Saliency Measure Research Articles

Evaluation of Local Feature Detectors for the Comparison of Thermal and Visual Low Altitude Aerial Images

Local features are key regions of an image suitable for applications such as image matching, and fusion. Detection of targets under varying atmospheric conditions, via aerial images is a typical defence application where multi spectral correlation is essential. Focuses on local features for the comparison of thermal and visual aerial images in this study. The state of the art differential and intensity comparison based features are evaluated over the dataset. An improved affine invariant feature is proposed with a new saliency measure. The performances of the existing and the proposed features are measured with a ground truth transformation estimated for each of the image pairs. Among the state of the art local features, Speeded Up Robust Feature exhibited the highest average repeatability of 57 per cent. The proposed detector produces features with average repeatability of 64 per cent. Future works include design of techniques for retrieval of corresponding regions.

Liberal Acceptance Bias, Momentary Aberrant Salience, and Psychosis: An Experimental Experience Sampling Study.

Cognitive models of psychosis posit that reasoning biases are an important mechanism contributing to the formation of psychotic symptoms, in part through transforming anomalous experiences of aberrant salience into frank psychotic symptoms. This study aimed to investigate the interplay of liberal acceptance (LA) bias, which is a specific type of reasoning bias, and momentary aberrant salience in the development of paranoid and psychotic experiences in daily life in first-episode psychosis patients (FEP), at-risk mental state participants (ARMS), and controls. We used a novel experimental Experience Sampling Methodology (eESM) task for measuring LA bias (ie, decisions based on low probability estimates) and ESM measures of momentary aberrant salience and paranoid and psychotic experiences in 51 FEP, 46 ARMS, and 53 controls. We found evidence that LA bias was more likely to occur in FEP than in controls. Further, LA bias was associated with psychotic and paranoid experiences (all P < .007) and modified the association between momentary aberrant salience and psychotic experiences (χ2(df) = 7.4(2), P = .025) in ARMS, such that momentary salience was associated with more intense psychotic experiences in the presence of LA bias in ARMS, but not in FEP and controls. Our findings suggest that LA bias may be central for anomalous experiences such as momentary aberrant salience to increase intensity of psychotic experiences in at-risk individuals. Further, LA bias appears to be more likely to be present, but not directly linked to current intensity of psychotic experiences, in treated FEP. Novel eESM tasks open new avenues for targeting psychological processes under real-world conditions.

Assessing Dynamic Value for Safety Gear During a Rock Climbing Task

Value is a component of the SEEV model of attention combined with measures of salience, effort, and expectancy to characterize areas of interest in a visual workspace. In the current project, an assessment of dynamic value was created by prompting rock climbers to assess their post-hoc priority for placing safety gear at various points during a climb when the risk of injury from a fall varied. Analyses determined that, for expert and non-expert rock climbers, gear placement priority rose as time since the most recent gear placement increased, after controlling for climbing speed. Prediction of the attentional value of a target via dynamic estimates may be applied in other highly dynamic environments.

RGB-D image saliency detection from 3D perspective

With the advent of stereo camera saliency object detection for RGB-D image is attracting more and more interest. Most existing algorithms treat RGB-D image as one RGB image and one depth map, then measure saliency map independently, and last fuse them. They disregard the fact that human visual system operates in real 3D environments. The paper proposed saliency object detection for RGB-D image from 3D perspective. It regards object as three dimensional structures, and redefines boundary conception in RGB-D image, and regards space boundary including top, down, left, right, front, back plane in real 3D environment as background. It incorporates 3D compactness feature, in which salient objects typically have 3D compact spatial distributions, into color and depth feature to express similarity among supervoxels and applies manifold ranking by six boundary planes to generate six saliency maps, and then integrates them to get the RGB-D saliency map from background view. In the end it refines saliency map by high confident salient seeds from foreground view. Experiment results show that six planes of RGB-D image are superior to four sides of RGB image as background seeds and 3D compactness plays an important role in saliency measurement. Our approach outperforms other state-of-the-art algorithms on NLPR RGBD 1000 benchmark.

A salience measure for 3D shape decomposition and sub-parts classification

This paper introduces a measure of significance on a curve skeleton of a 3D piecewise linear shape mesh, allowing the computation of both the shape’s parts and their saliency. We begin by reformulating three existing pruning measures into a non-linear PCA along the skeleton. From this PCA, we then derive a volume-based salience measure, the 3D WEDF, that determines the relative importance to the global shape of the shape part associated to a point of the skeleton. First, we provide robust algorithms for computing the 3D WEDF on a curve skeleton, independent on the number of skeleton branches. Then, we cluster the WEDF values to partition the curve skeleton, and coherently map the decomposition to the associated surface mesh. Thus, we develop an unsupervised hierarchical decomposition of the mesh faces into visually meaningful shape regions that are ordered according to their degree of perceptual salience. The shape analysis tools introduced in this paper are important for many applications including shape comparison, editing, and compression.

Semisupervised Learning Based on a Novel Iterative Optimization Model for Saliency Detection.

In this paper, we propose a novel iterative optimization model for bottom-up saliency detection. By exploring bottom-up saliency principles and semisupervised learning approaches, we design a high-performance saliency analysis method for wide ranging scenes. The proposed algorithm consists of two stages: 1) we develop a boundary homogeneity model to characterize the general position and the contour of the salient objects and 2) we propose a novel iterative optimization model, termed gradual saliency optimization, for further performance improvement. Our main contribution falls on the second stage, where we propose an iterative framework with self-repairing mechanisms for refining saliency maps. In this framework, we further develop a more comprehensive optimization function applying a novel semisupervised learning scheme to enhance the traditional saliency measure. More elaborately, the iterative method can gradually improve the output in each iteration and finally converge to high-quality saliency maps. Based on our experiments on four different public data sets, it can be demonstrated that our approach significantly outperforms the state-of-the-art methods.

The relationships among aberrant salience, reward motivation, and reward sensitivity.

Change in reward processing and motivation may mediate the relationship between dopaminergic dysregulation and positive symptoms of schizophrenia. We sought to investigate the measurement of aberrant salience and its relationship with behavioural measures of reward and motivation. Participants (n=82) completed measures of aberrant salience (Aberrant Salience Inventory and Salience Attribution Task), motivation (Effort Expenditure for Rewards Task), and reinforcer sensitivity (Stimulus Chase Task). Hypotheses were tested using correlation and generalised linear modelling. Results indicated no relationship between aberrant salience measures. The Aberrant Salience Inventory was positively related to effort expenditure for lower less likely rewards and predicted the use of probability alone in decision-making. The only significant relationship between reward and motivation was a positive relationship between gain sensitivity and motivated behaviour for higher more likely rewards. Although some support for a relationship between measures of reward motivation and aberrant salience were found, there was no evidence that the aberrant salience measures had concurrent validity. Our results suggest caution is warranted when interpreting measures of aberrant salience.

Salient Object Detection via Recursive Sparse Representation

Object-level saliency detection is an attractive research field which is useful for many content-based computer vision and remote-sensing tasks. This paper introduces an efficient unsupervised approach to salient object detection from the perspective of recursive sparse representation. The reconstruction error determined by foreground and background dictionaries other than common local and global contrasts is used as the saliency indication, by which the shortcomings of the object integrity can be effectively improved. The proposed method consists of the following four steps: (1) regional feature extraction; (2) background and foreground dictionaries extraction according to the initial saliency map and image boundary constraints; (3) sparse representation and saliency measurement; and (4) recursive processing with a current saliency map updating the initial saliency map in step 2 and repeating step 3. This paper also presents the experimental results of the proposed method compared with seven state-of-the-art saliency detection methods using three benchmark datasets, as well as some satellite and unmanned aerial vehicle remote-sensing images, which confirmed that the proposed method was more effective than current methods and could achieve more favorable performance in the detection of multiple objects as well as maintaining the integrity of the object area.

26.3 SALIENCE SIGNALING AND THE EMERGENCE OF PSYCHOPATHOLOGY IN YOUTH AT CLINICAL HIGH RISK FOR PSYCHOTIC ILLNESS

BackgroundThe early identification of people who appear to be at high risk for conversion to psychosis has become a central thrust of mental health research, with the hope that early intervention may alter the course of psychotic illness. Importantly, both positive and negative symptom dimensions have been found relate to risk for conversion in clinical high risk (CHR) populations. Neuroimaging work points to a role for dopamine pathway activity in both the positive and negative symptoms of psychotic illness. A role for dopamine pathways in signaling various kinds of salience is well-established, and several authors have proposed that excessive dopamine transmission in the striatum might contribute to psychotic symptoms by bringing about erratic, or “aberrant”, salience signaling. By contrast, a reduced ability to identify salient events as such, and signal salience “adaptively”, could result in impairments in learning and motivation. I will describe results from a study in which we examined the impact of salient events on learning and behavior in adolescents and young adults, a subset of whom were identified as being at CHR for developing psychotic illness.MethodsParticipants were 98 adolescents and young adults (mean age = 16.1 ± 3.3 years), assessed clinically using the Structured Interview for Psychosis-Risk Syndromes (SIPS). Eighty-nine participants were receiving mental health services, with 30 of the 89 identified as CHR and 8 identified as already having a psychotic illness. We used two experimental paradigms to investigate the impact of salient events on learning and behavior: the probabilistic stimulus selection task (PSST; Frank et al., 2004) and the Salience Attribution Task (SAT; Roiser et al., 2009). Both adaptive and aberrant salience signals were operationalized in the context of each task. Successful performance of the PSST depends on the adaptive signaling of mismatches between expected and obtained outcomes, called reward prediction errors, which are one form of salient event. The SAT requires participants to respond as quickly as possible to a response prompt, which is preceded by conditioned stimuli that potentially predict reward availability for a fast response. The comparison of reaction time (RT) between responses following the frequently vs. infrequently rewarded conditioned stimuli offers a measure of adaptive salience coding with the expectation of faster RT for reward predicting stimuli. The comparison of RT between responses to the two levels of the irrelevant dimension offers a measure of aberrant salience coding with the expectation of equal RT for stimuli equally-predictive of reward. We assessed whether experimental measures of both adaptive and aberrant salience showed correspondences with SIPS ratings for symptoms along both the positive and negative dimensions.ResultsWe observed significant correlations between multiple performance measures from the PSST and measures of both positive and negative symptoms. We found that positive symptom severity, in help-seeking youth, correlated positively with an implicit measure of aberrant salience from the SAT, and negatively with an explicit measure of adaptive salience.DiscussionThese results, consistent with our previous findings in both first-episode psychosis patients and patients with chronic schizophrenia, suggest that experimental measures of salience signaling may provide a psychosis risk signal in treatment-seeking youth. Further research is necessary to understand the potential predictive role of these measures for conversion to psychosis.

T85. PRELIMINARY ANALYSES OF THE NEUROCOGNITIVE DATABASE OF PRONIA USING UNIVARIATE STATISTICS: CLINICAL GROUP DIFFERENCES

BackgroundNeuro-cognitive deficits are a core feature of psychosis. In the clinical high risk stages of psychosis, neuro-cognitive deficits qualitatively affect the same functions while being quantitatively less marked compared to those in full-blown disorder. Therefore, cognitive impairments are considered to be an important intermediate phenotype for transition to psychosis. Partially overlapping deficits were also reported in depressive disorders, so it is important to identify deficits specifically associated to psychotic symptoms from those common to other conditions. We aimed to identify and differentiate cognitive deficits specifically associated to [i] psychopathology in general (i.e., presence of clinical diagnosis); [ii] psychotic symptoms; [iii] sub- and threshold levels of psychotic symptoms.MethodsWe compared four groups of participants within the project Personalised Prognostic Tools for Early Psychosis Management (PRONIA; www.pronia.eu). The PRONIA Cognitive Battery (PCB) includes 10 tests selected as reliable measures of neuropsychological difficulties in patients at high-risk of psychosis. The scores were obtained from the PRONIA Discovery Sample, which included 707 participants: 278 healthy controls (HC); 138 recent-onset depression (ROD); 139 clinical high-risk (CHR); 152 recent-onset psychosis (ROP), tested in seven sites across Europe. At first the norms were calculated correcting the HC’s raw scores by sex, age, cognitive level, education, and mother language (English, Finnish, German, Italian, or other). Then, univariate analyses of variance with a priori contrasts were used for directly comparing [i] HC vs ROD/CHR/ROP; [ii] ROD vs CHR/ROP; [iii] CHR vs ROP.ResultsThe difference in cognitive performance between the clinical groups (ROD, CHR, ROP) as compared to the HC [i], was shown in measures of: speed of execution (ωP2 range 0.016–0.123; all p≤0.035); sustained attention (ωP2: 0.024–0.080; p≤0.022); verbal fluency (ωP2: 0.020–0.031; p≤0.002); emotion recognition (ωP2=0.026; p=0.001); visuo-spatial (ωP2: 0.018–0.049; p≤0.006) and verbal (ωP2: 0.038–0.075; p<0.001) both short- and long-term memory. Three clinical groups did not show significant difference in salience measures when compared with HC (p≥0.053), beyond a main effect of group (ωP2=0.015). Differences between ROD and CHR/ROP groups [ii] were detected in: speed of execution (all p≤0.001); sustained attention (p≤0.011); short-term and working memory (p≤0.004); long-term memory (p≤0.001); semantic verbal fluency (p=0.024); emotion recognition (p=0.005); and estimation of adaptive salience (p=0.021). When compared with ROP, CHR [iii] performed significantly better in the same domains that differentiated ROD from CHR/ROP, with the important exception of long-term memory measures (p≥0.094).DiscussionThese results are consistent with the expectations drawn from previous literature on the neuropsychological impairments in psychotic disorders and CHR participants. Furthermore, PCB showed to be useful in [i] psychopathology in general, [ii] differentiating between recent-onset depression and psychotic symptoms, and [iii] between threshold and sub-threshold psychotic symptoms. Interestingly, long-term memory deficits contributed more in differentiating psychotic symptoms from other psychopathological entities (ROD vs CHR/ROP comparison) than along the spectrum of attenuated psychotic symptoms, resulting in full clinical picture of psychosis (CHR vs ROP). Finally, salience attribution difficulties were confirmed to be associated with (sub-)threshold psychotic symptoms, more than to general psychopathology.

S70. ABERRANT SALIENCE: A COMPARISON OF DIFFERENT MEASURES IN ANXIETY AND SCHIZOPHRENIA

BackgroundAberrant salience is thought to play a role in the development of the symptoms of schizophrenia, but the hypothesis lacks consistent support. Previous research found no relationship in a population sample between two measures of aberrant salience: the self-report Aberrant Salience Inventory (ASI) and the computerised Salience Attribution Task (SAT), which measures implicit (behavioural) and explicit (self-report) aberrant and adaptive salience. We compared the ASI and SAT in individuals with schizophrenia, with anxiety, and with no mental disorder (unaffected).MethodsIndividuals with schizophrenia (n = 30), anxiety (n = 33), or unaffected (n = 30) completed the ASI and the SAT.ResultsASI scores were higher in the schizophrenia group than anxiety (t(90) = 2.72, p < .01) and unaffected groups (t (90) = 5.29, p < .001) and higher in the anxiety than unaffected group (t(90) = 2.69, p < .01). SAT explicit adaptive salience scores were lower in the schizophrenia group than the anxiety (t(90) = -3.79, p < .001) and unaffected groups (t(90)= -3.86, p < .001). The schizophrenia group also had higher SAT implicit aberrant salience than the anxiety group (t(90) = 2.57, p < .05) but not the unaffected group (t(90) = 3.75, p = .08); there was no difference between anxiety and control groups (t(90) -0.76, p = .45). Group did not affect SAT explicit aberrant salience (F(2,91) = 0.47, p =.63) or implicit adaptive salience (F(90) = 0.62, p = .54). We found no correlation between the ASI and the SAT (all τ < .218, p > .05).DiscussionHigher ASI scores were associated with, but not unique to, schizophrenia. Reduced SAT explicit adaptive salience was associated with schizophrenia, while SAT implicit aberrant salience scores differed between psychopathologies. Consistent with previous findings, there was no relationship between the ASI and the SAT. The ASI is designed to measure a trait associated with schizophrenia. Conversely, the SAT implicit aberrant salience measures response latency to irrelevant stimuli. The lack of relationship between ASI and SAT may, therefore, be due to construct divergence.

Correction: A parallel spatiotemporal saliency and discriminative online learning method for visual target tracking in aerial videos.

[This corrects the article DOI: 10.1371/journal.pone.0192246.].

The Myth of Mob Rule: Violent Crime and Democratic Politics. Lisa L. Miller. New York: Oxford University Press (2016). 272pp. £26.49hb ISBN 9780190228705

The Myth of Mob Rule: Violent Crime and Democratic Politics. Lisa L. Miller. New York: Oxford University Press (2016). 272pp. £26.49hb ISBN 9780190228705

A parallel spatiotemporal saliency and discriminative online learning method for visual target tracking in aerial videos.

Visual tracking in aerial videos is a challenging task in computer vision and remote sensing technologies due to appearance variation difficulties. Appearance variations are caused by camera and target motion, low resolution noisy images, scale changes, and pose variations. Various approaches have been proposed to deal with appearance variation difficulties in aerial videos, and amongst these methods, the spatiotemporal saliency detection approach reported promising results in the context of moving target detection. However, it is not accurate for moving target detection when visual tracking is performed under appearance variations. In this study, a visual tracking method is proposed based on spatiotemporal saliency and discriminative online learning methods to deal with appearance variations difficulties. Temporal saliency is used to represent moving target regions, and it was extracted based on the frame difference with Sauvola local adaptive thresholding algorithms. The spatial saliency is used to represent the target appearance details in candidate moving regions. SLIC superpixel segmentation, color, and moment features can be used to compute feature uniqueness and spatial compactness of saliency measurements to detect spatial saliency. It is a time consuming process, which prompted the development of a parallel algorithm to optimize and distribute the saliency detection processes that are loaded into the multi-processors. Spatiotemporal saliency is then obtained by combining the temporal and spatial saliencies to represent moving targets. Finally, a discriminative online learning algorithm was applied to generate a sample model based on spatiotemporal saliency. This sample model is then incrementally updated to detect the target in appearance variation conditions. Experiments conducted on the VIVID dataset demonstrated that the proposed visual tracking method is effective and is computationally efficient compared to state-of-the-art methods.

A novel search algorithm for quantifying news media coverage as a measure of environmental issue salience

News media plays an important role in shaping public opinions and attitudes about the environment. Tracking and analyzing media coverage can provide insight into public exposure to narratives that impact resource consumption, environmental behavior, and emergency response, which can help to inform model development or provide additional model inputs. This paper presents Articulate, an open-source, flexible tool for discovering, compiling, and quantifying newspaper coverage on a user-specified topic. Articulate is written in Python and interfaces with Google Custom Search Engine API. We demonstrate the tool's application and validate its performance on two case studies of news media coverage in the New York Times about drought in California and flooding in Houston, Texas in recent years. Our results show that Articulate can generate data similar to or better than proprietary databases. Thus, Articulate can help researchers and environmental managers gain important insights to better understand and quantify changing socio-environmental dynamics.

The role of ethnic identity, self-concept, and aberrant salience in psychotic-like experiences.

Social-cognitive models of psychosis suggest that aberrant salience and self-concept clarity are related to the development and maintenance of psychoticlike experiences (PLEs). People with high aberrant salience but low self-concept clarity tend to have the highest levels of PLEs. Ethnic identity may also be related to PLEs. The current research aimed to (a) replicate the interaction between aberrant salience and self-concept clarity in their association with PLEs in an ethnically diverse sample, (b) examine whether ethnic identity and aberrant salience interact in their association with PLEs, and (c) determine if self-concept clarity and ethnic identity independently interact with aberrant salience in their association with PLEs. An ethnically diverse group of undergraduates (n = 663) completed self-report measures of aberrant salience, self-concept clarity, ethnic identity, and PLEs. There was an interaction between aberrant salience and self-concept clarity such that people with high levels of aberrant salience and low levels of self-concept clarity had the highest levels of PLEs. Similarly, there was an interaction between aberrant salience and ethnic identity such that people with high aberrant salience but low ethnic identity had the highest PLEs. These interactions independently contributed to explaining variance in PLEs. This interaction was present for the Exploration but not Commitment subscales of ethnic identity. These results suggest that, in addition to low self-concept clarity, low ethnic identity may be a risk factor for the development of psychosis. (PsycINFO Database Record

Visual Saliency Modeling for River Detection in High-Resolution SAR Imagery

Accurate detection of rivers plays a significant role in water conservancy construction and ecological protection, where airborne synthetic aperture radar (SAR) data have already become one of the main sources. However, extracting river information from radar data efficiently and accurately still remains an open problem. The existing methods for detecting rivers are typically based on rivers’ edges, which are easily mixed with those of artificial buildings or farmland. In addition, pixel-based image processing approaches cannot meet the requirement of real-time processing. Inspired by the feature integration and target recognition capabilities of biological vision systems, in this paper, we present a hierarchical method for automated detection of river networks in the high-resolution SAR data using biologically visual saliency modeling. For effective saliency detection, the original image is first over-segmented into a set of primitive superpixels. A visual feature set is designed to extract a regional feature histogram, which is then quantized based on the optimal parameters learned from the labeled SAR images. Afterward, three saliency measurements based on the specificity of the rivers in the SAR images are proposed to generate a single layer saliency map, i.e., local region contrast, boundary connectivity, and edge density. Finally, by exploiting belief propagation, we propose a multi-layer saliency fusion approach to derive a high-quality saliency map. Extensive experimental results on three airborne SAR image data sets with the ground truth demonstrate that the proposed saliency model consistently outperforms the existing saliency target detection models.

Saliency Map Generation for SAR Images with Bayes Theory and Heterogeneous Clutter Model

Saliency map generation in synthetic aperture radar (SAR) imagery has become a promising research area, since it has a close relationship with quick potential target identification, rescue services, etc. Due to the multiplicative speckle noise and complex backscattering in SAR imagery, producing satisfying results is still challenging. This paper proposes a new saliency map generation approach for SAR imagery using Bayes theory and a heterogeneous clutter model, i.e., the G 0 model. With Bayes theory, the ratio of the probability density functions (PDFs) in the target and background areas contributes to the saliency. Local and global background areas lead to different saliency measures, i.e., local saliency and global saliency, which are combined to make a final saliency measure. To measure the saliency of targets of different sizes, multiscale saliency enhancement is conducted with different region sizes of target and background areas. After collecting all of the salient regions in the image, the result is refined by considering the image’s immediate context. The saliency of regions that are far away from the focus of attention is suppressed. Experimental results with two single-polarization and two multi-polarization SAR images demonstrate that the proposed method has better speckle noise robustness, higher accuracy, and more stability in saliency map generation both with and without the complex background than state-of-the-art methods. The saliency map accuracy can achieve above 95% with four datasets, which is about 5–20% higher than other methods.

Saliency Detection for Unconstrained Videos Using Superpixel-Level Graph and Spatiotemporal Propagation

This paper proposes an effective spatiotemporal saliency model for unconstrained videos with complicated motion and complex scenes. First, superpixel-level motion and color histograms as well as global motion histogram are extracted as the features for saliency measurement. Then a superpixel-level graph with the addition of a virtual background node representing the global motion is constructed, and an iterative motion saliency (MS) measurement method that utilizes the shortest path algorithm on the graph is exploited to reasonably generate MS maps. Temporal propagation of saliency in both forward and backward directions is performed using efficient operations on inter-frame similarity matrices to obtain the integrated temporal saliency maps with the better coherence. Finally, spatial propagation of saliency both locally and globally is performed via the use of intra-frame similarity matrices to obtain the spatiotemporal saliency maps with the even better quality. The experimental results on two video data sets with various unconstrained videos demonstrate that the proposed model consistently outperforms the state-of-the-art spatiotemporal saliency models on saliency detection performance.

Learning Optimal Seeds for Ranking Saliency

A variety of methods have been developed for visual saliency analysis, and it is a challenge to detect the most important scene from the input image. In this paper, to improve the shortage that the spatial connectivity of every node in model only via the k-regular graph and the idealistic boundary prior assumption is used in graph-based manifold ranking, we present a new optimal seed method to get saliency map. First, we evaluate the salience value of each region by global contrast-based spatial and color feature. Second, the salience values of the first stage are used to optimize the background and foreground queries (seeds); meanwhile, we tackle boundary cues from hierarchical graph to optimize background seeds. Then, we derive each stage saliency measure by the classical manifold ranking after obtaining optimal seeds. Finally, the final saliency map is obtained by combining the saliency results of two stages. Our algorithm is tested on the five public datasets and compared with nine state-of-the-art methods; the quantitative evaluation indicates that our method is effective and efficient. Our method can handle complex images with different details and can produce more accurate saliency maps than other state-of-the-art approaches.