A multiview, multimodal fusion algorithm for classifying marine plankton is described and its performance is evaluated on laboratory data from live animals. The algorithm uses support vector machines with softmax outputs to classify either acoustical or optical features. Outputs from these single-view classifiers are then combined together using a feedback network with confidence weighting. For each view or modality, the initial classification and classifications from all other views and modalities are confidence-weighted and combined to render a final, improved classification. Simple features are computed from acoustic and video data with an aim at noise robustness. The algorithm is tested on acoustic and video data collected in the laboratory from live, untethered copepods and mysids (two dominant crustacean zooplankton). It is shown that the algorithm is able to yield significant (> 50%) reductions in error by combining views together. In addition, it is shown that the algorithm is able boost performance by giving more weight to views or modalities that are more discriminant than others, without any a priori knowledge of which views are more discriminant.

A new method for designing signals with a given time-bandwidth product, such that all the signals in the set have a flat amplitude spectrum and have low cross-correlation function values is introduced. It is shown that these signals lie on a signal parameter space ellipse. For a given duration and bandwidth, it is possible to trade off the cardinality of the set for better cross-correlation properties between remaining members of the set. Proof that the cross-correlation between these signals is bounded from above by a factor proportional to the square-root of inverse distance between these signals along one of the axes of the ellipse is given. It is also shown that the cross-correlation between two signals that are furthest apart is bounded by the square-root of inverse time-bandwidth product of the signals in the set. An example design is then given and a comparison of this signal set to some well-known signal sets illustrates that it performs better when the metric is maximum cross-correlation value. The proposed signal set also has the advantage that it can be designed for any time-bandwidth product instead of discrete values.

The use of acoustical techniques in oceanography is well known. Traditional applications have concentrated on long range, low frequency propagation. Due a convergence of both technology and national interests, consideration of the possibilities for creating high frequency, real time, multi-dimensional imaging systems (2 and 3-D) has emerged. Unfortunately, due to a lack of experimental investigation, the propagation issues related to these frequencies are relatively unknown. The authors' group in underwater acoustic imaging applied to ocean exploration and oceanography has been taking high frequency acoustic pictures in the ocean for almost a decade. The first experiences were in using the traditional types of ocean going sonars such as side scans. More recently they have been developing a 3-dimensional multibeam imaging system for tracking small animals in the sea. As part of this effort they have been looking at both the spatial and temporal correlation scales for the propagation of high frequency sound.

The technique of optical serial sectioning is a method which can be used to infer the optical characteristics of a three-dimensional structure with minimal disruption of the media. The method consists of scanning a plane or stripe of light through a three dimensional object and concurrently recording the scattered radiation at a fixed angle. Under suitable circumstances, the 3-dimensional structure of the concentration of absorbers can be inferred by performing an inversion on the recorded data. The article explores both the noise propagation and also the uniqueness of the solution. A so called "low passed" inversion is shown to be unique and, by suitable approximation, it is inferred that the true, unfiltered inversion is also.>

The authors propose a stabilized method for limited angle reconstruction. The algorithm provides a framework that makes it possible to incorporate prior information and to account for the noise; this framework is the reason for the method's stability. The method entails the following steps: a space of reconstruction solutions is generated, either analytically or by Monte Carlo simulation; representatives of the solutions are found by clustering the solution space; and these representatives are scored, using the a priori information, to define the final solution. Simulation results are presented for a simple 1-D deconvolution problem.>

A sonar imaging technique which uses specially coded waveforms to project different sound signals in different directions was previously proposed. In this context, the problem of creating a set of frequency-hopped code words that have mutually small correlation properties has arisen. The authors report on their progress in designing these frequency-hopped signals for multibeam sonar imaging. The special case is considered in which medium spreading effects destroy the phase coherence of the imaging system. A design strategy is discussed which has allowed the design of a set of frequency-hopped code words that exhibit good performance bounds with respect to incoherent processing. The design strategy is a systematic constructions technique which can be used with any size Galois field GF(p) to generate p-1 codes of length p. Various theorems are presented and examples of the performance of the technique are illustrated for time-bandwidth products of interest in terms of sonar.>

This paper describes the design of frequency-hopped signals for a multi-beam imaging system. A frequency hopping pattern is a frequency-coded uniform pulse train. The signal is divided into M time intervals, with each interval assigned a different frequency chosen from a set of N frequencies. A set of N patterns composed of N-1 frequencies can be generated using first-order Reed-Solomon codewords. These patterns exhibit very good correlation properties. In a frequency-hopped multi-beam imaging system, each beam is associated with a pattern and transmits a coded waveform. All N beams can be transmitted simultaneously resulting in a high scan-rate, high resolution imaging device. Furthermore, in the presence of noise and medium spreading effects, a frequency-hopped imaging device performs better than conventional systems by showing better noise rejection and less sensitivity to spreading effects.

This paper describes the application of a super resolution algorithm to data collected in an ocean acoustic tomography experiment. In addition, the algorithm was modified in order to optimize its' performance. Our results indicate that the modified algorithm can be successfully applied to the acoustic tomography data. For a typical transfer function we have been able to achieve an increase in the resolution of the deconvolved function of a factor of 4 down to signal-to-noise ratios of 8:1.