Quadratic Gabor filters for object detection

We present a new class of quadratic filters that are capable of creating spherical, elliptical, hyperbolic and linear decision surfaces which result in better detection and classification capabilities than the linear decision surfaces obtained from correlation filters. Each filter comprises of a number of separately designed linear basis filters. These filters are linearly combined into several macro filters; the output from these macro filters are passed through a magnitude square operation and are then linearly combined using real weights to achieve the quadratic decision surface. For detection, the creation of macro filters (linear combinations of multiple single filters) allows for a substantial computational saving by reducing the number of correlation operations required. In this work, we consider the use of Gabor basis filters; the Gabor filter parameters are separately optimized. The fusion parameters to combine the Gabor filter outputs are optimized using an extended piecewise quadratic neural network (E-PQNN). We demonstrate methods for selecting the number of macro Gabor filters, the filter parameters and the linear and nonlinear combination coefficients. We present preliminary results obtained for an infrared (IR) vehicle detection problem.

Extracting input-line position from Hough transform data

The Hough transform (HT) detects lines in an input but not their location. We describe a new way to determine the position of a line from HT data. The line position information is extracted from the shape of the HT pattern around the HT peak. Results are shown illustrating this algorithm on single- and multiple-line input images.

Performance evaluation of minimum average correlation energy filters

A new SDF type correlation filter referred to as the minimum average correlation energy (MACE) filter has been recently described in the literature. In this paper, we experimentally address the distortion tolerance and noise properties of this filter. The MACE filter has attractive properties that include: easily detectable peaks, distortion invariance, simplified training set selection, solutions to input bias effects, performance in noise and real background clutter, and less clutter with its reduced number of training set images. Each of these properties is investigated in detail in this paper.

Acoustooptic linear heterodyned complex-valued matrix-vector processor

A new analog (linear) optical vector-inner-product and matrix-vector processor architecture is described. It employs linear input laser-diode arrays and a multichannel acoustooptic cell. Attention is given to the selection of the operating modes for all devices, error correction, the ability of the system to operate on complex-valued data, and its stability with respect to temperature, data history, and component variations. Laboratory data are provided verifying the potential of the system architecture to achieve 9-10-bit accuracy and complex-valued multiplication.

Adaptive-clustering optical neural net

Pattern recognition techniques (for clustering and linear discriminant function selection) are combined with neural net methods (that provide an automated method to combine linear discriminant functions into piecewise linear discriminant surfaces). The resulting adaptive-clustering neural net is suitable for optical implementation and has certain desirable properties in comparison with other neural nets. Simulation results are provided.

Effects of error sources on the parallelism of an optical matrix-vector processor

The error sources in a high accuracy optical matrix-vector processor are analyzed by numerical simulation in terms of their effects on the parallelism and speed of the processor. These effects are detailed for radices -2, -4 and -8. Radix -4 is shown to provide maximum parallel processing capabilities under the effects of the system's error sources. Processing speed is shown to be a function of matrix partitioning and the number of parallel processing channels. Consequently, radix -4 operation provides a higher processing speed than radix -2 and -8 for most matrix-vector multiplications when error source effects are considered.

Ho-Kashyap optical associative processors

A Ho-Kashyap (H-K) associative processor (AP) is shown to have a larger storage capacity than the pseudoinverse and correlation APs and to accurately store linearly dependent key vectors. Prior APs have not demonstrated good performance on linearly dependent key vectors. The AP is attractive for optical implementation. A new robust H-K AP is proposed to improve noise performance. These results are demonstrated both theoretically and by Monte Carlo simulation. The H-K AP is also shown to outperform the pseudoinverse AP in an aircraft recognition case study. A technique is developed to indicate the least reliable output vector elements and a new AP error correcting synthesis technique is advanced.

High accuracy optical processors: a new performance comparison

A hybrid time and space integrating processor is shown to provide more than one operation per analog to digital conversion. A preferred performance measure, operations per second per component, is used to compare bit-sliced digital and high accuracy optical processors using the digital multiplication-by-analogconvolution algorithm. Optical system performance is shown to increase nearly quadratically with the speed of its components, while digital systems increase only linearly.

Optical laboratory morphological inspection processor

Morphological transformations are applied to industrial inspection problems. A real time optical architecture to implement morphological transformations such as erosion, opening, closing, and skeletonization is described and analyzed. The first real time optical laboratory results of erosion and opening are presented for locating string in tobacco.

Optical correlation CGHs with modulated error diffusion

Error diffusion presents a very attractive method for manufacturing computer-generated holograms. However, it has certain defects that make it unsuitable for various optical processing applications, such as correlation. We present an extension of error diffusion which makes it suitable for the implementation of correlation filters.

Hough transform detection of 3-D curves and target trajectories

The problem of detecting curves and target trajectories in 3-D space is examined. Techniques based on the Hough transform and a 3-D extension of it are presented. The effectiveness of the techniques for the detection of circles and circular missile trajectories given two arbitrary 2-D views (or given sequences of one range image of point target trajectories) is illustrated. Simulation results indicate that performance is excellent, even in the presence of gaps in the input data.

Optical neural net for classifying imaging spectrometer data

A problem in surface mineralogy is addressed; namely, how does one determine the composition of a mixture from its spectrum? A neural net algorithm arises naturally, and we detail the state equations of this net. An optical architecture and simulation results are presented.

Correlation filters for target detection in a Markov model background clutter

The performance of distortion-invariant correlation filters in the presence of background clutter is addressed. Background images are modeled as Markov noise processes, and a synthesis procedure for the optimal filter is described. It is shown that spatially filtering the training set images eliminates the need for the inversion of large noise covariance matrices, thus leading to a computationally efficient filter realization. The effect of errors (in the estimation of clutter correlation coefficient) on filter performance is theoretically analyzed, and a bound on the relative degradation of the SNR due to such errors is presented.

Multitarget tracking with cubic energy optical neural nets

A neural net processor and its optical realization are described for a multitarget tracking application. A cubic energy function results and a new optical neural processor is required. Initial simulation data are presented.

Optical systems for digit-serial computation

Optical systems and algorithms are presented for implementing digit-serial (or on-line) computations. These systems achieve high accuracy without the need for A-Ds and incorporate parallelism and carry-free addition to achieve high processing speed. On-line arithmetic allows parallel calculations to be performed by concurrent execution of operations. (Consecutive operations can start before all digits of the previous operations are available.) The algorithms are problem- and step-invariant and inherently allow variable precision as we show. To achieve parallelism, we introduce the modified signed-digit number representation into these algorithms and architectures. This results in new arithmetic rules (new addition rules, a different number of bits needed, a different number of cycles required) from those in conventional digital digit-serial systems. We include sufficient detail (not readily available elsewhere) needed for the design of the units we include. New architectures using optical bistable devices and optical interconnects are described that can implement digit-serial addition, subtraction, multiplication, and division algorithms in this new approach.

Fast JPDA multitarget tracking algorithm

Multitarget tracking requires assigning probabilities that measurements are associated with particular targets. A new and efficient algorithm to achieve this is detailed. Its implementation on an analog optical processor using a new and cost-effective frequency-multiplexing technique is discussed.

Optical projection correlations

The correlation of several 1-D projections of a 2-D image are considered for pattern recognition. A theoretical analysis and SNR comparison to 2-D correlations are provided with successful simulated results that show that the use of two or three 1-D correlations can identify and discriminate the 26 characters in the alphabet. Several possible 1-D optical correlators to implement projection correlations are described.

Parallel holographic generation of multiple Hough transform slices

Many uses for the Hough transform (HT) require only several theta slices, not the entire 2-D Hough space. A new holographic optical architecture is advanced to achieve this in parallel. The computer-generated holograms and holographic optical elements used are discussed, and laboratory results are provided. As a case study, we consider determining printing skew and other defects on cigarette packs. We also discuss how these holograms can produce the entire 2-D HT space.

Optical track initiator for multitarget tracking

An optical processor for multitarget track initiation is presented. It uses the Hough transform to detect target tracks in a fixed time independent of the number of tracks present. In new results, we describe how the system produces target position and velocity estimates in linear time (with the number of targets). Tests of the system are presented for multiple targets with various velocities and target detection probabilities. A new thresholded Hough transform algorithm to reduce the effects of noise and false tracks in single data frames is introduced and tested. The optical implementation of these Hough transform track initiation algorithms is discussed.

Multiple degree of freedom object recognition using optical relational graph decision nets

Multiple degree of freedom object recognition concerns objects with no stable rest position with all scale, rotation, and aspect distortions possible. We assume that the objects are in a fairly benign background, so that feature extractors are usable. In-plane distortion invariance is provided by use of a polar-log coordinate transform feature space, and out-of-plane distortion invariance is provided by linear discriminant function design. Relational graph decision nets are considered for multiple degree of freedom pattern recognition. The design of Fisher linear discriminant functions and synthetic discriminant functions for use at the nodes of binary and multidecision nets is discussed. Case studies are detailed for two-class and multiclass problems. Simulation results demonstrate the robustness of the processors to quantization of the filter coefficients and to noise.

Factorized extended Kalman filter: case study results

Simulation results for an optical realization of a factorized extended Kalman filter algorithm are presented, minimum word lengths required for accurate tracking are empirically determined, and computation times for an optical realization are quantified.

Computer generated hologram recording using a laser printer

The use of a laser printer for recording various types of computer generated holograms is discussed, and initial results are presented.