Object Detection at the Optimal Scale with Hidden State Shape Models

Hidden State Shape Models (HSSMs) [2], a variant of Hidden Markov Models (HMMs) [9], were proposed to detect shape classes of variable structure in cluttered images. In this paper, we formulate a probabilistic framework for HSSMs which provides two major improvements in comparison to the previous method [2]. First, while the method in [2] required the scale of the object to be passed as an input, the method proposed here estimates the scale of the object automatically. This is achieved by introducing a new term for the observation probability that is based on a object-clutter feature model. Second, a segmental HMM [6, 8] is applied to model the "duration probability" of each HMM state, which is learned from the shape statistics in a training set and helps obtain meaningful registration results. Using a segmental HMM provides a principled way to model dependencies between the scales of different parts of the object. In object localization experiments on a dataset of real hand images, the proposed method significantly outperforms the method of [2], reducing the incorrect localization rate from 40% to 15%. The improvement in accuracy becomes more significant if we consider that the method proposed here is scale-independent, whereas the method of [2] takes as input the scale of the object we want to localize.

Laminar Cortical Dynamics of Visual Form and Motion Interactions During Coherent Object Motion Perception

How do visual form and motion processes cooperate to compute object motion when each process separately is insufficient? A 3D FORMOTION model specifies how 3D boundary representations, which separate figures from backgrounds within cortical area V2, capture motion signals at the appropriate depths in MT; how motion signals in MT disambiguate boundaries in V2 via MT-to-Vl-to-V2 feedback; how sparse feature tracking signals are amplified; and how a spatially anisotropic motion grouping process propagates across perceptual space via MT-MST feedback to integrate feature-tracking and ambiguous motion signals to determine a global object motion percept. Simulated data include: the degree of motion coherence of rotating shapes observed through apertures, the coherent vs. element motion percepts separated in depth during the chopsticks illusion, and the rigid vs. non-rigid appearance of rotating ellipses.

View-Invariant Object Category Learning, Recognition, and Search: How Spatial and Object Attention Are Coordinated Using Surface-Based Attentional Shrouds

Air Force Office of Scientific Research (F49620-01-1-0397); National Science Foundation (SBE-0354378); Office of Naval Research (N00014-01-1-0624)

Cortical Dynamics of Contextually-Cued Attentive Visual Learning and Search: Spatial and Object Evidence Accumulation

How do humans use predictive contextual information to facilitate visual search? How are consistently paired scenic objects and positions learned and used to more efficiently guide search in familiar scenes? For example, a certain combination of objects can define a context for a kitchen and trigger a more efficient search for a typical object, such as a sink, in that context. A neural model, ARTSCENE Search, is developed to illustrate the neural mechanisms of such memory-based contextual learning and guidance, and to explain challenging behavioral data on positive/negative, spatial/object, and local/distant global cueing effects during visual search. The model proposes how global scene layout at a first glance rapidly forms a hypothesis about the target location. This hypothesis is then incrementally refined by enhancing target-like objects in space as a scene is scanned with saccadic eye movements. The model clarifies the functional roles of neuroanatomical, neurophysiological, and neuroimaging data in visual search for a desired goal object. In particular, the model simulates the interactive dynamics of spatial and object contextual cueing in the cortical What and Where streams starting from early visual areas through medial temporal lobe to prefrontal cortex. After learning, model dorsolateral prefrontal cortical cells (area 46) prime possible target locations in posterior parietal cortex based on goalmodulated percepts of spatial scene gist represented in parahippocampal cortex, whereas model ventral prefrontal cortical cells (area 47/12) prime possible target object representations in inferior temporal cortex based on the history of viewed objects represented in perirhinal cortex. The model hereby predicts how the cortical What and Where streams cooperate during scene perception, learning, and memory to accumulate evidence over time to drive efficient visual search of familiar scenes.

ART-EMAP: A Neural Network Architecture for Object Recognition by Evidence Accumulation

A new neural network architecture is introduced for the recognition of pattern classes after supervised and unsupervised learning. Applications include spatio-temporal image understanding and prediction and 3-D object recognition from a series of ambiguous 2-D views. The architecture, called ART-EMAP, achieves a synthesis of adaptive resonance theory (ART) and spatial and temporal evidence integration for dynamic predictive mapping (EMAP). ART-EMAP extends the capabilities of fuzzy ARTMAP in four incremental stages. Stage 1 introduces distributed pattern representation at a view category field. Stage 2 adds a decision criterion to the mapping between view and object categories, delaying identification of ambiguous objects when faced with a low confidence prediction. Stage 3 augments the system with a field where evidence accumulates in medium-term memory (MTM). Stage 4 adds an unsupervised learning process to fine-tune performance after the limited initial period of supervised network training. Each ART-EMAP stage is illustrated with a benchmark simulation example, using both noisy and noise-free data. A concluding set of simulations demonstrate ART-EMAP performance on a difficult 3-D object recognition problem.

A Neural Theory of Attentive Visual Search: Interactions of Boundary, Surface, Spatial, and Object Representations

Visual search data are given a unified quantitative explanation by a model of how spatial maps in the parietal cortex and object recognition categories in the inferotemporal cortex deploy attentional resources as they reciprocally interact with visual representations in the prestriate cortex. The model visual representations arc organized into multiple boundary and surface representations. Visual search in the model is initiated by organizing multiple items that lie within a given boundary or surface representation into a candidate search grouping. These items arc compared with object recognition categories to test for matches or mismatches. Mismatches can trigger deeper searches and recursive selection of new groupings until a target object io identified. This search model is algorithmically specified to quantitatively simulate search data using a single set of parameters, as well as to qualitatively explain a still larger data base, including data of Aks and Enns (1992), Bravo and Blake (1990), Chellazzi, Miller, Duncan, and Desimone (1993), Egeth, Viri, and Garbart (1984), Cohen and Ivry (1991), Enno and Rensink (1990), He and Nakayarna (1992), Humphreys, Quinlan, and Riddoch (1989), Mordkoff, Yantis, and Egeth (1990), Nakayama and Silverman (1986), Treisman and Gelade (1980), Treisman and Sato (1990), Wolfe, Cave, and Franzel (1989), and Wolfe and Friedman-Hill (1992). The model hereby provides an alternative to recent variations on the Feature Integration and Guided Search models, and grounds the analysis of visual search in neural models of preattentive vision, attentive object learning and categorization, and attentive spatial localization and orientation.

The What-And-Where Filter: A Spatial Mapping Neural Network for Object Recognition and Image Understanding

The What-and-Where filter forms part of a neural network architecture for spatial mapping, object recognition, and image understanding. The Where fllter responds to an image figure that has been separated from its background. It generates a spatial map whose cell activations simultaneously represent the position, orientation, ancl size of all tbe figures in a scene (where they are). This spatial map may he used to direct spatially localized attention to these image features. A multiscale array of oriented detectors, followed by competitve and interpolative interactions between position, orientation, and size scales, is used to define the Where filter. This analysis discloses several issues that need to be dealt with by a spatial mapping system that is based upon oriented filters, such as the role of cliff filters with and without normalization, the double peak problem of maximum orientation across size scale, and the different self-similar interpolation properties across orientation than across size scale. Several computationally efficient Where filters are proposed. The Where filter rnay be used for parallel transformation of multiple image figures into invariant representations that are insensitive to the figures' original position, orientation, and size. These invariant figural representations form part of a system devoted to attentive object learning and recognition (what it is). Unlike some alternative models where serial search for a target occurs, a What and Where representation can he used to rapidly search in parallel for a desired target in a scene. Such a representation can also be used to learn multidimensional representations of objects and their spatial relationships for purposes of image understanding. The What-and-Where filter is inspired by neurobiological data showing that a Where processing stream in the cerebral cortex is used for attentive spatial localization and orientation, whereas a What processing stream is used for attentive object learning and recognition.

3-D Object Recognition by the ART-EMAP Evidence Accumulation Network

ART-EMAP synthesizes adaptive resonance theory (AHT) and spatial and temporal evidence integration for dynamic predictive mapping (EMAP). The network extends the capabilities of fuzzy ARTMAP in four incremental stages. Stage I introduces distributed pattern representation at a view category field. Stage 2 adds a decision criterion to the mapping between view and object categories, delaying identification of ambiguous objects when faced with a low confidence prediction. Stage 3 augments the system with a field where evidence accumulates in medium-term memory (MTM). Stage 4 adds an unsupervised learning process to fine-tune performance after the limited initial period of supervised network training. Simulations of the four ART-EMAP stages demonstrate performance on a difficult 3-D object recognition problem.

Laminar Cortical Dynamics of Visual Form and Motion Interactions During Coherent Object Motion Perception

How do visual form and motion processes cooperate to compute object motion when each process separately is insufficient? Consider, for example, a deer moving behind a bush. Here the partially occluded fragments of motion signals available to an observer must be coherently grouped into the motion of a single object. A 3D FORMOTION model comprises five important functional interactions involving the brain’s form and motion systems that address such situations. Because the model’s stages are analogous to areas of the primate visual system, we refer to the stages by corresponding anatomical names. In one of these functional interactions, 3D boundary representations, in which figures are separated from their backgrounds, are formed in cortical area V2. These depth-selective V2 boundaries select motion signals at the appropriate depths in MT via V2-to-MT signals. In another, motion signals in MT disambiguate locally incomplete or ambiguous boundary signals in V2 via MT-to-V1-to-V2 feedback. The third functional property concerns resolution of the aperture problem along straight moving contours by propagating the influence of unambiguous motion signals generated at contour terminators or corners. Here, sparse “feature tracking signals” from, e.g., line ends, are amplified to overwhelm numerically superior ambiguous motion signals along line segment interiors. In the fourth, a spatially anisotropic motion grouping process takes place across perceptual space via MT-MST feedback to integrate veridical feature-tracking and ambiguous motion signals to determine a global object motion percept. The fifth property uses the MT-MST feedback loop to convey an attentional priming signal from higher brain areas back to V1 and V2. The model's use of mechanisms such as divisive normalization, endstopping, cross-orientation inhibition, and longrange cooperation is described. Simulated data include: the degree of motion coherence of rotating shapes observed through apertures, the coherent vs. element motion percepts separated in depth during the chopsticks illusion, and the rigid vs. non-rigid appearance of rotating ellipses.

Working Memory Networks for Learning Multiple Groupings of Temporarily Ordered Events: Applications to 3-D Visual Object Recognition

Working memory neural networks are characterized which encode the invariant temporal order of sequential events that may be presented at widely differing speeds, durations, and interstimulus intervals. This temporal order code is designed to enable all possible groupings of sequential events to be stably learned and remembered in real time, even as new events perturb the system. Such a competence is needed in neural architectures which self-organize learned codes for variable-rate speech perception, sensory-motor planning, or 3-D visual object recognition. Using such a working memory, a self-organizing architecture for invariant 3-D visual object recognition is described that is based on the model of Seibert and Waxman [1].

Working Memory Networks for Learning Temporal Order, with Application to 3-D Visual Object Recognition

Working memory neural networks are characterized which encode the invariant temporal order of sequential events. Inputs to the networks, called Sustained Temporal Order REcurrent (STORE) models, may be presented at widely differing speeds, durations, and interstimulus intervals. The STORE temporal order code is designed to enable all emergent groupings of sequential events to be stably learned and remembered in real time, even as new events perturb the system. Such a competence is needed in neural architectures which self-organize learned codes for variable-rate speech perception, sensory-motor planning, or 3-D visual object recognition. Using such a working memory, a self-organizing architecture for invariant 3-D visual object recognition is described. The new model is based on the model of Seibert and Waxman (1990a), which builds a 3-D representation of an object from a temporally ordered sequence of its 2-D aspect graphs. The new model, called an ARTSTORE model, consists of the following cascade of processing modules: Invariant Preprocessor --> ART 2 --> STORE Model --> ART 2 --> Outstar Network.

Enhancing the health-status of processed meats through ingredient manipulation and its effects on sensory and physiochemical product attributes

An important component of this Ph.D. thesis was to determine the European consumers’ views on processed meats and bioactive compounds. Thus a survey gathered information form over 500 respondents and explored their perceptions on the healthiness and purchase-ability for both traditional and functional processed meats. This study found that the consumer was distrustful towards processed meat, especially high salt and fat content. Consumers were found to be very pro-bioactive compounds in yogurt style products but unsure of their feelings on the idea of them in meat based products, which is likely due to the lack of familiarity to these products. The work in this thesis also centred on the applied acceptable reduction of salt and fat in terms of consumer sensory analysis. The products chosen ranged in the degree of comminution, from a coarse beef patty to a more fine emulsion style breakfast sausage and frankfurter. A full factorial design was implemented which saw the production of twenty beef patties with varying concentrations of fat (30%, 40%, 50%, 60% w/w) and salt (0.5%, 0.75%, 1.0%, 1.25%, 1.5% w/w). Twenty eight sausage were also produced with varying concentrations of fat (22.5%, 27.5%, 32.5%, 37.5% w/w) and salt (0.8%, 1%, 1.2%, 1.4%, 1.6%, 2%, 2.4% w/w). Finally, twenty different frankfurters formulations were produced with varying concentrations of fat (10%, 15%, 20%, 25% w/w) and salt (1%, 1.5%, 2%, 2.5%, 3% w/w). From these products it was found that the most consumer acceptable beef patty was that containing 40% fat with a salt level of 1%. This is a 20% decrease in fat and a 50% decrease in salt levels when compared to commercial patty available in Ireland and the UK. For sausages, salt reduced products were rated by the consumers as paler in colour, more tender and with greater meat flavour than higher salt containing products. The sausages containing 1.4 % and 1.0 % salt were significantly (P<0.01) found to be more acceptable to consumers than other salt levels. Frankfurter salt levels below 1.5% were shown to have a negative effect on consumer acceptability, with 2.5% salt concentration being the most accepted (P<0.001) by consumers. Samples containing less fat and salt were found to be tougher, less juicy and had greater cooking losses. Thus salt perception is very important for consumer acceptability, but fat levels can be potentially reduced without significantly affecting overall acceptability. Overall it can be summarised that the consumer acceptability of salt and fat reduced processed meats depends very much on the product and generalisations cannot be assumed. The study of bio-actives in processed meat products found that the reduced salt/fat patties fortified with CoQ10 were rated as more acceptable than commercially available products for beef patties. The reduced fat and salt, as well as the CoQ10 fortified, sausages were found to compare quite well to their commercial counterparts for overall acceptability, whereas commercial frankfurters were found to be the more favoured in comparison to reduced fat and CoQ10 fortified Frankfurters.

Optical micro- and nanofibres: Higher mode generation and particle manipulation studies

This thesis is centred on two experimental fields of optical micro- and nanofibre research; higher mode generation/excitation and evanescent field optical manipulation. Standard, commercial, single-mode silica fibre is used throughout most of the experiments; this generally produces high-quality, single-mode, micro- or nanofibres when tapered in a flame-heated, pulling rig in the laboratory. Single mode fibre can also support higher transverse modes, when transmitting wavelengths below that of their defined single-mode regime cut-off. To investigate this, a first-order Laguerre-Gaussian beam, LG01 of 1064 nm wavelength and doughnut-shaped intensity profile is generated free space via spatial light modulation. This technique facilitates coupling to the LP11 fibre mode in two-mode fibre, and convenient, fast switching to the fundamental mode via computer-generated hologram modulation. Following LP11 mode loss when exponentially tapering 125μm diameter fibre, two mode fibre with a cladding diameter of 80μm is selected fir testing since it is more suitable for satisfying the adiabatic criteria for fibre tapering. Proving a fruitful endeavour, experiments show a transmission of 55% of the original LP11 mode set (comprising TE01, TM01, HE21e,o true modes) in submicron fibres. Furthermore, by observing pulling dynamics and progressive mode-lass behaviour, it is possible to produce a nanofibre which supports only the TE01 and TM01 modes, while suppressing the HE21e,o elements of the LP11 group. This result provides a basis for experimental studies of atom trapping via mode-interference, and offers a new set of evanescent field geometries for sensing and particle manipulation applications. The thesis highlights the experimental results of the research unit’s Cold Atom subgroup, who successfully integrated one such higher-mode nanofibre into a cloud of cold Rubidium atoms. This led to the detection of stronger signals of resonance fluorescence coupling into the nanofibre and for light absorption by the atoms due to the presence of higher guided modes within the fibre. Theoretical work on the impact of the curved nanofibre surface on the atomic-surface van der Waals interaction is also presented, showing a clear deviation of the potential from the commonly-used flat-surface approximation. Optical micro- and nanofibres are also useful tools for evanescent-field mediated optical manipulation – this includes propulsion, defect-induced trapping, mass migration and size-sorting of micron-scale particles in dispersion. Similar early trapping experiments are described in this thesis, and resulting motivations for developing a targeted, site-specific particle induction method are given. The integration of optical nanofibres into an optical tweezers is presented, facilitating individual and group isolation of selected particles, and their controlled positioning and conveyance in the evanescent field. The effects of particle size and nanofibre diameter on pronounced scattering is experimentally investigated in this systems, as are optical binding effects between adjacent particles in the evanescent field. Such inter-particle interactions lead to regulated self-positioning and particle-chain speed enhancements.

Photonic integrated circuit for the manipulation of coherent optical combs

Photonic integration has become an important research topic in research for applications in the telecommunications industry. Current optical internet infrastructure has reached capacity with current generation dense wavelength division multiplexing (DWDM) systems fully occupying the low absorption region of optical fibre from 1530 nm to 1625 nm (the C and L bands). This is both due to an increase in the number of users worldwide and existing users demanding more bandwidth. Therefore, current research is focussed on using the available telecommunication spectrum more efficiently. To this end, coherent communication systems are being developed. Advanced coherent modulation schemes can be quite complex in terms of the number and array of devices required for implementation. In order to make these systems viable both logistically and commercially, photonic integration is required. In traditional DWDM systems, arrayed waveguide gratings (AWG) are used to both multiplex and demultiplex the multi-wavelength signal involved. AWGs are used widely as they allow filtering of the many DWDM wavelengths simultaneously. However, when moving to coherent telecommunication systems such as coherent optical frequency division multiplexing (OFDM) smaller FSR ranges are required from the AWG. This increases the size of the device which is counter to the miniaturisation which integration is trying to achieve. Much work was done with active filters during the 1980s. This involved using a laser device (usually below threshold) to allow selective wavelength filtering of input signals. By using more complicated cavity geometry devices such as distributed feedback (DFB) and sampled grating distributed Bragg gratings (SG-DBR) narrowband filtering is achievable with high suppression (>30 dB) of spurious wavelengths. The active nature of the devices also means that, through carrier injection, the index can be altered resulting in tunability of the filter. Used above threshold, active filters become useful in filtering coherent combs. Through injection locking, the coherence of the filtered wavelengths with the original comb source is retained. This gives active filters potential application in coherent communication system as demultiplexers. This work will focus on the use of slotted Fabry-Pérot (SFP) semiconductor lasers as active filters. Experiments were carried out to ensure that SFP lasers were useful as tunable active filters. In all experiments in this work the SFP lasers were operated above threshold and so injection locking was the mechanic by which the filters operated. Performance of the lasers under injection locking was examined using both single wavelength and coherent comb injection. In another experiment two discrete SFP lasers were used simultaneously to demultiplex a two-line coherent comb. The relative coherence of the comb lines was retained after demultiplexing. After showing that SFP lasers could be used to successfully demultiplex coherent combs a photonic integrated circuit was designed and fabricated. This involved monolithic integration of a MMI power splitter with an array of single facet SFP lasers. This device was tested much in the same way as the discrete devices. The integrated device was used to successfully demultiplex a two line coherent comb signal whilst retaining the relative coherence between the filtered comb lines. A series of modelling systems were then employed in order to understand the resonance characteristics of the fabricated devices, and to understand their performance under injection locking. Using this information, alterations to the SFP laser designs were made which were theoretically shown to provide improved performance and suitability for use in filtering coherent comb signals.

Manipulation of magnetic anisotropy in nanostructures

Of late, the magnetic properties of micro/nano-structures have attracted intense research interest both fundamentally and technologically particularly to address the question that how the manipulation in the different layers of nanostructures, geometry of a patterned structure can affect the overall magnetic properties, while generating novel applications such as in magnetic sensors, storage devices, integrated inductive components and spintronic devices. Depending on the applications, materials with high, medium or low magnetic anisotropy and their possible manipulation are required. The most dramatic manifestation in this respect is the chance to manipulate the magnetic anisotropy over the intrinsic preferential direction of the magnetization, which can open up more functionality particularly for device applications. Types of magnetic anisotropies of different nanostructured materials and their manipulation techniques are investigated in this work. Detail experimental methods for the quantitative determination of magnetic anisotropy in nanomodulated Ni45Fe55 thin film are studied. Magnetic field induced in-plane rotations within the nanomodulated Ni45Fe55 continuous films revealed various rotational symmetries of magnetic anisotropy due to dipolar interactions showing a crossover from lower to higher fold of symmetry as a function of modulation geometry. In a second approach, the control of exchange anisotropy at ferromagnetic (FM) – aniferomagnetic (AFM) interface in multifferoic nanocomposite materials, where two different phase/types of materials were simultaneously synthesized, was investigated. The third part of this work was to study the electroplated thin films of metal alloy nanocomposite for enhanced exchange anisotropy. In this work a unique observation of an anti-clock wise as well as a clock wise hysteresis loop formation in the Ni,Fe solid solution with very low coercivity and large positive exchange anisotropy/exchange bias have been investigated. Hence, controllable positive and negative exchange anisotropy has been observed for the first time which has high potential applications such as in MRAM devices.