Eddie L. Jacobs

Video compressive sensing using spatial domain sparsity

We present a compressive sensing video acquisition scheme that relies on the sparsity properties of video in the spatial domain. In this scheme, the video sequence is represented by a reference frame, followed by the difference of measurement results between each pair of neighboring frames. The video signal is reconstructed by first reconstructing the frame differences using 1 minimization algorithm, then adding them sequentially to the reference frame. Simulation results on both simulated and real video sequences show that when the spatial changes between neighboring frames are small, this scheme provides better reconstruction results than existing compressive sensing video acquisition schemes, such as 2-D or 3-D wavelet methods and the minimum total-variance (TV) method. This scheme is suitable for compressive sensing acquisition of video sequences with relatively small spatial changes. A method that estimates the amount of spatial change based on the statistical properties of measurement results is also presented.

Active and passive imaging in the THz spectral region : phenomenology, dynamic range, modes, and illumination

The useful compromise between resolution and penetration power of the submillimeter or terahertz (THz) spectral region has long made it attractive for a variety of imaging applications. However, many of the demonstrations of imaging in this spectral region have used strategically oriented targets, especially favorable concealment materials, proximate imaging geometries, etc. This paper reports the results of studies aimed at better understanding the phenomenology of targets, the impact of this phenomenology on various active and passive imaging strategies, and most importantly, the development of imaging strategies that do not require the aforementioned special circumstances. Particular attention is paid to the relationship between active and passive images, especially with respect to how they interact with the illumination- and detector-mode structures of various imaging scenarios. It is concluded that the very large dynamic range that can be obtained with active single-mode systems (including focal-plane arrays) can be used in system designs to overcome the deleterious effects that result from the dominance of specular reflections in single-mode active systems as well as to strategically orient targets to obtain recognition. This will aid in the development of a much more robust and generally useful imaging technology in this spectral region.

Modeling the target acquisition performance of active imaging systems

Recent development of active imaging system technology in the defense and security community have driven the need for a theoretical understanding of its operation and performance in military applications such as target acquisition. In this paper, the modeling of active imaging systems, developed at the U.S. Army RDECOM CERDEC Night Vision & Electronic Sensors Directorate, is presented with particular emphasis on the impact of coherent effects such as speckle and atmospheric scintillation. Experimental results from human perception tests are in good agreement with the model results, validating the modeling of coherent effects as additional noise sources. Example trade studies on the design of a conceptual active imaging system to mitigate deleterious coherent effects are shown.

Active and Passive Millimeter and Sub-Millimeter-Wave Imaging

We have developed several millimeter/submillimeter/terahertz systems to study active and passive imaging and associated phenomenology. For measuring the transmission and scattering properties of materials, we have developed a dual rotary stage scattering system with active illumination and a Fourier Transform spectrometer. For imaging studies, we have developed a system based on a 12-inch diameter raster-scanned mirror. By interchange of active sources and both heterodyne and bolometric detectors, this system can be used in a variety of active and passive configurations. The laboratory measurements are used as inputs for, and model calibration and validation of, a terahertz imaging system performance model used to evaluate different imaging modalities for concealed weapon identification. In this paper, we will present examples of transmission and scattering measurements for common clothing as well as active imaging results that used a 640 GHz source and receiver.

Concealed weapon identification using terahertz imaging sensors

Terahertz imaging sensors are being considered for providing a concealed weapon identification capability for military and security applications. In this paper the difficulty of this task is assessed in a systematic way. Using imaging systems operating at 640 GHz, high resolution imagery of possible concealed weapons has been collected. Information in this imagery is removed in a controlled and systematic way and then used in a human observer perception experiment. From the perception data, a calibration factor describing the overall difficulty of this task was derived. This calibration factor is used with a general model of human observer performance developed at the US Army Night Vision and Electronic Sensors Directorate to predict the task performance of observers using terahertz imaging sensors. Example performance calculations for a representative imaging sensor are shown.

Characteristics of infrared imaging systems that benefit from superresolution reconstruction

There have been numerous applications of superresolution reconstruction algorithms to improve the range performance of infrared imagers. These studies show there can be a dramatic improvement in range performance when superresolution algorithms are applied to undersampled imager outputs. These occur when the imager is moving relative to the target, which creates different spatial samplings of the field of view for each frame. The degree of performance benefit is dependent on the relative sizes of the detector/spacing and the optical blur spot in focal plane space. The minimum blur spot size achievable on the focal plane is dependent on the system F/number. Hence, we provide a range of these sensor characteristics, for which there is a benefit from superresolution reconstruction algorithms. Additionally, we quantify the potential performance improvements associated with these algorithms. We also provide three infrared sensor examples to show the range of improvements associated with provided guidelines.

Current Infrared Target Acquisition Approach for Military Sensor Design and Wargaming

The U.S. Armys infrared target acquisition models have been used for many years by the military sensor community, and there have been significant improvements to these models over the past few years. Significant improvements are the Target Task Performance (TTP) metric for all imaging sensors, the ACQUIRE-LC approach for low contrast infrared targets, and the development of discrimination criteria for the urban environment. This paper is intended to provide an overview of the current infrared target acquisition modeling approach. This paper will discuss recent advances and changes to the models and methodologies used to: (1) design and compare sensors, (2) predict expected target acquisition performance in the field, (3) predict target detection performance for combat simulations, (4) measure and characterize human operator performance in an operational environment (field performance), and (5) relate the models to target acquisition tasks and address targets that are relevant to urban operations. Finally, we present a catalog of discrimination criteria, characteristic dimensions, and target contrasts.

Comparison of tilted slit and tilted edge superresolution modulation transfer function techniques

With staring infrared sensors becoming more predominant, an increasing number of system Modulation Transfer Functions MTFs are being extracted from undersampled imagery. Contamination of the MTF measurement caused by aliasing must be avoided by using a su- perresolution technique. We compare differences in the utility of two of these techniques: the tilted edge technique, which is commonly used, and the tilted slit technique, which is not. While the tilted edge MTF measurement is taken for granted, the use of the tilted slit is both inno- vative and instructive. The differences between the two techniques are examined to determine their ability to successfully extract the MTF from an undersampled image without noise, with white noise, with fixed pat- tern noise, and with various angles of tilt. Ease of implementation is also considered. All things considered, the tilted edge technique is more eas- ily accomplished and gives superior results. The tilted slit technique, however, gives valuable insight into the nature of aliasing in the real world. It is shown to be an excellent tool by which the difficult concept of aliasing may be easily illustrated.

Near-IR Sparse Detector Sensor for Intelligent Electronic Fence Applications

A proof-of-concept, active near-IR sensor coupled with a classification algorithm that uses the Mahalanobis distance has been shown to be a feasible approach for discriminating among humans, animals, and vehicles for intelligent electronic fence applications. Analysis shows that only a sparse vertical array of detectors is required to sense minimal features from moving objects for reliable discrimination.

Adaptive deblurring of noisy images

We propose a practical sensor deblurring filtering method for images that are contaminated with noise. A sensor blurring function is usually modeled via a Gaussian-like function having a bell shape. The straightforward inverse function results in the magnification of noise at high frequencies. To address this issue, we apply a special spectral window to the inverse blurring function. This special window is called the power window, which is a Fourier-based smoothing window that preserves most of the spatial frequency components in the passband and attenuates quickly at the transition band. The power window is differentiable at the transition point, which gives a desired smooth property and limits the ripple effect. Utilizing the properties of the power window, we design the deblurring filter adaptively by estimating the energy of the signal and the noise of the image to determine the passband and the transition band of the filter. The deblurring filter design criteria are (a) the filter magnitude is less than 1 at the frequencies where the noise is stronger than the desired signal (the transition band), and (b) the filter magnitude is greater than 1 at the other frequencies (the passband). Therefore the adaptively designed deblurring filter is able to deblur the image by a desired amount based on the estimated or known blurring function while suppressing the noise in the output image. The deblurring filter performance is demonstrated by a human perception experiment in which 10 observers are to identify 12 military targets with 12 aspect angles. The results of comparing target identification probabilities with blurred and deblurred images and adding two levels of noise to blurred and deblurred noisy images are reported.