformation and a simple multilayer perceptron. The same architecture is used to operate on either single band or dualband FLIR input images, so that the net effects of dualband fusion can be demonstrated. When the dualband inputs are used, the component bands are combined at either pixel or feature level, thus providing insight into methods of performing data fusion in this particular application. A large set of real FLIR images is used in two series of experiments, one for clutter rejection tasks and the other for target detection tasks. In both series, the results indicate that the dualband input images do improve the performance of the clutter rejecters and target detectors over their single band counterparts. On the other hand, results of the pixel and feature level fusions are quite similar, suggesting that dimensionality reduction by the eigenspace transformation can be performed independently on the two bands.

Eigenspace transformation for automatic clutter rejection

The goal of our research is to develop an effective and efficient clutter rejector with the use of an eigenspace transformation and a multilayer perceptron (MLP) that can be incorporated into an automatic target recognition system. An eigenspace transformation is used for feature extraction and dimensionality reduction. The transformations considered in this research are principal-component analysis (PCA) and the eigenspace separation transformation (EST). We fed the result of the eigenspace transformation to an MLP that predicts the identity of the input, which is either a target or clutter. Our proposed clutter rejector was tested on two huge and realistic datasets of second-generation forwardlooking infrared imagery for the Comanche helicopter. In general, both the PCA and EST methods performed satisfactorily with minor differences. The EST method performed slightly better when a smaller amount of transformed data was fed to the MLP, or when the positive and negative EST eigentargets were used together.

Bipolar eigenspace separation transformation for automatic clutter rejection

A major problem for a detection algorithm is the vast amount of false alarms normally generated. This amount of false alarms has to be substantially reduced so that a typical target classifier in the subsequent stage may work reasonably. We use the bipolar eigenspace separation transformation (BEST) and neural network techniques to improve the clutter rejection performance of an automatic target detector. Experiments have been conducted on huge and realistic datasets of forward looking infrared (FLIR) imagery. Compared to the performance of the unipolar EST and principal component analysis (PCA) with the same datasets, significant improvement in clutter rejection rates has been achieved with BEST.

Automatic target recognition using vector quantization and neural networks

We propose an automatic target recognition (ATR) algorithm that uses a set of dedicated vector quantizers (VQs) and multilayer per- ceptrons (MLPs). For each target class at a specific range of aspects, the background pixels of an input image are first removed. The extracted target area is then subdivided into several subimages. A dedicated VQ codebook is constructed for each of the resulting subimages. Using the K-means algorithm, each VQ codebook learns a set of patterns repre- senting the local features of a particular target for a specific range of aspects. The resulting codebooks are further trained by a modified learn- ing vector quantization (LVQ) algorithm, which enhances the discrimina- tory power of the codebooks. Each final codebook is expected to give the lowest mean squared error (MSE) for its correct target class and range of aspects. These MSEs are then input to an array of window-level MLPs (WMLPs), where each WMLP is specialized in recognizing its in- tended target class for a specific range of aspects. The outputs of these WMLPs are manipulated and passed to a target-level MLP, which pro- duces the final recognition results. We trained and tested the proposed ATR algorithm on large and realistic data sets and obtained impressive results using the wavelet-based adaptive product VQs configuration.

Discriminative eigen targets for automatic target recognition

Three different linear transformations have been examined for their potential use as feature extractors for an automatic target recognition classifier. These transformations are based on a set of eigen targets, which are obtained through one of the following three methods: principal component analysis, the eigen separation transform, or the Fisher linear discriminant. From the sets of eigen targets obtained through each of the above methods, projection values of an input image are computed and fed to one or more multilayer perceptrons (MLPs) for training and testing purposes. With a fixed-structure MLP, each of the different eigen target sets are examined for their effects on the final recognition performance.

Improved target detector for FLIR imagery

Algorithms are considered for searching wide area forward-looking infrared imagery for military vehicles. Wide area search has typically been handled by using a simple detection algorithm with low computational cost to search the entire image or set of images, followed by a clutter rejection algorithm that analyzes only those portions of the image that are marked by the detection algorithm. We start with a feature based detector and a eigen-neural based clutter rejecter, and examine a number of architectures for combining these modules to maximize joint performance. The architectures considered include a clutter rejection threshold method and a nonlinear learning-based combination. The performance of the architectures are compared using a set of several thousand real images.

Automatic target detection using dualband infrared imagery

An automatic target detector often produces too many false alarms that could bog down the performance of a subsequent target classifier. Therefore, we need a good clutter rejector to remove as many clutterers as possible, before feeding the most likely target detections to the classifier. We investigate the benefits of using dual-band forward-looking infrared images to improve the performance of an eigen-neural based clutter rejector. With individual or combined bands as input, we use either principal component analysis or the eigenspace separation transform to perform feature extraction and dimensionality reduction. The transformed data is then fed to a properly trained multilayer perceptron that predicts the identity of the input, which is either a target or clutter. Experimental results are presented on a dataset of real dualband images.

Clutter Rejection Using Eigenspace Transformation

An effective clutter rejection scheme is needed to distinguish between clutter and targets in a high-performance automatic target recognition system. In this paper, we present a clutter rejection scheme that consists of an eigenspace transformation and a multilayer perceptron (MLP). The input to the clutter rejector module is the output of the detector that provides the potential regions (target chips). We first use an eigen transformation for feature extraction and dimensionality reduction. The transformations considered in this research are principal component analysis (PCA) and the eigenspace separation transform (EST). These transformations differ in their capabilities to enhance the class separability and to compact the information (energy) for a given training set. The result of the eigenspace transformation is then fed to an MLP that predicts the identity of the input, which is either a target or clutter. To search for the optimal performance, we use different sets of eigentargets and construct the matching MLPs. Modified from the popular Qprop algorithm, we devise an MLP training algorithm that seeks to maximize the class separation at a given false-alarm rate, which does not necessarily minimize the average deviation of the MLP outputs from their target values. Experimental results are presented on a huge and realistic data set of forward-looking infrared (FLIR) imagery.

Dual-band FLIR fusion for target detection

Passive infrared imagers have long been used to detect military targets in operational scenarios. The proliferation of sensors on the battlefield has increased the need for automatic detection algorithms with low false alarm rates and high detection rates. Most infrared imagers currently operate in a single band. We are investigating the utility of dualband passive infrared sensors for target detection, and attempting to quantify the performance improvement over single band sensors. The two bands used in this research were broadband longwave and broadband midwave. The performance differences were observed using a similar set of neural-based target detectors, each of which consists of an eigenspace transformation and a simple multilayer perceptron (MLP) with different inputs. The detectors were trained with midwave-only, longwave-only, as well as signal-level and feature-level dualband inputs. Experimental results indicate significant performance improvement by the dualband inputs over single band data.