Perceptually-based Comparison of Image Similarity Metrics

The image comparison operation ??sessing how well one image matches another ??rms a critical component of many image analysis systems and models of human visual processing. Two norms used commonly for this purpose are L1 and L2, which are specific instances of the Minkowski metric. However, there is often not a principled reason for selecting one norm over the other. One way to address this problem is by examining whether one metric better captures the perceptual notion of image similarity than the other. With this goal, we examined perceptual preferences for images retrieved on the basis of the L1 versus the L2 norm. These images were either small fragments without recognizable content, or larger patterns with recognizable content created via vector quantization. In both conditions the subjects showed a consistent preference for images matched using the L1 metric. These results suggest that, in the domain of natural images of the kind we have used, the L1 metric may better capture human notions of image similarity.

Global Depth Perception from Familiar Scene Structure

In the absence of cues for absolute depth measurements as binocular disparity, motion, or defocus, the absolute distance between the observer and a scene cannot be measured. The interpretation of shading, edges and junctions may provide a 3D model of the scene but it will not inform about the actual "size" of the space. One possible source of information for absolute depth estimation is the image size of known objects. However, this is computationally complex due to the difficulty of the object recognition process. Here we propose a source of information for absolute depth estimation that does not rely on specific objects: we introduce a procedure for absolute depth estimation based on the recognition of the whole scene. The shape of the space of the scene and the structures present in the scene are strongly related to the scale of observation. We demonstrate that, by recognizing the properties of the structures present in the image, we can infer the scale of the scene, and therefore its absolute mean depth. We illustrate the interest in computing the mean depth of the scene with application to scene recognition and object detection.

Classification-based probabilistic modeling of texture transition for fast line search tracking and delineation

We introduce a classification-based approach to finding occluding texture boundaries. The classifier is composed of a set of weak learners, which operate on image intensity discriminative features that are defined on small patches and are fast to compute. A database that is designed to simulate digitized occluding contours of textured objects in natural images is used to train the weak learners. The trained classifier score is then used to obtain a probabilistic model for the presence of texture transitions, which can readily be used for line search texture boundary detection in the direction normal to an initial boundary estimate. This method is fast and therefore suitable for real-time and interactive applications. It works as a robust estimator, which requires a ribbon-like search region and can handle complex texture structures without requiring a large number of observations. We demonstrate results both in the context of interactive 2D delineation and of fast 3D tracking and compare its performance with other existing methods for line search boundary detection.

Segmentação fuzzy de imagens coloridas com características texturais: uma aplicação a rochas sedimentares

Image segmentation is the process of labeling pixels on di erent objects, an important step in many image processing systems. This work proposes a clustering method for the segmentation of color digital images with textural features. This is done by reducing the dimensionality of histograms of color images and using the Skew Divergence to calculate the fuzzy a nity functions. This approach is appropriate for segmenting images that have colorful textural features such as geological, dermoscopic and other natural images, as images containing mountains, grass or forests. Furthermore, experimental results of colored texture clustering using images of aquifers' sedimentary porous rocks are presented and analyzed in terms of precision to verify its e ectiveness.

Looking left with left neglect: the role of spatial attention when active vision selects local image features for fixation

When we actively explore the visual environment, our gaze preferentially selects regions characterized by high contrast and high density of edges, suggesting that the guidance of eye movements during visual exploration is driven to a significant degree by perceptual characteristics of a scene. Converging findings suggest that the selection of the visual target for the upcoming saccade critically depends on a covert shift of spatial attention. However, it is unclear whether attention selects the location of the next fixation uniquely on the basis of global scene structure or additionally on local perceptual information. To investigate the role of spatial attention in scene processing, we examined eye fixation patterns of patients with spatial neglect during unconstrained exploration of natural images and compared these to healthy and brain-injured control participants. We computed luminance, colour, contrast, and edge information contained in image patches surrounding each fixation and evaluated whether they differed from randomly selected image patches. At the global level, neglect patients showed the characteristic ipsilesional shift of the distribution of their fixations. At the local level, patients with neglect and control participants fixated image regions in ipsilesional space that were closely similar with respect to their local feature content. In contrast, when directing their gaze to contralesional (impaired) space neglect patients fixated regions of significantly higher local luminance and lower edge content than controls. These results suggest that intact spatial attention is necessary for the active sampling of local feature content during scene perception.

Progressive Image Denoising

Image denoising continues to be an active research topic. Although state-of-the-art denoising methods are numerically impressive and approch theoretical limits, they suffer from visible artifacts.While they produce acceptable results for natural images, human eyes are less forgiving when viewing synthetic images. At the same time, current methods are becoming more complex, making analysis, and implementation difficult. We propose image denoising as a simple physical process, which progressively reduces noise by deterministic annealing. The results of our implementation are numerically and visually excellent. We further demonstrate that our method is particularly suited for synthetic images. Finally, we offer a new perspective on image denoising using robust estimators.

Propagation, structural similarity and image quality

Póster presentado en SPIE Photonics Europe, Brussels, 16-19 April 2012.

Retinal image quality assessment through a visual similarity index

Retinal image quality is commonly analyzed through parameters inherited from instrumental optics. These parameters are defined for ‘good optics’ so they are hard to translate into visual quality metrics. Instead of using point or artificial functions, we propose a quality index that takes into account properties of natural images. These images usually show strong local correlations that help to interpret the image. Our aim is to derive an objective index that quantifies the quality of vision by taking into account the local structure of the scene, instead of focusing on a particular aberration. As we show, this index highly correlates with visual acuity and allows inter-comparison of natural images around the retina. The usefulness of the index is proven through the analysis of real eyes before and after undergoing corneal surgery, which usually are hard to analyze with standard metrics.

The role of local luminance amplitude in the interpretation of shape-from-shading in textured surfaces

Luminance changes within a scene are ambiguous; they can indicate reflectance changes, shadows, or shading due to surface undulations. How does vision distinguish between these possibilities? When a surface painted with an albedo texture is shaded, the change in local mean luminance (LM) is accompanied by a similar modulation of the local luminance amplitude (AM) of the texture. This relationship does not necessarily hold for reflectance changes or for shading of a relief texture. Here we concentrate on the role of AM in shape-from-shading. Observers were presented with a noise texture onto which sinusoidal LM and AM signals were superimposed, and were asked to indicate which of two marked locations was closer to them. Shape-from-shading was enhanced when LM and AM co-varied (in-phase), and was disrupted when they were out-of-phase. The perceptual differences between cue types (in-phase vs out-of-phase) were enhanced when the two cues were present at different orientations within a single image. Similar results were found with a haptic matching task. We conclude that vision can use AM to disambiguate luminance changes. LM and AM have a positive relationship for rendered, undulating, albedo textures, and we assess the degree to which this relationship holds in natural images. [Supported by EPSRC grants to AJS and MAG].

Spatial and temporal dependencies of cross-orientation suppression in human vision.

A well-known property of orientation-tuned neurons in the visual cortex is that they are suppressed by the superposition of an orthogonal mask. This phenomenon has been explained in terms of physiological constraints (synaptic depression), engineering solutions for components with poor dynamic range (contrast normalization) and fundamental coding strategies for natural images (redundancy reduction). A common but often tacit assumption is that the suppressive process is equally potent at different spatial and temporal scales of analysis. To determine whether it is so, we measured psychophysical cross-orientation masking (XOM) functions for flickering horizontal Gabor stimuli over wide ranges of spatio-temporal frequency and contrast. We found that orthogonal masks raised contrast detection thresholds substantially at low spatial frequencies and high temporal frequencies (high speeds), and that small and unexpected levels of facilitation were evident elsewhere. The data were well fit by a functional model of contrast gain control, where (i) the weight of suppression increased with the ratio of temporal to spatial frequency and (ii) the weight of facilitatory modulation was the same for all conditions, but outcompeted by suppression at higher contrasts. These results (i) provide new constraints for models of primary visual cortex, (ii) associate XOM and facilitation with the transient magno- and sustained parvostreams, respectively, and (iii) reconcile earlier conflicting psychophysical reports on XOM.

What is second-order vision for? Discriminating illumination versus material changes

The human visual system is sensitive to second-order modulations of the local contrast (CM) or amplitude (AM) of a carrier signal. Second-order cues are detected independently of first-order luminance signals; however, it is not clear why vision should benet from second-order sensitivity. Analysis of the first-and second-order contents of natural images suggests that these cues tend to occur together, but their phase relationship varies. We have shown that in-phase combinations of LM and AM are perceived as a shaded corrugated surface whereas the anti-phase combination can be seen as corrugated when presented alone or as a flat material change when presented in a plaid containing the in-phase cue. We now extend these findings using new stimulus types and a novel haptic matching task. We also introduce a computational model based on initially separate first-and second-order channels that are combined within orientation and subsequently across orientation to produce a shading signal. Contrast gain control allows the LM + AM cue to suppress responses to the LM-AM when presented in a plaid. Thus, the model sees LM -AM as flat in these circumstances. We conclude that second-order vision plays a key role in disambiguating the origin of luminance changes within an image. © ARVO.

A new unsupervised convolutional neural network model for Chinese scene text detection

As one of the most popular deep learning models, convolution neural network (CNN) has achieved huge success in image information extraction. Traditionally CNN is trained by supervised learning method with labeled data and used as a classifier by adding a classification layer in the end. Its capability of extracting image features is largely limited due to the difficulty of setting up a large training dataset. In this paper, we propose a new unsupervised learning CNN model, which uses a so-called convolutional sparse auto-encoder (CSAE) algorithm pre-Train the CNN. Instead of using labeled natural images for CNN training, the CSAE algorithm can be used to train the CNN with unlabeled artificial images, which enables easy expansion of training data and unsupervised learning. The CSAE algorithm is especially designed for extracting complex features from specific objects such as Chinese characters. After the features of articficial images are extracted by the CSAE algorithm, the learned parameters are used to initialize the first CNN convolutional layer, and then the CNN model is fine-Trained by scene image patches with a linear classifier. The new CNN model is applied to Chinese scene text detection and is evaluated with a multilingual image dataset, which labels Chinese, English and numerals texts separately. More than 10% detection precision gain is observed over two CNN models.

Designing an adaptive salutogenic care environment

Humans are profoundly affected by the surroundings which they inhabit. Environmental psychologists have produced numerous credible theories describing optimal human environments, based on the concept of congruence or “fit” (1, 2). Lack of person/environment fit can lead to stress-related illness and lack of psychosocial well-being (3). Conversely, appropriately designed environments can promote wellness (4) or “salutogenesis” (5). Increasingly, research in the area of Evidence-Based Design, largely concentrated in the area of healthcare architecture, has tended to bear out these theories (6). Patients and long-term care residents, because of injury, illness or physical/ cognitive impairment, are less likely to be able to intervene to modify their immediate environment, unless this is designed specifically to facilitate their particular needs. In the context of care settings, detailed design of personal space therefore takes on enormous significance. MyRoom conceptualises a personalisable room, utilising sensoring and networked computing to enable the environment to respond directly and continuously to the occupant. Bio-signals collected and relayed to the system will actuate application(s) intended to positively influence user well-being. Drawing on the evidence base in relation to therapeutic design interventions (7), real-time changes in ambient lighting, colour, image, etc. respond continuously to the user’s physiological state, optimising congruence. Based on research evidence, consideration is also given to development of an application which uses natural images (8). It is envisaged that actuation will require machine-learning based on interpretation of data gathered by sensors; sensoring arrangements may vary depending on context and end-user. Such interventions aim to reduce inappropriate stress/ provide stimulation, supporting both instrumental and cognitive tasks.

Hierarchically grouped 2D local features applied to edge contour localisation

One of the most significant research topics in computer vision is object detection. Most of the reported object detection results localise the detected object within a bounding box, but do not explicitly label the edge contours of the object. Since object contours provide a fundamental diagnostic of object shape, some researchers have initiated work on linear contour feature representations for object detection and localisation. However, linear contour feature-based localisation is highly dependent on the performance of linear contour detection within natural images, and this can be perturbed significantly by a cluttered background. In addition, the conventional approach to achieving rotation-invariant features is to rotate the feature receptive field to align with the local dominant orientation before computing the feature representation. Grid resampling after rotation adds extra computational cost and increases the total time consumption for computing the feature descriptor. Though it is not an expensive process if using current computers, it is appreciated that if each step of the implementation is faster to compute especially when the number of local features is increasing and the application is implemented on resource limited ”smart devices”, such as mobile phones, in real-time. Motivated by the above issues, a 2D object localisation system is proposed in this thesis that matches features of edge contour points, which is an alternative method that takes advantage of the shape information for object localisation. This is inspired by edge contour points comprising the basic components of shape contours. In addition, edge point detection is usually simpler to achieve than linear edge contour detection. Therefore, the proposed localization system could avoid the need for linear contour detection and reduce the pathological disruption from the image background. Moreover, since natural images usually comprise many more edge contour points than interest points (i.e. corner points), we also propose new methods to generate rotation-invariant local feature descriptors without pre-rotating the feature receptive field to improve the computational efficiency of the whole system. In detail, the 2D object localisation system is achieved by matching edge contour points features in a constrained search area based on the initial pose-estimate produced by a prior object detection process. The local feature descriptor obtains rotation invariance by making use of rotational symmetry of the hexagonal structure. Therefore, a set of local feature descriptors is proposed based on the hierarchically hexagonal grouping structure. Ultimately, the 2D object localisation system achieves a very promising performance based on matching the proposed features of edge contour points with the mean correct labelling rate of the edge contour points 0.8654 and the mean false labelling rate 0.0314 applied on the data from Amsterdam Library of Object Images (ALOI). Furthermore, the proposed descriptors are evaluated by comparing to the state-of-the-art descriptors and achieve competitive performances in terms of pose estimate with around half-pixel pose error.

Definition of linear color models in the RGB vector color space to detect red peaches in orchard images taken under natural illumination

This work proposes the detection of red peaches in orchard images based on the definition of different linear color models in the RGB vector color space. The classification and segmentation of the pixels of the image is then performed by comparing the color distance from each pixel to the different previously defined linear color models. The methodology proposed has been tested with images obtained in a real orchard under natural light. The peach variety in the orchard was the paraguayo (Prunus persica var. platycarpa) peach with red skin. The segmentation results showed that the area of the red peaches in the images was detected with an average error of 11.6%; 19.7% in the case of bright illumination; 8.2% in the case of low illumination; 8.6% for occlusion up to 33%; 12.2% in the case of occlusion between 34 and 66%; and 23% for occlusion above 66%. Finally, a methodology was proposed to estimate the diameter of the fruits based on an ellipsoidal fitting. A first diameter was obtained by using all the contour pixels and a second diameter was obtained by rejecting some pixels of the contour. This approach enables a rough estimate of the fruit occlusion percentage range by comparing the two diameter estimates.