This PDF file contains the front matter associated with SPIE Proceedings Volume 6971, including the Title Page, Copyright information, Table of Contents, Introduction (if any), and the Conference Committee listing.

The Zero Instruction Set Computer (ZISC) is an integrated circuit devised by IBM to realize a restricted Coulomb energy neural network. In our application, it functions as a parallel computer that calculates the correlation coefficients between an input pattern and patterns stored in its neurons. We explored the possibility of using the ZISC in a target tracking system by devising algorithms to take advantage of the ZISC's parallelism and testing them on real video sequences. Our experiments indicate that the ZISC does improve appreciably the computing time compared to a sequential version of the algorithm.

Low resolution and un-sharp facial images are always captured from surveillance videos because of long human-camera distance and human movements. Previous works addressed this problem by using an active camera to capture close-up facial images without considering human movements and mechanical delays of the active camera. In this paper, we proposed a unified framework to capture facial images in video surveillance systems by using one static and active camera in a cooperative manner. Human faces are first located by a skin-color based real-time face detection algorithm. A stereo camera model is also employed to approximate human face location and his/her velocity with respect to the active camera. Given the mechanical delays of the active camera, the position of a target face with a given delay can be estimated using a Human-Camera Synchronization Model. By controlling the active camera with corresponding amount of pan, tilt, and zoom, a clear close-up facial image of a moving human can be captured then. We built the proposed system in an 8.4-meter indoor corridor. Results show that the proposed stereo camera configuration can locate faces with average error of 3%. In addition, it is capable of capturing facial images of a walking human clearly in first instance in 90% of the test cases.

A variety of unmanned air vehicles (UAVs) have been developed for both military and civilian use. The typical large UAV is typically state owned, whereas small UAVs (SUAVs) may be in the form of remote controlled aircraft that are widely available. The potential threat of these SUAVs to both the military and civilian populace has led to research efforts to counter these assets via track, ID, and attack. Difficulties arise from the small size and low radar cross section when attempting to detect and track these targets with a single sensor such as radar or video cameras. In addition, clutter objects make accurate ID difficult without very high resolution data, leading to the use of an acoustic array to support this function. This paper presents a multi-sensor architecture that exploits sensor modes including EO/IR cameras, an acoustic array, and future inclusion of a radar. A sensor resource management concept is presented along with preliminary results from three of the sensors.

We propose a real-time vehicle detection and tracking system from an electro-optical (EO) surveillance camera. Real time object detection remains a challenging computer vision problem in uncontrolled environments. The state-of-the-art adaboosting technique is used to serve as a robust object detector. In addition to the generally-used Haar features, we propose to include corner features to improve the detection performance of the vehicles. Having the objects of interest detected, we use the detection results to initialize the object tracking module. We propose an advanced, adaptive particle-filtering based algorithm to robustly track multiple mobile targets by adaptively changing the appearance model of the selected targets. We use the affine transformation to describe the motion of the object across frames. By drawing multiple particles on the transformation parameters, our approach provides high performance while facilitating implementation of this algorithm in hardware with parallel processing. In order to resume from the lost track case, which may result from the objects' out of boundary or being occluded, we utilize the prior information (height-to-width ratio) and the temporal information of the objects to estimate if the tracking is reliable. Object detectors will be evoked at the frames which fail in tracking the objects reliably. We also check for occlusion by comparing hue values within the rectangular region for the current frame with that of the previous frame. Detection is re-initialized for the next frame if an occlusion is claimed for the current frame. The system works very well in terms of speed and performance for the real surveillance video.

As target tracking is arousing more and more interest, the necessity to reliably assess tracking algorithms in any conditions is becoming essential. The evaluation of such algorithms requires a database of sequences representative of the whole range of conditions in which the tracking system is likely to operate, together with its associated ground truth. However, building such a database with real sequences, and collecting the associated ground truth appears to be hardly possible and very time-consuming. Therefore, more and more often, synthetic sequences are generated by complex and heavy simulation platforms to evaluate the performance of tracking algorithms. Some methods have also been proposed using simple synthetic sequences generated without such complex simulation platforms. These sequences are generated from a finite number of discriminating parameters, and are statistically representative, as regards these parameters, of real sequences. They are very simple and not photorealistic, but can be reliably used for low-level tracking algorithms evaluation in any operating conditions. The aim of this paper is to assess the reliability of these non-photorealistic synthetic sequences for evaluation of tracking systems on complex-textured objects, and to show how the number of parameters can be increased to synthesize more elaborated scenes and deal with more complex dynamics, including occlusions and three-dimensional deformations.

The communication architecture for most pointing, tracking, and high order adaptive optics control systems has been based on a centralized point-to-point and bus based approach. With the increased use of larger arrays and multiple sensors, actuators and processing nodes, these evolving systems require decentralized control, modularity, flexibility redundancy, integrated diagnostics, dynamic resource allocation, and ease of maintenance to support a wide range of experiments. Network control systems provide all of these critical functionalities. This paper begins with a quick overview of adaptive optics as a control system and communication architecture. It then provides an introduction to network control systems, identifying the key design areas that impact system performance. The paper then discusses the performance test results of a fielded network control system used to implement an adaptive optics system comprised of: a 10KHz, 32x32 spatial selfreferencing interferometer wave front sensor, a 705 channel "Tweeter" deformable mirror, a 177 channel "Woofer" deformable mirror, ten processing nodes, and six data acquisition nodes. The reconstructor algorithm utilized a modulo-2pi wave front phase measurement and a least-squares phase un-wrapper with branch point correction. The servo control algorithm is a hybrid of exponential and infinite impulse response controllers, with tweeter-to-woofer saturation offloading. This system achieved a first-pixel-out to last-mirror-voltage latency of 86 microseconds, with the network accounting for 4 microseconds of the measured latency. Finally, the extensibility of this architecture will be illustrated, by detailing the integration of a tracking sub-system into the existing network.

Future space based deployable telescopes will be subject to non-atmospheric disturbances. Jitter and optical misalignment on a spacecraft can be caused by mechanical noise of the spacecraft, and settling after maneuvers. The introduction of optical misalignment and jitter can reduce the performance of an optical system resulting in pointing error and contributing to higher order aberrations. Adaptive optics can be used to control jitter and higher order aberrations in an optical system. In this paper, wavefront control methods for the Naval Postgraduate School adaptive optics testbed are developed. The focus is on removing structural noise from the flexible optical surface using discrete time proportional integral control with second order filters. Experiments using the adaptive optics testbed successfully demonstrate wavefront control methods, including a combined iterative feedback and gradient control technique. This control technique results in a three time improvement in RMS wavefront error over the individual controllers correcting from a biased mirror position. Second order discrete time notch filters are also used to remove induced low frequency actuator and sensor noise at 2Hz. Additionally a 2 Hz structural disturbance is simulated on a Micromachined Membrane Deformable Mirror and removed using discrete time notch filters combined with an iterative closed loop feedback controller, showing a 36 time improvement in RMS wavefront error over the iterative closed loop feedback alone.

Future space based deployable telescopes will be subject to non-atmospheric disturbances. Jitter and optical misalignment on a spacecraft can be caused by mechanical noise of the spacecraft, and settling after maneuvers. The introduction of optical misalignment and jitter can reduce the performance of an optical system resulting in pointing error and contributing to higher order aberrations. Adaptive optics can be used to control jitter and higher order aberrations in an optical system. In this paper, wavefront control methods for the Naval Postgraduate School adaptive optics testbed are developed. The focus is on removing structural noise from the flexible optical surface using discrete time proportional integral control with second order filters. Experiments using the adaptive optics testbed successfully demonstrate wavefront control methods, including a combined iterative feedback and gradient control technique. This control technique results in a three time improvement in RMS wavefront error over the individual controllers correcting from a biased mirror position. Second order discrete time notch filters are also used to remove induced low frequency actuator and sensor noise at 2Hz. Additionally a 2 Hz structural disturbance is simulated on a Micromachined Membrane Deformable Mirror and removed using discrete time notch filters combined with an iterative closed loop feedback controller, showing a 36 time improvement in RMS wavefront error over the iterative closed loop feedback alone.

Pointing and tracking applications usually require relative gimbal angles to be measured for reporting and controlling the line-of-sight angular position. Depending on the application, angular resolution and/or accuracy might jointly or independently determine the angle transducer requirements. In the past decade, encoders have been increasingly taking the place of inductive devices where the measurement of angles over a wide range is required. This is primarily due to the fact that encoders are now achieving very high resolution in smaller sizes than was previously possible. These advances in resolution are primarily due to improved encoder disk and detector technology along with developments in interpolation techniques. Measurement accuracy, on the other hand, is primarily determined by mounting and bearing eccentricity as it is with all angular measurement devices. For very demanding accuracy requirements, some type of calibration of the assembled system may be the only solution, in which case transducer repeatability is paramount. This paper describes a unique encoder-to-digital tracking converter concept for improving interpolation of optical encoders. The new method relies on Fraunhofer diffraction models to correct the non-ideal sin/cos outputs of the encoder detectors. Diffraction model concepts are used in the interpolation filters to predict the phase of non-ideal sin and cosine encoder outputs. The new method also minimizes many of the open loop pre-processing requirements and assumptions that limit interpolation accuracy and rate loop noise performance in ratiometric tracking converter designs.

We are presenting a novel electro-optic architecture for non-mechanical laser beam steering with a demonstrated 80 degrees of steering in a chip-scale package. To our knowledge this is the largest angular coverage ever achieved by non-mechanical means. Even higher angular deflections are possible with our architecture both in the plane of the waveguide and out of the waveguide plane. In the present paper we describe the steering in the plane of the waveguide leaving the out-of-plane scanning mechanism to be detailed in a subsequent publication. In order to realize this performance we exploit an entirely new electro-optic architecture. Specifically, we utilize liquid crystals (LCs), which have the largest known electro-optic response, as an active cladding layer in an LC-waveguide geometry. This architecture exploits the benefits of liquid crystals (large tunable index), while circumventing historic LC limitations. LC-waveguides provide unprecedented macroscopic (>1 mm) electro-optic phase delays. When combined with patterned electrodes, this provides a truly analog, "Snell's-law-type" beam-steerer. With only two control electrodes we have realized an 80 degree field of view for 1550 nm light. Furthermore, the waveguide geometry keeps the light from ever coming into contact with an ITO electrode, thereby permitting high optical power transmission. Finally, the beamsteering devices have sub-millisecond response times.

We present a novel algorithm taking measurements of time, solar irradiance, wind speed, peak wind speed, temperature gradient, and relative humidity to derive a predictive differential equation for mean C_n². Our method derives individual control terms and forcing functions by modeling macro-structure, micro-structure, and fine structure terms independently. The final model is suitable for analysis and able to be used as a baseline expectation model for in situ battlefield use for predictive optical correction or slewing, and possibly for mitigating the effects of wind shear on artillery shells downrange.

In this work we examine the dynamic implications of active and attentive scanning for LADAR based automatic target/object recognition and show that a dynamically constrained, scanner based, ATR system's ability to identify objects in real-time is improved through attentive scanning. By actively and attentively scanning only salient regions of an image at the density required for recognition, the amount of time it takes to find a target object in a random scene is reduced. A LADAR scanner's attention is guided by identifying areas-of-interest using a visual saliency algorithm on electro-optical images of a scene to be scanned. Identified areas-of-interest are inspected in order of decreasing saliency by scanning the most salient area and saccading to the next most salient area until the object-of-interest is recognized. No ATR algorithms are used; instead, an object is considered to be recognized when a threshold density of pixels-on-target is reached.

Target tracking is one of the main applications of wireless sensor networks. Optimized computation and energy dissipation are critical requirements to save the limited resource of sensor nodes. A new energy efficient collaborative target tracking algorithm via particle filtering (PF) is presented. Assuming the network infrastructure is cluster-based, collaborative scheme is implemented through passing sensing and computation operations from one active cluster to another and an event driven cluster reforming approach is also proposed for evening energy consumption distribution. At each time step, measurements from three sensors are chosen at the current active cluster head to estimate and predict the target motion and the results are propagated among cluster heads to the sink. In order to save the communication and computation resource, we present a new particle filter algorithm called Gaussian Rao-Blackwellised Particle Filter (GRBPF), which approximate the posterior distributions by Gaussians and only the mean and covariance of the Gaussians need to be communicated among cluster heads when target enter another cluster. The GRBPF algorithm is also more computation efficient than generic PF by dropping the resampling step. In the simulation comparison, a target moves through the sensor network field and is tracked by both generic PF and the GRBPF algorithm using our proposed collaborative scheme. The results show that the latter works very well for target tracking in wireless sensor networks and the total communication burden is substantially reduced, so as to prolong the lifetime of wireless sensor networks.