Feature Cues Research Articles

Bumblebees spontaneously map location of conspecific using geometry and features

Researchers of vertebrate navigation have investigated and debated intensely the existence of independent processes underlying spatial representation of environmental geometry and featural cues (e.g., landmarks). One of the most well-known behavioural instantiations of this dissociation originates from the differences in the way animals map locations in the surrounding environment with respect to the external cues such as boundaries or landmarks. Invertebrate researchers have joined this debate in the past few years, showing that insects can be trained to encode spatial locations using a variety of cues within small enclosures. In the present paper, we test insects in a non-rewarded, spontaneous navigation task. We observed bumblebees’ spatial navigation in response to the observed location of a conspecific, over manipulations of cue type and arena size. The results indicate that bumblebees spontaneously use both boundaries and features to guide their spatial behaviour. We discuss the similarities and differences in their behaviours to those of vertebrates and speculate on the implications of these findings.

Differential Processing of Vertical Interfeature Relations Due to Real-Life Experience with Personally Familiar Faces

Identification of personally familiar faces is possibly the most complex and likewise efficient task achieved by the human visual system, yet to date the mechanisms underlying this extreme proficiency remain largely unknown. Building on empirical evidence from unfamiliar face processing in healthy populations and neuropsychological patients, the present work aimed to determine the type of information processed differently due to repeated, real-life experience with faces. A modulatory effect of familiarity was observed for processing of vertical interfeature distances, which have been suggested to rely on holistic processing skills. Contrariwise, no such effect was found for processing of information that can be discriminated locally (ie featural cues, interocular distances). The results indicate that familiarity-related advantages in face processing may arise from more efficient, or increased, holistic processing.

Unconscious analyses of visual scenes based on feature conjunctions.

To efficiently process a cluttered scene, the visual system analyzes statistical properties or regularities of visual elements embedded in the scene. It is controversial, however, whether those scene analyses could also work for stimuli unconsciously perceived. Here we show that our brain performs the unconscious scene analyses not only using a single featural cue (e.g., orientation) but also based on conjunctions of multiple visual features (e.g., combinations of color and orientation information). Subjects foveally viewed a stimulus array (duration: 50 ms) where 4 types of bars (red-horizontal, red-vertical, green-horizontal, and green-vertical) were intermixed. Although a conscious perception of those bars was inhibited by a subsequent mask stimulus, the brain correctly analyzed the information about color, orientation, and color-orientation conjunctions of those invisible bars. The information of those features was then used for the unconscious configuration analysis (statistical processing) of the central bars, which induced a perceptual bias and illusory feature binding in visible stimuli at peripheral locations. While statistical analyses and feature binding are normally 2 key functions of the visual system to construct coherent percepts of visual scenes, our results show that a high-level analysis combining those 2 functions is correctly performed by unconscious computations in the brain.

Stimulus-driven attentional capture by subliminal onset cues

In two experiments, we tested whether subliminal abrupt onset cues capture attention in a stimulus-driven way. An onset cue was presented 16 ms prior to the stimulus display that consisted of clearly visible color targets. The onset cue was presented either at the same side as the target (the valid cue condition) or on the opposite side of the target (the invalid cue condition). Because the onset cue was presented 16 ms before other placeholders were presented, the cue was subliminal to the participant. To ensure that this subliminal cue captured attention in a stimulus-driven way, the cue's features did not match the top-down attentional control settings of the participants: (1) The color of the cue was always different than the color of the non-singleton targets ensuring that a top-down set for a specific color or for a singleton would not match the cue, and (2) colored targets and distractors had the same objective luminance (measured by the colorimeter) and subjective lightness (measured by flicker photometry), preventing a match between the top-down set for target and cue contrast. Even though a match between the cues and top-down settings was prevented, in both experiments, the cues captured attention, with faster response times in valid than invalid cue conditions (Experiments 1 and 2) and faster response times in valid than the neutral conditions (Experiment 2). The results support the conclusion that subliminal cues capture attention in a stimulus-driven way.

An age-related deficit in spatial–feature reference memory in homing pigeons (Columba livia)

Age-related memory decline in mammals has been well documented. By contrast, very little is known about memory decline in birds as they age. In the current study we trained younger and older homing pigeons on a reference memory task in which a goal location could be encoded by spatial and feature cues. Consistent with a previous working memory study, the results revealed impaired acquisition of combined spatial–feature reference memory in older compared to younger pigeons. Following memory acquisition, we used cue-conflict probe trials to provide an initial assessment of possible age-related differences in cue preference. Both younger and older pigeons displayed a similarly modest preference for feature over spatial cues.

Goal orientation by geometric and feature cues: spatial learning in the terrestrial toad Rhinella arenarum.

Although of crucial importance in vertebrate evolution, amphibians are rarely considered in studies of comparative cognition. Using water as reward, we studied whether the terrestrial toad, Rhinella arenarum, is also capable of encoding geometric and feature information to navigate to a goal location. Experimental toads, partially dehydrated, were trained in either a white rectangular box (Geometry-only, Experiment 1) or in the same box with a removable colored panel (Geometry-Feature, Experiment 2) covering one wall. Four water containers were used, but only one (Geometry-Feature), or two in geometrically equivalent corners (Geometry-only), had water accessible to the trained animals. After learning to successfully locate the water reward, probe trials were carried out by changing the shape of the arena or the location of the feature cue. Probe tests revealed that, under the experimental conditions used, toads can use both geometry and feature to locate a goal location, but geometry is more potent as a navigational cue. The results generally agree with findings from other vertebrates and support the idea that at the behavioral-level geometric orientation is a conserved feature shared by all vertebrates.

Quantifying surface phonetic variation using acoustic landmarks as feature cues

Acoustic landmarks, which are abrupt spectral changes associated with certain feature sequences in spoken utterances, are highly informative and have been proposed as the initial analysis stage in human speech perception, and for automatic speech recognition (Stevens, JASA 111(4), 2002, 1872–1891). These feature cues and their parameter values also provide an effective tool for quantifying systematic context-governed surface phonetic variation (Shattuck-Hufnagel and Veilleux, ICPhS XVI, 2007, 925–928). However, few studies have provided landmark-based information about the full range of variation in continuous communicative speech. The current study examines landmark modification patterns in a corpus of maptask-elicited speech, hand annotated for whether the landmarks were realized as predicted from the word forms or modified in context. Preliminary analyses of a single conversation (400 s, one speaker) show that the majority of landmarks (about 84%) exhibited the canonical form predicted from their lexical specifications, and that modifications were distributed systematically across segment types. For example, 90% of vowel landmarks (at amplitude/F1 peaks) were realized as predicted, but only 70% of the closures for non-strident fricatives /v/ and /dh/, and 40–50% of /t/ closures and releases. Further quantification of landmark modification patterns will provide useful information about the processing of surface phonetic variation.

Mixed Feature Selection Based Affective Classification in Ink-wash Paintings

As one of the most valuable eastern cultural heritages, ink-wash paintings are attracting more and more concerns from people around the world. In this context, affective understanding plays an important role in analysis, classification, and indexing of Chinese ink-wash paintings. However, due to the semantic gap between low-level features and high-level feelings, that problem is far from addressed. Existing researches on analysis and recognition of the paintings mostly focus on artists’ styles rather than hidden affects or feelings of paintings. Therefore, we propose a mixed feature-based algorithm in this paper to perform the automatic affects analysis and understanding, where a set of affective categories could be identified. After an image is segmented into several sub images by quad tree, a number of feature cues are extracted. We mixed and refined them together by using interclass distance method. The effectively selected cues are applied to SVM classifier to justify which emotion class the image belongs to. Results and evaluations via experiments validate the proposed approach. This method maintains great theoretical significance and practical value in personalized retrieval of ink-wash paintings.

A critical role of holistic processing in face gender perception.

Whether face gender perception is processed by encoding holistic (whole) or featural (parts) information is a controversial issue. Although neuroimaging studies have identified brain regions related to face gender perception, the temporal dynamics of this process remain under debate. Here, we identified the mechanism and temporal dynamics of face gender perception. We used stereoscopic depth manipulation to create two conditions: the front and behind condition. In the front condition, facial patches were presented stereoscopically in front of the occluder and participants perceived them as disjoint parts (featural cues). In the behind condition, facial patches were presented stereoscopically behind the occluder and were amodally completed and unified in a coherent face (holistic cues). We performed three behavioral experiments and one electroencephalography experiment, and compared the results of the front and behind conditions. We found faster reaction times (RTs) in the behind condition compared with the front, and observed priming effects and aftereffects only in the behind condition. Moreover, the EEG experiment revealed that face gender perception is processed in the relatively late phase of visual recognition (200–285 ms). Our results indicate that holistic information is critical for face gender perception, and that this process occurs with a relatively late latency.

Female hummingbirds do not relocate rewards using colour cues

Males generally outperform females in spatial tasks. This difference in spatial performance may reflect differences in cue preference because males often use both spatial cues (distance and direction) and feature cues, whereas females prefer to use feature cues. However, studies in birds are few and results are conflicting. As wild male rufous hummingbirds, Selasphorus rufus, prefer to use spatial cues to relocate a rewarded flower, in the present study we tested free-flying wild female hummingbirds of three different species (rufous, white-eared, Hylocharis leucotis, and magnificent hummingbird, Eugenes fulgens) and males of one species (white-eared) for their cue preference in the same task in which rufous males have been tested previously. Birds were allowed to feed once from a four-flower array in which only one flower was rewarded. When the birds returned, the colour and the spatial cue designating the rewarded flower had been dissociated. Although we had expected females to visit the flower of the correct colour (feature cue) first, during the test phase most of the birds (males and females) went to the correct spatial location (spatial cue). It appears, then, that preference for spatial cues is not specific to males and it seems more likely to depend on the relevance or value of a cue to the solution of the task.

3D Traffic Scene Understanding From Movable Platforms.

In this paper, we present a novel probabilistic generative model for multi-object traffic scene understanding from movable platforms which reasons jointly about the 3D scene layout as well as the location and orientation of objects in the scene. In particular, the scene topology, geometry, and traffic activities are inferred from short video sequences. Inspired by the impressive driving capabilities of humans, our model does not rely on GPS, lidar, or map knowledge. Instead, it takes advantage of a diverse set of visual cues in the form of vehicle tracklets, vanishing points, semantic scene labels, scene flow, and occupancy grids. For each of these cues, we propose likelihood functions that are integrated into a probabilistic generative model. We learn all model parameters from training data using contrastive divergence. Experiments conducted on videos of 113 representative intersections show that our approach successfully infers the correct layout in a variety of very challenging scenarios. To evaluate the importance of each feature cue, experiments using different feature combinations are conducted. Furthermore, we show how by employing context derived from the proposed method we are able to improve over the state-of-the-art in terms of object detection and object orientation estimation in challenging and cluttered urban environments.

A fuzzy logic-based system for the automation of human behavior recognition using machine vision in intelligent environments

The recent years have witnessed significant progress in the automation of human behavior recognition using machine vision in order to realize intelligent environments which are capable of detecting users' actions and gestures so that the needed services can be provided automatically and instantly for maximizing the user comfort and safety as well as minimizing energy. However, the majority of traditional human behavior machine vision-based recognition approaches rely on assumptions (such as known spatial locations and temporal segmentations) or computationally expensive approaches (such as sliding window search through a spatio-temporal volume). Hence, it is difficult for such methods to scale up and handle the high uncertainty levels and complexities available in real-world applications. This paper proposes a novel fuzzy machine vision-based framework for efficient humans' behavior recognition. A model-based feature set is utilized to extract visual feature cues including silhouette slices and movement speed from the human silhouette in video sequences which are analyzed as inputs by the proposed fuzzy system. We have employed fuzzy c-means clustering to learn the membership functions of the proposed fuzzy system. The behavior recognition was implemented via selecting the best candidate's behavior category with the highest output degree as the recognized behavior. We have successfully tested our system on the publicly available Weizmann human action dataset where our fuzzy-based system produced an average recognition accuracy of 94.03 %, which outperformed the traditional non-fuzzy systems based on hidden Markov models and other state-of-the-art approaches which were applied on the Weizmann dataset.

Kenneth Noble Stevens

Kenneth Noble Stevens

Size does not matter, but features do: Clark's nutcrackers (Nucifraga columbiana) weigh features more heavily than geometry in large and small enclosures

Two groups of Clark's nutcrackers (Nucifraga columbiana) were trained to locate a hidden goal which was consistently located at one corner of a fully enclosed rectangular environment with distinctive cues available at each corner. One group was trained in a small enclosure, whereas the second group was trained in a large enclosure. Once the birds were showing accurate search behavior, they were presented with non-reinforced tests in either the same sized environment as training or the novel sized environment, as well as in a square-shaped environment. The birds were able to accurately search at the two geometrically correct corners when the four distinctive features were removed showing that they had encoded geometry. Although accuracy was greater when tested in the same sized environment as during training, accuracy was above chance in both environments. Regardless of the size of training enclosure both groups showed primary control by features along with secondary control by geometry. Furthermore, when the features and geometric cues provided conflicting information as to the goal location, both groups weighed featural cues over geometry, and this was independent of whether the size of the testing environment was maintained or manipulated. These results show that for Clark's nutcrackers the size of the environment had little effect on the weighing of featural and geometric cues. Furthermore, although nutcrackers encoded both features and geometry, when spatial cues provided discrepant information as to the goal location, nutcrackers relied primarily on features.This article is part of a Special Issue entitled: CO3 2013.

The Role of Surface Color Feature Cue in Object Persistence under Two Different Spatiotemporal Conditions

The Role of Surface Color Feature Cue in Object Persistence under Two Different Spatiotemporal Conditions

The influence of left and right hemisphere brain damage on configural and featural processing of affective faces

The literature about the lateralization of facial emotion perception according to valence (positive, negative) is conflicting; investigating the underlying processes may shed light on why some studies show right-hemisphere dominance across valence and other studies demonstrate hemispheric differences according to valence. This is the first clinical study to examine whether the use of configural and featural cues underlies hemispheric differences in affective face perception. Right brain-damaged (RBD; n = 17), left brain-damaged (LBD; n = 17) and healthy control (HC; n = 34) participants completed an affective face discrimination task that tested configural processing using whole faces and featural processing using partial faces. No group differences in expression perception according to valence or processing strategy were found. Across emotions, the RBD group was less accurate than the HC group in discriminating whole faces, whilst the RBD and LBD groups were less accurate than HCs in discriminating partial faces. This suggests that the right hemisphere processes facial expressions from configural and featural information, whereas the left hemisphere relies more heavily on featural facial information.

Moving Object Segmentation Using Level Set Method with Shape Prior, Color and Texture

This paper presents a method for the combination of different feature cues in a level set based moving object segmentation framework. To distinguish object from background, motion detection is firstly adopted to locate object position and obtain coarse shape prior. Moreover, the color and texture feature descriptors that represent object contour are designed in this dissertation. Then the finer segmentation solution based on the color and texture difference between the object and background is proposed, which avoids the invalid feature components to hamper segmentation and improves the accuracy. Extensive experiments have been carried out on surveillance video sequences to validate the proposed method.

A MOTION ATTENTION MODEL BASED ON RARITY WEIGHTING AND MOTION CUES IN DYNAMIC SCENES

Nowadays, motion attention model is a controversial topic in the biological computer vision area. The computational attention model can be decomposed into a set of features via predefined channels. Here we designed a bio-inspired vision attention model, and added the rarity measurement onto it. The priority of rarity is emphasized under the assumption of weighting effect upon the features logic fusion. At this stage, a final saliency map at each frame is adjusted by the spatiotemporal and rarity values. By doing this, the process of mimicking human vision attention becomes more realistic and logical to the real circumstance. The experiments are conducted on the benchmark dataset of static images and video sequences. We simulated the attention shift based on several dataset. Most importantly, our dynamic scenes are mostly selected from the objects moving on the highway and dynamic scenes. The former one can be developed on the detection of car collision and will be a useful tool for further application in robotics. We also conduct experiment on the other video clips to prove the rationality of rarity factor and feature cues fusion methods. Finally, the evaluation results indicate our visual attention model outperforms several state-of-the-art motion attention models.

Space, color, and direction of movement: How do they affect attention?

Paying attention improves performance, but is this improvement regardless of what we attend to? We explored the differences in performance between attending to a location and attending to a feature when perceiving global motion. Attention was first cued to one of four locations that had coherently moving dots, while the remaining three had randomly moving distracter dots. Participants then viewed a colored display, wherein the color of the coherently moving dots was cued instead of location. In the third task, participants identified the location that had a particular cued direction of motion. Most observers reported reductions of motion threshold in all three tasks compared to when no cue was provided. However, the attentional bias generated by location cues was significantly larger than the bias resulting from feature cues of direction or color. This effect is consistent with the idea that attention is largely controlled by a fronto-parietal network where spatial relations are preferentially processed. On the other hand, color could not be used as a cue to focus attention and integrate motion. This finding suggests that color relies heavily on processing by ventral temporal cortical areas, which may have little control over the global motion areas in the dorsal part of the brain.

Apparent color–orientation bindings in the periphery can be influenced by feature binding in central vision

A previous study reported the misbinding illusion in which visual features belonging to overlapping sets of items were erroneously integrated (Wu, Kanai, & Shimojo, 2004, Nature, 429, 262). In this illusion, central and peripheral portions of a transparent motion field combined color and motion in opposite fashions. When observers saw such stimuli, their perceptual color–motion bindings in the periphery were re-arranged in such a way as to accord with the bindings in the central region, resulting in erroneous color–motion pairings (misbinding) in peripheral vision. Here we show that this misbinding illusion is also seen in the binding of color and orientation. When the central field of a stimulus array was composed of objects that had coherent (regular) color–orientation pairings, subjective color–orientation bindings in the peripheral stimuli were automatically altered to match the coherent pairings of the central stimuli. Interestingly, the illusion was induced only when all items in the central field combined color and orientation in an orthogonal fashion (e.g. all red bars were horizontal and all green bars were vertical). If this orthogonality was disrupted (e.g. all red and green bars were horizontal), the central field lost its power to induce the misbinding illusion in the peripheral stimuli. The original misbinding illusion study proposed that the illusion stemmed from a perceptual extrapolation that resolved peripheral ambiguity with clear central vision. However, our present results indicate that visual analyses of the correlational structure between two features (color and orientation) are critical for the illusion to occur, suggesting a rapid integration of multiple featural cues in the human visual system.