Attention in the processing of complex visual displays: detecting features and their combinations.

In Earth observation, cloud severely affects the interpretation of optical satellites generated high-resolution images. Cloud-free optical images are vital for downstream tasks such as semantic segmentation and object detection. Thus, the elimination of clouds from optical imagery has emerged as a significant topic in remote sensing. Currently, most existing methods are proposed to leverage the texture information from auxiliary synthetic aperture radar (SAR) images to restore cloud-free images via direct channel merging. However, such a unified feature extraction approach often neglects the inherent distribution disparity between SAR and optical images—the result of differing imaging principles-potentially leading to significant feature loss. To this end, we introduce a network by jointing SAR and optical images multimodal and cross-attention network (MMCANet) to effectively extract multiscale contextual features from SAR imagery and integrate them with optical features. Specifically, instead of simple concatenation of the channels of SAR and optical images, we obtain high-dimensional features from them through independent feature extractors. The integration of these features is facilitated by a cross-attention mechanism that provides a more fine-grained amalgamation of information. Meanwhile, an atrous spatial pyramid pooling (ASPP) module is introduced into the integration of high-level features, which captures multiscale contextual information around clouded areas. In addition, we propose four advanced remote sensing image restoration algorithms that approach image restoration as a series of subtasks, gradually eliminating clouds to enhance performance. Comprehensive assessments show that MMCANet performs well on the SEN 12 MS-CR dataset with peak signal-to-noise ratio (PSNR) of 39.8871, structural similarity index (SSIM) of 0.9672, mean absolute error (MAE) of 0.0081, and spectral angle mapper (SAM) of 2.9884.

Decision letter: Visual and motor signatures of locomotion dynamically shape a population code for feature detection in Drosophila

Perception of novel objects is of enormous importance in our lives. People have to perceive or understand novel objects when seeing an original painting, admiring an unconventional construction, and using an inventive device. However, very little is known about neural mechanisms underlying the perception for novel objects. Perception of novel objects relies on the integration of unusual features of novel objects in order to identify what such objects are. In the present study, functional Magnetic Resonance Imaging (MRI) was employed to investigate neural correlates of perception of novel objects. The neuroimaging data on participants engaged in novel object viewing versus ordinary object viewing revealed that perception of novel objects involves significant activation in the left precuneus (Brodmann area 7) and the right visual cortex. The results suggest that the left precuneus is associated with the integration of unusual features of novel objects, while the right visual cortex is sensitive to the detection of such features. Our findings highlight the left precuneus as a crucial component of the neural circuitry underlying perception of novel objects.

Major depressive disorder (MDD) is a complex psychiatric disorder that affects the lives of hundreds of millions of individuals around the globe. Even today, researchers debate if morphological alterations in the brain are linked to MDD, likely due to the heterogeneity of this disorder. The application of deep learning tools to neuroimaging data, capable of capturing complex non-linear patterns, has the potential to provide diagnostic and predictive biomarkers for MDD. However, previous attempts to demarcate MDD patients and healthy controls (HC) based on segmented cortical features via linear machine learning approaches have reported low accuracies. In this study, we used globally representative data from the ENIGMA-MDD working group containing 7,012 participants from 30 sites (N=2,772 MDD and N=4,240 HC), which allows a comprehensive analysis with generalizable results. Based on the hypothesis that integration of vertex-wise cortical features can improve classification performance, we evaluated the classification of a DenseNet and a Support Vector Machine (SVM), with the expectation that the former would outperform the latter. As we analyzed a multi-site sample, we additionally applied the ComBat harmonization tool to remove potential nuisance effects of site. We found that both classifiers exhibited close to chance performance (balanced accuracy DenseNet: 51%; SVM: 53%), when estimated on unseen sites. Slightly higher classification performance (balanced accuracy DenseNet: 58%; SVM: 55%) was found when the cross-validation folds contained subjects from all sites, indicating site effect. In conclusion, the integration of vertex-wise morphometric features and the use of the non-linear classifier did not lead to the differentiability between MDD and HC. Our results support the notion that MDD classification on this combination of features and classifiers is unfeasible. Future studies are needed to determine whether more sophisticated integration of information from other MRI modalities such as fMRI and DWI will lead to a higher performance in this diagnostic task.

Dependence in visual feature processing was studied with two identification experiments using briefly exposed stimuli. The basic stimuli consisted of two orthogonal line segments which formed either one of the four angles of a square, and the subject had to identify the location of the horizontal and vertical line segment making up a stimulus. In the first experiment, the two orthogonal line segments were sometimes separated by a gap. In the second experiment, either a 45° diagonal or a 135° diagonal or both were added to the right angles. With the observed response frequencies two forms of independence in feature detection were tested. Feature detection is called state independent if the detection of a feature is independent of the detection of another feature. According to context independence the detection of a feature is independent of the orientation and location of other features in the stimulus. Feature detection was shown to be context dependent whereas the hypothesis of state independence was not rejected, and state independence was unrelated to the size of the gap between the two orthogonal line segments in the first experiment. It was argued that the context effects obtained (context dependence) had occurred early in processing during the automatic activation of representational units corresponding to features before controlled search, interpretation and decision processes became involved.

Object files play important roles in visual cognition, one of which is feature integration. However, their internal structure remains unknown. The current study investigated how multiple features of an object—color and shape—are integrated using a modified object-reviewing paradigm. In particular, we tested whether independent features or color-shape conjunction are bound to the file. A preview of two-colored objects was followed by a linking display. Then, a target was presented to judge whether the target contains any feature of the previews, regardless of their locations. Conditions in which both color and shape are matched with different corresponding preview locations were compared. If features are independently bound, facilitation in response time (RT) will increase additively as the number of location-shared features. If features are bundled, the RT facilitation occurs only when both features share the location. A series of experiments different in stimuli, the number of placeholders, and object motion showed different patterns of mean RT, but RT distribution analysis indicated that estimated nondecision time is consistent with the feature bundle hypothesis. The conjunction-based object preview benefit suggests that in object file formation, color and shape form a perceptual unit, and the unit is kept through object motion.

TV white spaces exploitation for multimedia signal distribution

We propose a method to control a process of feature detection (flexible feature detection) and to produce different representations, depending on given targets. This feature detection is realized by controlling several types of information called structural information, corresponding to different features. The structural information method was applied to simple pattern classification for demonstrating intuitively how the method controls internal representations. Then the method was applied to an alphabet feature detection problem. In this problem we show that several different features can be detected by changing the structural information. Finally in a consonant cluster problem, it is shown that explicit features can be detected by the flexible control of the structural information.

Simultaneous localization and mapping (SLAM) is one of the cornerstones of autonomous navigation systems in robotics and the automotive industry. Visual SLAM (V-SLAM), which relies on image features, such as keypoints and descriptors to estimate the pose transformation between consecutive frames, is a highly efficient and effective approach for gathering environmental information. With the rise of representation learning, feature detectors based on deep neural networks (DNNs) have emerged as an alternative to handcrafted solutions. This work examines the integration of sparse learned features into a state-of-the-art SLAM framework and benchmarks handcrafted and learning-based approaches by comparing the two methods through in-depth experiments. Specifically, we replace the ORB detector and BRIEF descriptor of the ORBSLAM3 pipeline with those provided by Superpoint, a DNN model that jointly computes keypoints and descriptors. Experiments on three publicly available datasets from different application domains were conducted to evaluate the pose estimation performance and resource usage of both solutions.

Spiking neurons can perform spatiotemporal feature detection by nonlinear\nsynaptic and dendritic integration of presynaptic spike patterns.\nMulticompartment models of non-linear dendrites and related neuromorphic\ncircuit designs enable faithful imitation of such dynamic integration\nprocesses, but these approaches are also associated with a relatively high\ncomputing cost or circuit size. Here, we investigate synaptic integration of\nspatiotemporal spike patterns with multiple dynamic synapses on point-neurons\nin the DYNAP-SE neuromorphic processor, which offers a complementary\nresource-efficient, albeit less flexible, approach to feature detection. We\ninvestigate how previously proposed excitatory--inhibitory pairs of dynamic\nsynapses can be combined to integrate multiple inputs, and we generalize that\nconcept to a case in which one inhibitory synapse is combined with multiple\nexcitatory synapses. We characterize the resulting delayed excitatory\npostsynaptic potentials (EPSPs) by measuring and analyzing the membrane\npotentials of the neuromorphic neuronal circuits. We find that biologically\nrelevant EPSP delays, with variability of order 10 milliseconds per neuron, can\nbe realized in the proposed manner by selecting different synapse combinations,\nthanks to device mismatch. Based on these results, we demonstrate that a single\npoint-neuron with dynamic synapses in the DYNAP-SE can respond selectively to\npresynaptic spikes with a particular spatiotemporal structure, which enables,\nfor instance, visual feature tuning of single neurons.\n

Presently, cognitive Internet of Things (CIoT) with cloud computing (CC) enabled intelligent healthcare models are developed, which enables communication with intelligent devices, sensor modules, and other stakeholders in the healthcare sector to avail effective decision making. On the other hand, Alzheimer disease (AD) is an advanced and degenerative illness which injures the brain cells, and its earlier detection is necessary for suitable interference by healthcare professional. In this aspect, this paper presents a new Oriented Features from Accelerated Segment Test (FAST) with Rotated Binary Robust Independent Elementary Features (BRIEF) Detector (ORB) with optimal artificial neural network (ORB-OANN) model for AD diagnosis and classification on the CIoT based smart healthcare system. For initial pre-processing, bilateral filtering (BLF) based noise removal and region of interest (RoI) detection processes are carried out. In addition, the ORB-OANN model includes ORB based feature extractor and principal component analysis (PCA) based feature selector. Moreover, artificial neural network (ANN) model is utilized as a classifier and the parameters of the ANN are optimally chosen by the use of salp swarm algorithm (SSA). A comprehensive experimental analysis of the ORB-OANN model is carried out on the benchmark database and the obtained results pointed out the promising outcome of the ORB-OANN technique in terms of different measures.

In this paper, we present a face registration approach, in which alignment is done by minimizing the closest distance at the classification step. This method eliminates the need of a feature localization step that exists in traditional face recognition systems and formulates alignment as an optimization process during classification. In other words, instead of performing a separate facial feature localization step and localizing facial features according to some type of feature matching score, in the proposed method, alignment is done by directly optimizing the classification score. Moreover, a feature detector can still be integrated to the system. In this case, the output of the feature detector is used as the initial point of the optimization process. Results of extensive experiments have shown that the proposed approach leads very high correct recognition rates, especially in the case of partial face occlusion, where it is not possible to precisely detect the facial feature locations. It has been also found that, in the case of using a facial feature detector, the approach can tolerate localization errors of up to 18% of the interocular distance.

Background: Holistic processing is defined as the perceptual integration of facial features, and plays an important role in face recognition. While researchers recognize the crucial role played by holistic processing in face perception, a complete delineation of the underlying mechanisms is impending. Very few studies have examined the effects of perceptual discrimination and spatial perception on holistic processing. Hence, the present study aimed to examine the influence of perceptual discrimination and spatial perception on face recognition.Methods: We conducted two experiments by manipulating the perceptual discriminability of the target (the top-half faces) and non-target face (the bottom-half faces) parts in the composite-face task and examined how perceptual discriminability may affect holistic processing of faces.Results: The results of Experiment 1 illustrated that holistic processing was modulated by the perceptual discriminability of the face. Furthermore, differential patterns of perceptual discriminability with the target and non-target parts suggested that different mechanisms may be responsible for the influence of target and non-target parts on face perception. The results of Experiment 2 illustrated that holistic processing was modulated by spatial distance between two faces, implicating that feature-by-feature strategy might decrease the magnitude of holistic processing.Conclusion: The results of the present study suggest that holistic processing may lead to augmented perception effect exaggerating the differences between the two faces and may also be affected by the feature-by-feature strategy.

Biologically inspired visual object tracking has been very popular in the last decades. Several methods have exploited some visual cognitive mechanisms to outperform trackers purely based on machine vision in accuracy and speed. Although they have shown outstanding performance on recent benchmarks, they are still far from achieving the performance of a human in tracking. Visual attention is an important mechanism, which can be applied by saliency detection models. Saliency detection models can be determined in the spatial domain or the frequency domain. Spatial domain based models mostly have complex computations and are time-consuming. Thus, we have improved and recruited Frequency Divisive Normalization (FDN) model as a saliency detection method to propose a bio-inspired object tracking algorithm. This algorithm has been developed based on Fast Fourier Transform (FFT) and uses spectral features in order to achieve a high-speed tracker. It also can be launched on normal hardware. FDN model is based on Feature Integration Theory (FIT) and operates all the processes of saliency detection in the frequency domain, including decomposing the image to the features, lateral inhibition interaction, combining all features, and computing the saliency map. In addition to, extracting color, intensity and orientation it uses spatial frequency features, which is rarely used by other trackers. Extensive evaluations on large-scale benchmark datasets show that the proposed method has good performance with respect to its runtime.

Research is a very personal matter. On the basis of experiences in different countries with researchers from different cultures over many years, some observations will be described. The conceptual frame of this attempt is to look for anthropological universals and cultural specifics. Much can be learned from spatial representations in the arts. Whereas in the West since Renaissance time the central perspective has become dominant in visual art, in Eastern landscape paintings the "floating view" is typical. The claim that the central perspective corresponds to geometric laws and matches how we see the world is misleading for at least two reasons: It violates mechanisms of size constancy, and the visual world is spatially reduced in pictures to the perifoveal region only. Research on spatial attention has disclosed two different attentional systems being responsible either for near-fovea vision or for the far periphery. This fundamental principle as a global characteristic of visual processing is neglected in Western art. In Eastern art with a floating view geometric laws are violated, and different potential perspectives are integrated within a holistic pattern. The semantics of what shall be expressed becomes important irrespective of physical parameters. The latter may also create the unique phenomenon of becoming subjectively part of the picture confirming personal identity. Cultural specifics like in the arts (what one might expect) can surprisingly also be observed in theoretical considerations about visual processing. Whereas in the tradition of Western science visual percepts are built up with local elements like feature detectors, in an important Chinese theory global topological features are analyzed first. An important task of the brain is to create the identity of a percept on the basis of spatially and temporally distributed neural activities. It is, thus, an important theoretical question how to deal with the challenge to create and maintain the identity of a percept for some time. It is suggested that one should leave behind a monocausal reasoning for such explanations but adopt for analytical strategies the concept of complementarity as a generative principle.