Salient Contours Research Articles

CEDFlow: Latent Contour Enhancement for Dark Optical Flow Estimation

Accurately computing optical flow in low-contrast and noisy dark images is challenging, especially when contour information is degraded or difficult to extract. This paper proposes CEDFlow, a latent space contour enhancement for estimating optical flow in dark environments. By leveraging spatial frequency feature decomposition, CEDFlow effectively encodes local and global motion features. Importantly, we introduce the 2nd-order Gaussian difference operation to select salient contour features in the latent space precisely. It is specifically designed for large-scale contour components essential in dark optical flow estimation. Experimental results on the FCDN and VBOF datasets demonstrate that CEDFlow outperforms state-of-the-art methods in terms of the EPE index and produces more accurate and robust flow estimation. Our code is available at: https://github.com/xautstuzfy.

Few-parameter learning for a hierarchical perceptual grouping system.

Perceptual grouping along well-established Gestalt laws provides one set of traditional methods that provide a tiny set of meaningful parameters to be adjusted for each application field. More complex and challenging tasks require a hierarchical setting, where the results aggregated by a first grouping process are later subject to further processing on a larger scale and with more abstract objects. This can be several steps deep. An example from the domain of forestry provides insight into the search for suitable parameter settings providing sufficient performance for the machine-vision module to be of practical use within a larger robotic control setting in this application domain. This sets a stark contrast in comparison to the state-of-the-art deep-learning neural nets, where many millions of obscure parameters must be adjusted properly before the performance suffices. It is the opinion of the author that the huge freedom for possible settings in such a high-dimensional inscrutable parameter space poses an unnecessary risk. Moreover, few-parameter learning is getting along with less training material. Whereas the state-of-the-art networks require millions of images with expert labels, a single image can already provide good insight into the nature of the parameter domain of the Gestalt laws, and a domain expert labeling just a handful of salient contours in said image yields already a proper goal function, so that a well working sweet spot in the parameter domain can be found in a few steps. As compared to the state-of-the-art neural nets, a reduction of six orders of magnitude in the number of parameters results. Almost parameter-free statistical test methods can reduce the number of parameters to be trained further by one order of magnitude, but they are less flexible and currently lack the advantages of hierarchical feature processing.

Deep image retrieval using artificial neural network interpolation and indexing based on similarity measurement

In content-based image retrieval (CBIR), primitive image signatures are critical because they represent the visual characteristics. Image signatures, which are algorithmically descriptive and accurately recognized visual components, are used to appropriately index and retrieve comparable results. To differentiate an image in the category of qualifying contender, feature vectors must have image information's like colour, objects, shape, spatial viewpoints. Previous methods such as sketch-based image retrieval by salient contour (SBIR) and greedy learning of deep Boltzmann machine (GDBM) used spatial information to distinguish between image categories. This requires interest points and also feature analysis emerged image detection problems. Thus, a proposed model to overcome this issue and predict the repeating pattern as well as series of pixels that conclude similarity has been necessary. In this study, a technique called CBIR-similarity measure via artificial neural network interpolation (CBIR-SMANN) has been presented. By collecting datasets, the images are resized then subject to Gaussian filtering in the pre-processing stage, then by permitting them to the Hessian detector, the interesting points are gathered. Based on Skewness, mean, kurtosis and standard deviation features were extracted then given to ANN for interpolation. Interpolated results are stored in a database for retrieval. In the testing stage, the query image was inputted that is subjected to pre-processing, and feature extraction was then fed to the similarity measurement function. Thus, ANN helps to get similar images from the database. CBIR-SMANN have been implemented in the python tool and then evaluated for its performance. Results show that CBIR-SMANN exhibited a high recall value of 78% with a minimum retrieval time of 980 ms. This showed the supremacy of the proposed model was comparatively greater than the previous ones.

Joint Learning of Salient Object Detection, Depth Estimation and Contour Extraction

Benefiting from color independence, illumination invariance and location discrimination attributed by the depth map, it can provide important supplemental information for extracting salient objects in complex environments. However, high-quality depth sensors are expensive and can not be widely applied. While general depth sensors produce the noisy and sparse depth information, which brings the depth-based networks with irreversible interference. In this paper, we propose a novel multi-task and multi-modal filtered transformer (MMFT) network for RGB-D salient object detection (SOD). Specifically, we unify three complementary tasks: depth estimation, salient object detection and contour estimation. The multi-task mechanism promotes the model to learn the task-aware features from the auxiliary tasks. In this way, the depth information can be completed and purified. Moreover, we introduce a multi-modal filtered transformer (MFT) module, which equips with three modality-specific filters to generate the transformer-enhanced feature for each modality. The proposed model works in a depth-free style during the testing phase. Experiments show that it not only significantly surpasses the depth-based RGB-D SOD methods on multiple datasets, but also precisely predicts a high-quality depth map and salient contour at the same time. And, the resulted depth map can help existing RGB-D SOD methods obtain significant performance gain.

Contour detection and salient feature line regularization for printed circuit board in point clouds based on geometric primitives

Effective extraction of the printed circuit board (PCB)’s key contour feature line is of great significance in the optoelectronic industry, especially for defect detection and accurate dimensional measurement. Here, we develop a method to convert the fold edge points into boundary points and then extract the salient contour feature lines of the PCB. The method is performed directly on the real PCB point cloud, and no image processing or 3D reconstruction is required. Our approach involves three key steps: point cloud preprocessing, feature point detection, and contour feature line regularization. Firstly, the PCB point cloud’s preprocessing is completed using the k-d tree and pass-through filtering algorithms. Secondly, through the least-median-of-squares (LMS) and the Euclidean-Clustering-Based-on-Gaussian-Mapping (ECGM), the fold edge points are converted into boundary points. Then, the contour feature points are identified by angle criterion between lines formed by the query point with respect to its neighboring points. Finally, the feature points are regularized by the self-pruning B-spline curve based on squared distance minimization (SDM). Compared with the other methods, our approach extracts more salient contour points and fewer redundant points. Experiments on real test data also suggest that this method has good potential in extracting feature lines, which ensures accuracy and strong applicability.

Singing Transcription from Polyphonic Music Using Melody Contour Filtering

Automatic singing transcription and analysis from polyphonic music records are essential in a number of indexing techniques for computational auditory scenes. To obtain a note-level sequence in this work, we divide the singing transcription task into two subtasks: melody extraction and note transcription. We construct a salience function in terms of harmonic and rhythmic similarity and a measurement of spectral balance. Central to our proposed method is the measurement of melody contours, which are calculated using edge searching based on their continuity properties. We calculate the mean contour salience by separating melody analysis from the adjacent breakpoint connective strength matrix, and we select the final melody contour to determine MIDI notes. This unique method, combining audio signals with image edge analysis, provides a more interpretable analysis platform for continuous singing signals. Experimental analysis using Music Information Retrieval Evaluation Exchange (MIREX) datasets shows that our technique achieves promising results both for audio melody extraction and polyphonic singing transcription.

Phasic pupillary responses modulate object-based attentional prioritization

Visual attention studies have demonstrated that the shape of space-based selection can be governed by salient object contours: when a portion of an enclosed space is cued, the selected region extends to the full enclosure. Although this form of object-based attention (OBA) is well established, one continuing investigation focuses on whether this selection is obligatory or under voluntary control. We attempt to dissociate between these alternatives by interrogating the locus coeruleus-norepinephrine (LC-NE) system – known to fluctuate with top-down attention – during a classic two-rectangle paradigm in a sample of healthy human participants (N = 36). An endogenous spatial pre-cue directed voluntary space-based attention (SBA) to one end of a rectangular frame. We manipulated the reliability of the cue, such that targets appearing at an uncued location within the frame occurred at low or moderate frequencies. Phasic pupillary responses time-locked to the cue display served to noninvasively measure LC-NE activity, reflecting top-down processing of the spatial cue. If OBA is controlled analogously to SBA, then object selection should emerge only when it is behaviorally expedient and when LC-NE activity reflects a high degree of top-down attention to the cue display. Our results bore this out. Thus, we conclude that OBA was voluntarily controlled, and furthermore show that phasic norepinephrine may modulate attentional strategy.

Improving the Contour Saliency by Tensor Superposition

Contour integration is an important intermediate stage of the object recognition, in which line segments belonging to an object boundary are perceptually linked and segmented from complex backgrounds. This paper puts forward an improved contour integration algorithm. The method computes the vector sum of all neighboring ball voting vectors for an object image pixel to improve the saliency of edges. Four groups of experiments are performed on typical contours to compare the contour in painting effect and the precision. Experimental results show that the contour saliency can be improved by tensor superposition.

A contour self-compensated network for salient object detection

Given that existing salient object detection methods cannot effectively predict the fine contours of salient objects when extracting local or global contexts and features, we propose a novel contour self-compensated network (CSCNet) to generate a more accurate saliency map with complete contour. Unlike the common binary saliency detection, we reconstruct the salient object detection problem into a multi-classification problem of the background, the salient object, and the salient contour, where the salient contour is used as the third label for ground truth. Meanwhile, the image and its superpixel map are concatenated as the input of our network to add more edge information. Also, a penalty loss is proposed to restrict the spatial relationship between the background, objects, and their contours. Experimentally, we evaluate the proposed CSCNet on six benchmark datasets in both accuracy and efficiency and evaluate the attribute-based performance on the SOC dataset. Compared with 13 state-of-the-art algorithms, our CSCNet can detect salient objects more accurately and completely without adding too many convolutional layers and parameters.

Salient contour detection on the basis of the mechanism of bilateral asymmetric receptive fields

Salient contour detection is the key step of visual perception and is very important for visual detection and object recognition. In this paper, a new contour detection method, which is based on the bilateral asymmetric receptive field mechanism of the visual pathway, is proposed. First, the classical receptive field of primary visual cortex neurons was simulated, and the 2D Gaussian derivative was used to detect the primary contour response of the input image. Then, the asymmetric receptive field structure was introduced to enhance the contrast difference in local regions. Assuming that unilateral asymmetric receptive fields will suppress the intensity imbalance of the primary contour in the image, the strategy of weight information fusion, which is based on the bilateral asymmetric receptive field multi-scale inhibition, was proposed. Finally, texture suppression was performed with varying intensity in the local regions of the primary contour in the image, and the salient contour was detected. The salient contour detection method, which is based on bilateral asymmetric receptive fields proposed in this study, provides new ideas for the subsequent image understanding and analysis that are on the basis of the visual mechanism.

Infrared Moving Small-Target Detection via Spatiotemporal Consistency of Trajectory Points

Effective detection of infrared (IR) moving small targets in complex cluttered environments plays a key role in IR search and track systems for self-defense or attacks. In this letter, an IR moving small-target detection algorithm utilizing a spatiotemporal consistency of motion trajectories is proposed. First, feature points are densely sampled and tracked using the dense optical flow algorithm to compute dense trajectories. Second, suspected trajectories are deleted by utilizing the moving characteristics of the target. Third, under the assumption that each small target is defined as a compact space region, a binary image is created depending on the image coordinates of the trajectory points, from which salient contours are extracted as candidate target regions. Finally, a coding mechanism for contour numbering is introduced, and the moving targets are distinguished from the backgrounds by the temporal consistency of contour codewords. Several experiments were conducted, and their results demonstrate that our proposed method can detect small moving IR targets with higher detection rate, lower false alarm rate, and less running time compared with the state-of-the-art methods.

Application of the center–surround mechanism to contour detection

Physiological studies have revealed that the center–surround mechanism widely exists in the primary stages of the human visual system, such as the retina, lateral geniculate nucleus (LGN), and primary visual cortex (V1). In retina ganglion cells (RGC) and the LGN, the mechanism is well known to have two types: center “on” and center “off.” However, this mechanism in V1 is shown as classical receptive field (CRF) stimulation and surrounding non-CRF suppression. Although these two manifestations differ in function and appear in different areas of the visual pathway, from the perspective of computational simulation, they simply compute the differences between the center and its surrounding information. In the past decade, many bio-inspired computational models have demonstrated that the center–surround mechanism is good at extracting salient contours while suppressing textures. Based on this mechanism, we propose a method for extracting local center–surround contrast information from nature images by using a normalized difference of Gaussian (DoG) function and a sigmoid activated function. Compared with previous contour detection models (especially bio-motivated ones), the proposed method can efficiently suppress textures more quickly and accurately. More importantly, the proposed algorithm yields even better contour detection, yet the computational complexity is similar to the classical Canny operator.

Robust Cooperative Target Detection for a Vision-Based UAVs Autonomous Aerial Refueling Platform via the Contrast Sensitivity Mechanism of Eagle's Eye

In this paper, a robust cooperative target detection method inspired by contrast sensitivity mechanism of eagle's eye is proposed to extract the cooperative target, aiming at reducing the computation load and improving the accuracy of feature extraction. The contrast sensitivity mechanism is simulated based on the attenuation effect of CSF of an eagle to suppress the texture edges. The comparison experimental results compared with the Canny edge detection operator demonstrated the excellent performance of our proposed method in extracting salient contours and suppressing texture edges. Besides, a binocular vision based AAR platform for UAV is designed and implemented in this paper. Applying our proposed detection method to a series of outdoor experiment images captured from two challenging lighting conditions, the cooperative targets are successfully extracted while the other disturbance areas are removed dramatically. The outdoor flight tests are conducted to verify the visual measurement system and the hardware platform. The experimental results verified the feasibility and effectiveness of our developed visual measurement algorithms. In the future, we will focus on the influence of other factors on the marker detection to further improve the accuracy of marker detection, like illumination, the image quality influenced by the motion of tanker UAV.

Composite description based on Salient Contours and Color information for CBIR tasks.

This paper introduces a novel image descriptor for content based image retrieval tasks that integrates contour and color information into a compact vector. Loosely inspired by the human visual system and its mechanisms in efficiently identifying visual saliency, operations are performed on a fixed lattice of discrete positions by a set of edge detecting kernels that calculate region derivatives at different scales and orientation. The description method utilizes a weighted edge histogram where bins are populated on the premise of whether the regions contain edges belonging to the salient contours, while the discriminative power is further enhanced by integrating regional quantized color information. The proposed technique is both efficient and adaptive to the specifics of each depiction, while it does not need any training data to adjust parameters. Experimental evaluation conducted on seven benchmarking datasets against 13 well known global descriptors along with SIFT, SURF implementations (both in VLAD and BOVW), highlight the effectiveness and efficiency of the proposed descriptor.

Intact illusory contour formation but equivalently impaired visual shape completion in first- and later-episode schizophrenia.

Visual shape completion is a fundamental process that constructs contours and shapes on the basis of the geometric relations between spatially separated edge elements. People with schizophrenia are impaired at distinguishing visually completed shapes, but when does the impairment emerge and how does it evolve with illness duration? The question bears on the debate as to whether cognition declines after illness onset. To address the issue, we tested healthy controls (n = 48), first-episode psychosis patients (n = 23), and chronic schizophrenia patients (n = 49) on a classic psychophysical task in which subjects discriminated the relative orientations of four sectored circles that either formed or did not form visually completed shapes (illusory and fragmented conditions, respectively). Visual shape completion was quantified as the extent to which performance in the illusory condition exceeded that of the fragmented. Half of the trials incorporated wire edge elements, which augment contour salience and improve shape completion. Each patient group exhibited large visual shape completion deficits that could not be explained by differences in age, motivation, or orientation tuning. Patients responded normally to changes in illusory contour salience, indicating that they were forming but not adequately employing such contours for discriminating shapes. Shape completion deficits were most apparent for patients with cognitive disorganization, poor premorbid early adolescent functioning, and normal orientation discrimination. Visual shape completion deficits emerge maximally by the first psychotic episode and arise from higher-level disturbances that are related to premorbid functioning and disorganization. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Biologically Inspired Hierarchical Contour Detection with Surround Modulation and Neural Connection.

Contour is a very important feature in biological visual cognition and has been extensively investigated as a fundamental vision problem. In connection with the limitations of conventional models in detecting image contours in complex scenes, a hierarchical image contour extraction method is proposed based on the biological vision mechanism that draws on the perceptual characteristics of the early vision for features such as edges, shapes, and colours. By simulating the information processing mechanisms of the cells’ receptive fields in the early stages of the biological visual system, we put forward a computational model that combines feedforward, lateral, and feedback neural connections to decode and obtain the image contours. Our model simulations and their results show that the established hierarchical contour detection model can adequately fit the characteristics of the biological experiment, quickly and effectively detect the salient contours in complex scenes, and better suppress the unwanted textures.

A Fast Contour Detection Model Inspired by Biological Mechanisms in Primary Vision System.

Compared to computer vision systems, the human visual system is more fast and accurate. It is well accepted that V1 neurons can well encode contour information. There are plenty of computational models about contour detection based on the mechanism of the V1 neurons. Multiple-cue inhibition operator is one well-known model, which is based on the mechanism of V1 neurons' non-classical receptive fields. However, this model is time-consuming and noisy. To solve these two problems, we propose an improved model which integrates some additional other mechanisms of the primary vision system. Firstly, based on the knowledge that the salient contours only occupy a small portion of the whole image, the prior filtering is introduced to decrease the running time. Secondly, based on the physiological finding that nearby neurons often have highly correlated responses and thus include redundant information, we adopt the uniform samplings to speed up the algorithm. Thirdly, sparse coding is introduced to suppress the unwanted noises. Finally, to validate the performance, we test it on Berkeley Segmentation Data Set. The results show that the improved model can decrease running time as well as keep the accuracy of the contour detection.

Contrast-Oriented Deep Neural Networks for Salient Object Detection.

Deep convolutional neural networks (CNNs) have become a key element in the recent breakthrough of salient object detection. However, existing CNN-based methods are based on either patchwise (regionwise) training and inference or fully convolutional networks. Methods in the former category are generally time-consuming due to severe storage and computational redundancies among overlapping patches. To overcome this deficiency, methods in the second category attempt to directly map a raw input image to a predicted dense saliency map in a single network forward pass. Though being very efficient, it is arduous for these methods to detect salient objects of different scales or salient regions with weak semantic information. In this paper, we develop hybrid contrast-oriented deep neural networks to overcome the aforementioned limitations. Each of our deep networks is composed of two complementary components, including a fully convolutional stream for dense prediction and a segment-level spatial pooling stream for sparse saliency inference. We further propose an attentional module that learns weight maps for fusing the two saliency predictions from these two streams. A tailored alternate scheme is designed to train these deep networks by fine-tuning pretrained baseline models. Finally, a customized fully connected conditional random field model incorporating a salient contour feature embedding can be optionally applied as a postprocessing step to improve spatial coherence and contour positioning in the fused result from these two streams. Extensive experiments on six benchmark data sets demonstrate that our proposed model can significantly outperform the state of the art in terms of all popular evaluation metrics.

A man-made object detection algorithm based on contour complexity evaluation

Man-made object detection is of great significance in both military and civil areas, such as search-and-rescue missions at sea, traffic signs recognition during visual navigation, and targets location in a military strike. Contours of man-made objects usually consist of straight lines, corner points, and simple curves. Motivated by this observation, a man-made object detection method is proposed based on complexity evaluation of object contours. After salient contours which keep the crucial information of objects are accurately extracted using an improved mean-shift clustering algorithm, a novel approach is presented to evaluate the complexity of contours. By comparing the entropy values of contours before/after sampling and linear interpolation, it is easy to distinguish between man-made objects and natural ones according to the complexity of their contours. Experimental results show that the presented method can effectively detect man-made objects when compared to the existing ones.

New Hierarchical Saliency Filtering for Fast Ship Detection in High-Resolution SAR Images

The increasing resolution of synthetic aperture radar (SAR) images makes ship targets appear to be more structured and shaped and nevertheless contain many weak echoes in their resolution cells, which brings great challenges for accurate scene understanding. In this paper, inspired by the multilayer selective cognition property of the human visual system, we advance a new hierarchical saliency filtering method for fast and accurate ship detection in high-resolution SAR images. The saliency of targets is first explored to develop a random-forest-based hierarchical sparse model (HSM) for the selection of candidate target regions. Then, a dynamic constant-false-alarm-rate-based contour saliency model (CSM) is proposed to gradually filter out the false alarms from candidate regions and extract the target outlines for accurate detection. Because of a rapid capture of regions of interest in the HSM and dynamic false alarm removal in the CSM, our method can make efficient ship detection in high-resolution SAR images possible by working in a coarse-to-fine manner. Finally, the proposed ship detector is tested on high-resolution SAR data collected from TerraSAR and RADARSAT satellites, showing significant agreement with the ground truth. It is also compared with other classical ship detectors, in terms of both speed and accuracy, and shows superior performance, particularly in complex scenes.