Actual Recognition Research Articles

Image Access, the Semantic Gap, and Social Tagging as a Paradigm Shift

The recent phenomenon of “social tagging” or “distributed indexing” raises a number of questions regarding long-held beliefs and practices of the classification and indexing community. This workshop paper covers several of these issues, such as locus of authority, control, and meaning, and suggests we may be observing the emergence of a new paradigm of knowledge organization. The Semantic Gap The “semantic gap” is mentioned frequently in the literature of image access. The term originated in computer science (Smeulders et al., 2000) and is still used in the CS literature today to refer to the difference between two descriptions of an object using different languages, specifically the difference between a human-readable description and a computational representation. In a computational representation, a simple image of an object moves from the level of individual pixels to assemblages of image primitives such as color, shape/region, and texture, to the assemblage and recognition of an object, at least at the level of a simple “basic object.” Object recognition necessitates a level of “understanding” of what is being represented; this is achieved by inferring what different combinations of primitives may represent, e.g. black spots or black stripes on an orangetan background and an assemblage of potential “leg,” “body,” “tail,” and “head” shapes, perhaps combined with “nature colors,” could be interpreted as a leopard or tiger. The process is fraught with stumbling blocks such as occlusion, angle of view, scale, shadow, and lack of uniqueness, to mention a few. However, it is at the level of object recognition that human image description often begins. With the development of automated methods of content-based image retrieval the term “semantic gap” has come to refer to the larger issue of the gap between these image primitives, or low-level features, and the context-sensitive meanings human beings associate with these. This brings us beyond object recognition and understanding into more abstract levels of semantic meaning, and the meanings or emotions associated with even one image can be many, and can vary across time and place. For a human, recognition of familiar objects is instantaneous, and an image of a tiger, once recognized, can represent multiple concepts such power, ferocity, freedom (or a lack thereof, as in a caged tiger), or even endangered species. These concepts form a gestalt of the object, gestalt being a German word roughly translated as a complete pattern or configuration. There are three parts to a definition of gestalt: a thing, its context or environment, and the relationship between them (Wymore 2002). Studies in cognitive science suggest that this gestalt may have equal importance with sensory stimuli in the process of actual recognition of the object.

Modeling and Predicting Face Recognition System Performance Based on Analysis of Similarity Scores

This paper presents methods of modeling and predicting face recognition (FR) system performance based on analysis of similarity scores. We define the performance of an FR system as its recognition accuracy, and consider the intrinsic and extrinsic factors affecting its performance. The intrinsic factors of an FR system include the gallery images, the FR algorithm, and the tuning parameters. The extrinsic factors include mainly query image conditions. For performance modeling, we propose the concept of "perfect recognition," based on which a performance metric is extracted from perfect recognition similarity scores (PRSS) to relate the performance of an FR system to its intrinsic factors. The PRSS performance metric allows tuning FR algorithm parameters offline for near optimal performance. In addition, the performance metric extracted from query images is used to adjust face alignment parameters online for improved performance. For online prediction of the performance of an FR system on query images, features are extracted from the actual recognition similarity scores and their corresponding PRSS. Using such features, we can predict online if an individual query image can be correctly matched by the FR system, based on which we can reduce the incorrect match rates. Experimental results demonstrate that the performance of an FR system can be significantly improved using the presented methods.

On the Problem for Damage Detection of Vibrating Cracked Plates

This work was motivated by the recent NATO funded research on preventing disasters from collapse and improving the safety of aircraft structures. It considers the problem for vibrationbased damage detection in aircraft panels modelled as isotropic plates. The explored method does not use any assumptions of model or linearity, it is simply based on pure signal analysis of the vibration response of plates. FE modelling is used to model the plate’s dynamic response in its intact and in its damaged state. The signals obtained are analysed using multivariate analysis applied in the measured frequency domain. This reduces the data dimensionality and is expected to have a clustering effect. At this stage the measured data is transformed into features – new variables- which have smaller dimension than the initial ones and make the categories more distinguishable. Then a very simple pattern recognition (PR) method is applied to discriminate between the two categories of data -data coming from the undamaged plate and data coming from the damaged plate. This is the second stage when the obtained features are used for the actual recognition between the defined categories. The paper suggests the use of the Karhunen-Loeve transform in order to extract features from the measured frequency response functions of the plate. When the data dimensionality is brought down to two the response of the plate can be visualised. The clustering effect on the features coming from undamaged plate and those from the damaged is obvious.

A smart camera with gesture recognition and SLAM capabilities for mobile robots

This paper presents a smart camera for a real-time gesture recognition system. The smart camera has been designed and implemented as a system-on-a-chip (SoC), using reconfigurable computing technology. In this system the gesture images are captured by means of a CMOS digital camera. After some pre-processing steps, those are sent to a fault tolerant module (FTM) for the actual recognition process. The FTM implements a RAM-based neural network, using three knowledge bases. In addition, a majority voting technique is used to improve the confidence level in the recognition step. A number of experiments using a prototype implementation and selected gestures resulted in a rate of 100% true recognition. The system also showed to be robust, and flexible as new gesture patterns can be easily included by means of the on-chip training capabilities. An application for this systems is also presented, consisting of four smart cameras used in simultaneous localization and mapping (SLAM) tasks for robotics.

Interpersonal Recognition and Responsiveness to Relevant Differences

This essay defends a three‐dimensional response‐model theory of recognition of persons, and discusses the related phenomenon of recognition of reasons, values and principles. The theory is three‐dimensional in endorsing recognition of the equality of persons and two kinds of relevant differences: merits and special relationships. It defends a ‘response‐model’ which holds that adequacy of recognition of persons is a matter of adequate responsiveness to situation‐specific reasons and requirements. This three‐dimensional response‐model is compared to Peter Jones’s view, which draws the distinction between status and merit recognition, and mediated and unmediated recognition. The essay discusses a number of questions related to how recognition of situation‐specific reasons, and more general values and principles, is related to recognition of persons. The three‐dimensional response‐model of recognition of persons is in principle compatible with a constructivist view, which holds that the validity of values and principles is dependent on acknowledgement or endorsement. But even if one is a realist on that issue and thinks of validity as independent of acknowledgement, acknowledging relevant values, reasons and principles is a hugely important precondition of actual interpersonal recognition. The essay analyses these connections.

Psychothérapie et phénomènes de langage : une application clinique de la phénoménologie

The question is what possible interaction can be found between a personal interest in phenomenological philosophy and the practice of psychotherapy? To begin, applying a phenomenological thought process enables one to point out the distinction between the symptoms, rooted in reasoning by inference, a by-product of somatic medicine, and the phenomenon, as sensed by the psychiatrist, often without actual recognition. The intuitive grasp of phenomena requires the qualification of a sustained distancing as regards any established knowledge, which puts one in the position of being surprised by events, just as you can let yourself be surprised by children. In this way, it can be illuminating to compare the “sense of realisation” experienced through the psychotherapeutic dialogue with the account of a dream such as a child may supply. This is the kind of discourse which may witness the birth of “meaning in the making”, lack completion, which differs from the intentional meaning of acts of consciousness, such as traditionally described in phenomenology. In its manifestations, this meaning can be related to artistic significances, and particularly those, which are displayed, live during of an improvisation. Finally, applying a phenomenological method makes it possible to reveal as a true moment of crisis the moment when the continuity of discourse is interrupted, for example when experiencing boredom during an exchange. Such an interruption brings to light a fundamental feature of time and affectivity.

You need hands [fingerprint-based biometrics

Advanced optics promise to transform the sometimes shaky business of fingerprint-based biometrics by overcoming various spoofing techniques. If two US-based companies, Lumidigm and Aprilis, have anything to do with it, the biometric community will find its answers in better optoelectronic design. What is surprising, is that this may involve extending the optical system beyond the detector to the actual fingerprint database and recognition system.

Targeting of olfactory neurons.

Olfactory sensory neurons detect an enormous variety of small volatile molecules with extremely high sensitivity and specificity. The actual recognition and discrimination of odorous compounds is accomplished by specific receptor proteins located in the ciliary membrane of the sensory neurons. Axonal connections into the olfactory bulb, the first relay station for odor processing in the brain, are organized such that all neurons expressing the same odorant receptor converge their axons onto common glomeruli which are located at similar positions in all individuals from one species. For the establishment of this precise targeting of olfactory axons to their appropriate glomeruli, combinatorial functions of axon-associated cell adhesion molecules and odorant receptor proteins appear to be required. Odorants that stimulate distinct receptor cell populations will thereby activate a specific combination of glomeruli in the bulb; this characteristic activity pattern may be used by the system to encode the quality of a particular odorant.

Neural substrates for the recognition of newly learned faces: a functional MRI study

Face recognition is critical to the appreciation of our social and physical relations. Functional magnetic resonance imaging (fMRI) was used to identify brain regions involved in the recognition of newly learned faces. Two experiments were conducted. Experiment 1 contrasted a fixation control task with a face recognition task in which subjects were exposed solely to previously viewed faces (all-target). Experiment 2 compared a fixation control with another face recognition task in which subjects were presented with both novel and viewed faces (half-target). Compared to the fixation control, the all-target face recognition was associated with activation in the bilateral occipital and occipitotemporal regions, whereas the half-target face recognition produced activation in the right parietal and prefrontal regions, in addition to the occipital and occipitotemporal. The all-target minus half-target comparison revealed significant activation in the bilateral fusiform gyrus, suggesting stronger fusiform activity during the all-target than the half-target face recognition. The half-target minus all-target comparison showed significant activation in the superior and inferior parietal lobules and several regions in the right frontal lobe. These findings demonstrated that the bilateral fusiform gyrus is involved, not only in face perception, but in a certain aspect of face recognition memory and that this aspect is related to the actual recognition of previously viewed faces rather than the processing of novel ones, which results are consistent with previous lesion work. The right parietal and frontal regions, in contrast, are differentially more associated with the processes related to the detection of novel faces or retrieval effort.

Sequence-specific High Mobility Group Box Factors Recognize 10–12-Base Pair Minor Groove Motifs

Sequence-specific High Mobility Group Box Factors Recognize 10–12-Base Pair Minor Groove Motifs

Large vocabulary speech recognition with multispan statistical language models

Multispan language modeling refers to the integration of various constraints, both local and global, present in the language. It was recently proposed to capture global constraints through the use of latent semantic analysis, while taking local constraints into account via the usual n-gram approach. This has led to several families of data-driven, multispan language models for large vocabulary speech recognition. Because of the inherent complementarity in the two types of constraints, the multispan performance, as measured by perplexity, has been shown to compare favorably with the corresponding n-gram performance. The objective of this work is to characterize the behavior of such multispan modeling in actual recognition. Major implementation issues are addressed, including search integration and context scope selection. Experiments are conducted on a subset of the Wall Street Journal (WSJ) speaker-independent, 20000-word vocabulary, continuous speech task. Results show that, compared to standard n-gram, the multispan framework can lead to a reduction in average word error rate of over 20%. The paper concludes with a discussion of intrinsic multi-span tradeoffs, such as the influence of training data selection on the resulting performance.

Toll and immunity in Drosophila

In Drosophila, the Toll signaling pathway regulates the antifungal host defense. In this immune response, when the ligand Spaetzle (Spz) binds to the Toll receptor, the antimicrobial peptide drosomycin is up-regulated. Levashina et al. now add Spn43Ac, a member of the serpin family of serine protease inhibitors, to the Toll cascade. When Spn43Ac is eliminated, Spz is continuously cleaved, which leads to the constitutive activation of Toll signaling; Spn43Ac negatively regulates Toll. In bacterial infection, the lipopolysaccharide (LPS) component of the Gram-negative bacterial cell wall is sensed by the host cells, which triggers an immune response involving the Toll pathway. Toll has been reported to act as a pattern recognition receptor in mammals for LPS. Because the Spn43Ac is located upstream of Toll in Drosophila, Toll does not appear to be its pattern recognition signal. The actual pattern recognition signal in Drosophila is still unknown. Levashina, E.A., Langley, E., Green, C., Gubb, D., Ashburner, M., Hoffman, J.A., Reichart, J-M. (1999) Constitutive activation of Toll-mediated antifungal defense in serpin-deficient Drosophila. Science 285 : 1917-1919. [Abstract] [Full Text]

PLANNING SEQUENCES OF VIEWS FOR 3-D OBJECT RECOGNITION AND POSE DETERMINATION

We present a method for planning sequences of views for recognition and pose (orientation) determination of 3-D objects of arbitrary shape. The approach consists of a learning stage in which we derive a recognition and pose identification plan and a stage in which actual recognition and pose identification take place. In the learning stage, the objects are observed from all possible views and each view is characterized by an extracted feature vector. These vectors are then used to structure the views into clusters based on their proximity in the feature space. To resolve the remaining ambiguity within each of the clusters, we designed a strategy which exploits the idea of taking additional views. We developed an original procedure which analyzes the transformation of individual clusters under changing viewpoints into several smaller clusters. This results in an optimal next-view planning when additional views are necessary to resolve the ambiguities. This plan then guides the actual recognition and pose determination of an unknown object in an unknown pose.

Multi-stroke relaxation matching method for handwritten chinese character recognition

Since Chinese characters can be represented by a set of basic line segments called sub-strokes, sub-strokes are often used as features to recognize handwritten Chinese Characters. However, the number of sub-strokes needed to evaluate a character are different due to handwriting variations and variations in the stroke extraction process. These differences in representation of the same character lower character recognition rate. A preliminary method to solve this problem is to merge several sub-strokes into a complete stroke. However, it is difficult to merge several sub-strokes into a correct and complete stroke because the Chinese character is not known in advance. In this paper, we propose a multi-stroke relaxation matching method to solve this problem. The proposed matching method can be divided into two parts; one is the multi-stroke relaxation process and the other is the multi-stroke select-match-pair process. The multi-stroke relaxation process will determine the optimal matching relations from the probability of each possible matching pair of sub-strokes and allow more than one of the sub-strokes to match with one or more additional sub-strokes by combining the merging steps into the relaxation process. The multi-stroke select-match-pair process is used to determine the stroke matching relation between the input and reference characters. Some experiments will be conducted to show the feasibility and correctness of the proposed algorithm. From the experimental results, we will prove that the proposed algorithm can solve the matching problem of different numbers of sub-strokes caused by handwriting variations and the stroke extraction process. For 2000 daily used Chinese characters, the actual recognition rate is 93.8 % and the cumulative recognition rate of the first five candidates is 98.9%.

Recognition in Excitable Chemical Reactor Networks. Experiments and Model-Simulations

A reactor network consisting of four electrically coupled Belousov−Zhabotinsky (BZ) oscillators is used for the recognition of chemical oscillation patterns. The four reactors show identical period-1 oscillations. The application of a constant electrical current drives the period-1 oscillations into an excitable nodal steady state via a saddle-node infinite period (SNIPER) bifurcation. Global electric coupling is achieved by periodic external pulses applied simultaneously to all four nodal steady states. Thus, any two of eight possible oscillation patterns can be established as initial conditions. Local electrical coupling according to a programming rule enables the network to carry out the actual recognition process of the two encoded patterns. As a result, the two encoded patterns are immediately selected if offered as initial conditions. A pattern with one error is corrected within one or two oscillations; however, patterns with two errors are rejected by this experimental network. Numerical simulation...

Theta synchronization and alpha desynchronization in a memory task.

In the present study, we examined the hypothesis that episodic encoding and retrieval processes are primarily reflected by a task-related increase in theta power. Individuals performed a recognition task with a total of 192 words. The electroencephalogram was recorded during the study and recognition phase. The results show that only those words that were later correctly recognized produced a significant increase in theta power during encoding. During the actual recognition processes too, a significant theta synchronization (increase in band power) was found for correctly remembered words only. In contrast to the theta band, remembered and not remembered words revealed a complex pattern of desynchronization in the lower and upper alpha band that was different during encoding and recognition.

Identities in conflict: The Aboriginal question and the politics of recognition in Quebec

Abstract The will to have one's own particular identity universally acknowledged and respected is now one of the most significant determinants of the sociopolitical dynamic of contemporary modern societies. However, the actual recognition of difference is more often than not an unsatisfactory and disappointing experience for many of the groups and communities striving for that recognition. The recent experience of Canadian Aboriginal peoples is a case in point. This article looks at the struggle of Aboriginal nations for recognition in one particular region of Canada, the province of Quebec. Indeed, Quebec offers a revealing case‐study of the peculiarities of the politics of recognition. For over three decades now, the French‐speaking majority of Quebec has striven to assert its ethnocultural distinctiveness and laid claims to a special status within the Canadian political and constitutional framework. Today, both Quebecers and Aboriginals are recognition seekers within the Canadian polity. As a result, their relationship is inevitably marked by their respective but competing attempt to draw the attention of the Canadian state to their particular identity claims. In recent years, this has led to a highly conflictual dynamic which considerably strains any hope of social and cultural coexistence. The article examines the particularities of the politics of recognition in the Quebec context. Its objective is twofold. First, it should serve to shed some light on the manifestations of Aboriginal ethnonationalism in Canada. Secondly, it seeks to illustrate the limitations and paradoxical nature of the politics of recognition within a liberal framework.

Supervised Factorial Learning

Factorial learning, finding a statistically independent representation of a sensory “image”—a factorial code—is applied here to solve multilayer supervised learning problems that have traditionally required backpropagation. This lends support to Barlow's argument for factorial sensory processing, by demonstrating how it can solve actual pattern recognition problems. Two techniques for supervised factorial learning are explored, one of which gives a novel distributed solution requiring only positive examples. Also, a new nonlinear technique for factorial learning is introduced that uses neural networks based on almost reversible cellular automata. Due to the special functional connectivity of these networks—which resemble some biological microcircuits—learning requires only simple local algorithms. Also, supervised factorial learning is shown to be a viable alternative to backpropagation. One significant advantage is the existence of a measure for the performance of intermediate learning stages.

3D object recognition from 2D images using geometric hashing

Abstract In this paper, a general technique for model-based recognition is discussed, called Geometric Hashing. Its purpose is to identify an object in the scene, together with its position and orientation. This technique is based on an intensive preprocessing stage, done off-line, where transformation invariant features of the models are indexed into a hash table. This makes the actual recognition particularly efficient. The algorithm stands out for its high inherent parallelism and its ability to deal with occluded scenes. This paper focuses on the use of Geometric Hashing for the case of 3D object recognition from 2D images. An efficient method to represent a 3D model by its 2D projections is proposed. Results are presented of experiments on random data and 3D objects. It has been found that distinguishing between different types of features in a model or scene results in a very efficient implementation of Geometric Hashing using a multidimensional hash table. The filtering ratio of this scheme turns out to be high enough to allow raliable recognition with the corerct feature correspondence between model and scene. The algorithm performed succesfully in dealing with scenes with up to 50% of occlusion and performed at speeds in the order of one second on a SPARC station.

Cis-acting sequences required for encapsidation of duck hepatitis B virus pregenomic RNA

Hepadnavirus reverse transcription requires that pregenomic RNA first be selectively packaged into a cytoplasmic core particle. This process presumably requires the presence of specific recognition sequences on the pregenomic RNA. To define the cis-acting sequences required for pregenome encapsidation in the duck hepatitis B virus (DHBV), we assayed the packaging efficiency of a series of pregenomic RNA deletion mutants and hybrid DHBV/lacZ fusion transcripts. The 5' boundary of the packaging signal lies within the precore region, starting approximately 35 nucleotides from the cap site of pregenomic RNA; thus, the DR1 sequence required for proper viral DNA synthesis is not included in this signal. To define the 3' boundary of the encapsidation signal, fusion transcripts bearing foreign (lacZ) sequences fused to DHBV at different sites 3' to the pregenomic RNA start site were examined. A surprisingly large region of the DHBV genome proved to be required for packaging of such chimeras, which are efficiently encapsidated only when they contain the first 1,200 to 1,400 nucleotides of DHBV pregenomic RNA. However, mutant genomes bearing insertions within this region are packaged efficiently, making it likely that the actual recognition elements for encapsidation are smaller discontinuous sequences located within this region.