Search

Eye-tracking technology is an innovative tool that holds promise for enhancing dementia screening. In this work, we introduce a novel way of extracting salient features directly from the raw eye-tracking data of a mixed sample of dementia patients during a novel instruction-less cognitive test. Our approach is based on self-supervised representation learning where, by training initially a deep neural network to solve a pretext task using well-defined available labels (e.g. recognising distinct cognitive activities in healthy individuals), the network encodes high-level semantic information which is useful for solving other problems of interest (e.g. dementia classification). Inspired by previous work in explainable AI, we use the Layer-wise Relevance Propagation (LRP) technique to describe our network's decisions in differentiating between the distinct cognitive activities. The extent to which eye-tracking features of dementia patients deviate from healthy behaviour is then explored, followed by a comparison between self-supervised and handcrafted representations on discriminating between participants with and without dementia. Our findings not only reveal novel self-supervised learning features that are more sensitive than handcrafted features in detecting performance differences between participants with and without dementia across a variety of tasks, but also validate that instruction-less eye-tracking tests can detect oculomotor biomarkers of dementia-related cognitive dysfunction. This work highlights the contribution of self-supervised representation learning techniques in biomedical applications where the small number of patients, the non-homogenous presentations of the disease and the complexity of the setting can be a challenge using state-of-the-art feature extraction methods.

Accurate and realistic simulation of high-dimensional medical images has become an important research area relevant to many AI-enabled healthcare applications. However, current state-of-the-art approaches lack the ability to produce satisfactory high-resolution and accurate subject-specific images. In this work, we present a deep learning framework, namely 4D-Degenerative Adversarial NeuroImage Net (4D-DANI-Net), to generate high-resolution, longitudinal MRI scans that mimic subject-specific neurodegeneration in ageing and dementia. 4D-DANI-Net is a modular framework based on adversarial training and a set of novel spatiotemporal, biologically-informed constraints. To ensure efficient training and overcome memory limitations affecting such high-dimensional problems, we rely on three key technological advances: i) a new 3D training consistency mechanism called Profile Weight Functions (PWFs), ii) a 3D super-resolution module and iii) a transfer learning strategy to fine-tune the system for a given individual. To evaluate our approach, we trained the framework on 9852 T1-weighted MRI scans from 876 participants in the Alzheimer’s Disease Neuroimaging Initiative dataset and held out a separate test set of 1283 MRI scans from 170 participants for quantitative and qualitative assessment of the personalised time series of synthetic images. We performed three evaluations: i) image quality assessment; ii) quantifying the accuracy of regional brain volumes over and above benchmark models; and iii) quantifying visual perception of the synthetic images by medical experts. Overall, both quantitative and qualitative results show that 4D-DANI-Net produces realistic, low-artefact, personalised time series of synthetic T1 MRI that outperforms benchmark models.

The use of hyperspectral imaging for medical applications is becoming more common in recent years. One of the main obstacles that researchers find when developing hyperspectral algorithms for medical applications is the lack of specific, publicly available, and hyperspectral medical data. The work described in this paper was developed within the framework of the European project HELICoiD (HypErspectraL Imaging Cancer Detection), which had as a main goal the application of hyperspectral imaging to the delineation of brain tumors in real-time during neurosurgical operations. In this paper, the methodology followed to generate the first hyperspectral database of in-vivo human brain tissues is presented. Data was acquired employing a customized hyperspectral acquisition system capable of capturing information in the Visual and Near InfraRed (VNIR) range from 400 to 1000 nm. Repeatability was assessed for the cases where two images of the same scene were captured consecutively. The analysis reveals that the system works more efficiently in the spectral range between 450 and 900 nm. A total of 36 hyperspectral images from 22 different patients were obtained. From these data, more than 300 000 spectral signatures were labeled employing a semi-automatic methodology based on the spectral angle mapper algorithm. Four different classes were defined: normal tissue, tumor tissue, blood vessel, and background elements. All the hyperspectral data has been made available in a public repository.

Purpose Probe-based confocal laser endomicroscopy (pCLE) enables performing an optical biopsy via a probe. pCLE probes consist of multiple optical fibres arranged in a bundle, which taken together generate signals in an irregularly sampled pattern. Current pCLE reconstruction is based on interpolating irregular signals onto an over-sampled Cartesian grid, using a naive linear interpolation. It was shown that convolutional neural networks (CNNs) could improve pCLE image quality. Yet classical CNNs may be suboptimal in regard to irregular data. Methods We compare pCLE reconstruction and super-resolution (SR) methods taking irregularly sampled or reconstructed pCLE images as input. We also propose to embed a Nadaraya–Watson (NW) kernel regression into the CNN framework as a novel trainable CNN layer. We design deep learning architectures allowing for reconstructing high-quality pCLE images directly from the irregularly sampled input data. We created synthetic sparse pCLE images to evaluate our methodology. Results The results were validated through an image quality assessment based on a combination of the following metrics: peak signal-to-noise ratio and the structural similarity index. Our analysis indicates that both dense and sparse CNNs outperform the reconstruction method currently used in the clinic. Conclusion The main contributions of our study are a comparison of sparse and dense approach in pCLE image reconstruction. We also implement trainable generalised NW kernel regression as a novel sparse approach. We also generated synthetic data for training pCLE SR.

Diffusion-Weighted Magnetic Resonance Imaging (DW-MRI) explores the random motion of diffusing water molecules in biological tissue and can provide information on the tissue structure at a microscopic scale. DW-MRI in used in many applications both in the brain and other parts of the body such as the breast and prostate, and novelcomputational methods are at the core of advancements in DW-MRI, both in terms of research and its clinical translation. This article reviews the ways in whichmachine learning anddeep learning is currently applied in DW-MRI. We will also discuss the more traditional methods used for processing diffusion MRI and the potential of deep learning in augmenting these existing methods in the future.

Simulating images representative of neurodegenerative diseases is important for predicting patient outcomes and for validation of computational models of disease progression. This capability is valuable for secondary prevention clinical trials where outcomes and screening criteria involve neuroimaging. Traditional computational methods are limited by imposing a parametric model for atrophy and are extremely resource-demanding. Recent advances in deep learning have yielded data-driven models for longitudinal studies (e.g., face ageing) that are capable of generating synthetic images in real-time. Similar solutions can be used to model trajectories of atrophy in the brain, although new challenges need to be addressed to ensure accurate disease progression modelling. Here we propose Degenerative Adversarial NeuroImage Net (DaniNet)—a new deep learning approach that learns to emulate the effect of neurodegeneration on MRI by simulating atrophy as a function of ages, and disease progression. DaniNet uses an underlying set of Support Vector Regressors (SVRs) trained to capture the patterns of regional intensity changes that accompany disease progression. DaniNet produces whole output images, consisting of 2D-MRI slices that are constrained to match regional predictions from the SVRs. DaniNet is also able to maintain the unique brain morphology of individuals. Adversarial training ensures realistic brain images and smooth temporal progression. We train our model using 9652 T1-weighted (longitudinal) MRI extracted from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset. We perform quantitative and qualitative evaluations on a separate test set of 1283 images (also from ADNI) demonstrating the ability of DaniNet to produce accurate and convincing synthetic images that emulate disease progression.

In this paper we introduce the DCT-GIST image representation model which is useful to summarize the context of the scene. The proposed image descriptor addresses the problem of real-time scene context classification on devices with limited memory and low computational resources (e.g., mobile and other single sensor devices such as wearable cameras). Images are holistically represented starting from the statistics collected in the Discrete Cosine Transform (DCT) domain. Since the DCT coefficients are usually computed within the digital signal processor for the JPEG conversion/storage, the proposed solution allows to obtain an instant and "free of charge" image signature. The novel image representation exploits the DCT coefficients of natural images by modelling them as Laplacian distributions which are summarized by the scale parameter in order to capture the context of the scene. Only discriminative DCT frequencies corresponding to edges and textures are retained to build the descriptor of the image. A spatial hierarchy approach allows to collect the DCT statistics on image sub-regions to better encode the spatial envelope of the scene. The proposed image descriptor is coupled with a Support Vector Machine classifier for context recognition purpose. Experiments on the well-known 8 Scene Context Dataset as well as on the MIT-67 Indoor Scene dataset demonstrate that the proposed representation technique achieves better results with respect to the popular GIST descriptor, outperforming this last representation also in terms of computational costs. Moreover, the experiments pointed out that the proposed representation model closely matches other state-of-the-art methods based on bag of Textons collected on spatial hierarchy. HighlightsA new image descriptor for scene context classification purpose.The descriptor is suitable for Image Generation Pipeline of single sensor devices.The descriptor is computed directly on compressed domain (JPEG).The descriptor is computed in realtime on platform with low computational resources.The extraction process does not need extra information (e.g., a visual vocabulary).

Hyperspectral imaging (HSI) allows for the acquisition of large numbers of spectral bands throughout the electromagnetic spectrum (within and beyond the visual range) with respect to the surface of scenes captured by sensors. Using this information and a set of complex classification algorithms, it is possible to determine which material or substance is located in each pixel. The work presented in this paper aims to exploit the characteristics of HSI to develop a demonstrator capable of delineating tumor tissue from brain tissue during neurosurgical operations. Improved delineation of tumor boundaries is expected to improve the results of surgery. The developed demonstrator is composed of two hyperspectral cameras covering a spectral range of 400-1700 nm. Furthermore, a hardware accelerator connected to a control unit is used to speed up the hyperspectral brain cancer detection algorithm to achieve processing during the time of surgery. A labeled dataset comprised of more than 300,000 spectral signatures is used as the training dataset for the supervised stage of the classification algorithm. In this preliminary study, thematic maps obtained from a validation database of seven hyperspectral images of in vivo brain tissue captured and processed during neurosurgical operations demonstrate that the system is able to discriminate between normal and tumor tissue in the brain. The results can be provided during the surgical procedure (~1 min), making it a practical system for neurosurgeons to use in the near future to improve excision and potentially improve patient outcomes.

Recent advances in hyperspectral imaging have made it a promising solution for intra-operative tissue characterization, with the advantages of being non-contact, non-ionizing and non-invasive. Working with hyperspectral images in vivo, however, is not straightforward as the high dimensionality of the data makes real-time processing challenging. In this paper, a novel dimensionality reduction scheme and a new processing pipeline are introduced to obtain a detailed tumour classification map for intra-operative margin definition during brain surgery. However, existing approaches to dimensionality reduction based on manifold embedding can be time consuming and may not guarantee a consistent result, thus hindering final tissue classification. The proposed framework aims to overcome these problems through a process divided into two steps: dimensionality reduction based on an extension of the T-distributed stochastic neighbour (t-SNE) approach is first performed and then a semantic segmentation technique is applied to the embedded results by using a Semantic Texton Forest (STF) for tissue classification. Detailed in vivo validation of the proposed method has been performed to demonstrate the potential clinical value of the system.

Recent studies have shown that air pollution has a negative impact on people's health, especially for patients with respiratory and cardiac diseases (e.g. COPD, asthma, ischemic heart disease). Although there are already many air quality monitoring stations in major cities, such as London, these stations are sparsely located, and the periodic collection of information is insufficient to provide the granularity needed to assess the environmental risk for an individual (e.g. to avoid exacerbation). Wearable devices, on the other hand, are more suitable in this context, providing a better estimation of the air quality in the proximity of the person. Therefore, relevant warnings and information on health risks can be provided in real-time. As a proof of concept, we have developed a wearable sensor for continuous monitoring of air quality around the user, and a preliminary study was conducted to validate the sensor and assess the air quality in London underground stations. Based on the PM2.5 (particulate matter with a diameter of 2.5 µm), temperature and location information, a model is generated for predicting the air quality of each station at different times. Our preliminary results have shown that there are significant differences in air quality among stations and metro lines. It also demonstrates that wearable sensors can provide necessary information for users to make travel arrangements that minimize their exposure to polluted air.

Content-aware image resizing techniques allow to take into account the visual content of images during the resizing process. The basic idea beyond these algorithms is the removal of vertical and/or horizontal paths of pixels (i.e., seams) containing low salient information. In this paper, we present a method which exploits the gradient vector flow (GVF) of the image to establish the paths to be considered during the resizing. The relevance of each GVF path is straightforward derived from an energy map related to the magnitude of the GVF associated to the image to be resized. To make more relevant, the visual content of the images during the content-aware resizing, we also propose to select the generated GVF paths based on their visual saliency properties. In this way, visually important image regions are better preserved in the final resized image. The proposed technique has been tested, both qualitatively and quantitatively, by considering a representative data set of 1000 images labeled with corresponding salient objects (i.e., ground-truth maps). Experimental results demonstrate that our method preserves crucial salient regions better than other state-of-the-art algorithms.

The detection of near duplicate images in large databases, such as the ones of popular social networks, digital investigation archives, and surveillance systems, is an important task for a number of image forensics applications. In digital investigation, hashing techniques are commonly used to index large quantities of images for the detection of copies belonging to different archives. In the last few years, different image hashing techniques based on the Bags of Visual Features paradigm appeared in literature. Recently, this paradigm has been augmented by using multiple descriptors (e.g., Bags of Visual Phrases) in order to exploit the coherence between different feature spaces. In this paper we propose to further improve the Bags of Visual Phrases approach considering the coherence between feature spaces not only at the level of image representation, but also during the codebook generation phase. Also we introduce a novel image database specifically designed for the development and benchmarking of near duplicate image retrieval techniques. The dataset consists of more than 3,300 images depicting more than 500 different scenes having at least three real near duplicates. The dataset has a huge variability in terms of geometric and photometric transformations between scenes and their corresponding near duplicates. Finally, we suggest a method to compress the proposed image representation for storage purposes. Experiments show the effectiveness of the proposed near duplicate retrieval technique, which outperforms the original Bags of Visual Phrases approach.

With a massive influx of multimodality data, the role of data analytics in health informatics has grown rapidly in the last decade. This has also prompted increasing interests in the generation of analytical, data driven models based on machine learning in health informatics. Deep learning, a technique with its foundation in artificial neural networks, is emerging in recent years as a powerful tool for machine learning, promising to reshape the future of artificial intelligence. Rapid improvements in computational power, fast data storage, and parallelization have also contributed to the rapid uptake of the technology in addition to its predictive power and ability to generate automatically optimized high-level features and semantic interpretation from the input data. This article presents a comprehensive up-to-date review of research employing deep learning in health informatics, providing a critical analysis of the relative merit, and potential pitfalls of the technique as well as its future outlook. The paper mainly focuses on key applications of deep learning in the fields of translational bioinformatics, medical imaging, pervasive sensing, medical informatics, and public health.

The increasing popularity of wearable devices in recent years means that a diverse range of physiological and functional data can now be captured continuously for applications in sports, wellbeing, and healthcare. This wealth of information requires efficient methods of classification and analysis where deep learning is a promising technique for large-scale data analytics. While deep learning has been successful in implementations that utilize high-performance computing platforms, its use on low-power wearable devices is limited by resource constraints. In this paper, we propose a deep learning methodology, which combines features learned from inertial sensor data together with complementary information from a set of shallow features to enable accurate and real-time activity classification. The design of this combined method aims to overcome some of the limitations present in a typical deep learning framework where on-node computation is required. To optimize the proposed method for real-time on-node computation, spectral domain preprocessing is used before the data are passed onto the deep learning framework. The classification accuracy of our proposed deep learning approach is evaluated against state-of-the-art methods using both laboratory and real world activity datasets. Our results show the validity of the approach on different human activity datasets, outperforming other methods, including the two methods used within our combined pipeline. We also demonstrate that the computation times for the proposed method are consistent with the constraints of real-time on-node processing on smartphones and a wearable sensor platform.

Human Activity Recognition provides valuable contextual information for wellbeing, healthcare, and sport applications. Over the past decades, many machine learning approaches have been proposed to identify activities from inertial sensor data for specific applications. Most methods, however, are designed for offline processing rather than processing on the sensor node. In this paper, a human activity recognition technique based on a deep learning methodology is designed to enable accurate and real-time classification for low-power wearable devices. To obtain invariance against changes in sensor orientation, sensor placement, and in sensor acquisition rates, we design a feature generation process that is applied to the spectral domain of the inertial data. Specifically, the proposed method uses sums of temporal convolutions of the transformed input. Accuracy of the proposed approach is evaluated against the current state-of-the-art methods using both laboratory and real world activity datasets. A systematic analysis of the feature generation parameters and a comparison of activity recognition computation times on mobile devices and sensor nodes are also presented.

This paper presents an approach for generating class-specific image segmentation. We introduce two novel features that use the quantized data of the Discrete Cosine Transform (DCT) in a Semantic Texton Forest based framework (STF), by combining together colour and texture information for semantic segmentation purpose. The combination of multiple features in a segmentation system is not a straightforward process. The proposed system is designed to exploit complementary features in a computationally efficient manner. Our DCT based features describe complex textures represented in the frequency domain and not just simple textures obtained using differences between intensity of pixels as in the classic STF approach. Differently than existing methods (e.g., filter bank) just a limited amount of resources is required. The proposed method has been tested on two popular databases: CamVid and MSRC-v2. Comparison with respect to recent state-of-the-art methods shows improvement in terms of semantic segmentation accuracy. HighlightsA method for semantic image segmentation based on random forests.Novel texture features based on Discrete Cosine Transform to describe image regions.The method uses a limited amount of resources and works in realtime.The approach shows good performance overcoming other state of the art.The system obtains a better accuracy on small classes (i.e., Pedestrians).

© 2015 IEEE.Assessment of food intake has a wide range of applications in public health and life-style related chronic disease management. In this paper, we propose a real-time food recognition platform combined with daily activity and energy expenditure estimation. In the proposed method, food recognition is based on hierarchical classification using multiple visual cues, supported by efficient software implementation suitable for realtime mobile device execution. A Fischer Vector representation together with a set of linear classifiers are used to categorize food intake. Daily energy expenditure estimation is achieved by using the built-in inertial motion sensors of the mobile device. The performance of the vision-based food recognition algorithm is compared to the current state-of-the-art, showing improved accuracy and high computational efficiency suitable for realtime feedback. Detailed user studies have also been performed to demonstrate the practical value of the software environment.

Red eye artifact is caused by the flash light reflected off a person’s retina. This effect often occurs when the flash light is very close to the camera lens, as in most compact imaging devices. To reduce these artifacts, most cameras have a red eye flash mode which fires a series of preflashes prior to picture capture. The major disadvantage of the preflash approach is power consumption (e.g., flash is the most power-consuming device on the camera). Alternatively, red eyes can be detected after photo acquisition. Some photo-editing softwares make use of red eye removal tools that require considerable user interaction. To overcome this problem, different techniques have been proposed in literature. Due to the growing interest of industry, many automatic algorithms, embedded on commercial software, have been patented in the last decade. The huge variety of approaches has permitted research to explore different aspects and problems of red eyes identification and correction. The big challenge now is to obtain the best results with the minimal number of visual errors. This chapter critically reviews some of the state-of-the-art approaches for red eye removal. We also discuss a recent technique whose strength is due to a multimodal classifier which is obtained by combining clustering and boosting in order to recognize red eyes represented in the gray codes feature space.

Content-aware image resizing is an effective technique that allows to take into account the visual content of images during the resizing process. The basic idea beyond these algorithms is the resizing of an image by considering vertical and/or horizontal paths of pixels (i.e., seams) which contain low salient information. In this paper we exploit the Gradient Vector Flow (GVF) of the image to establish the paths to be considered during the resizing. The relevance of each path is derived from a saliency map obtained by considering the magnitude of the GVF associated to the image under consideration. The proposed technique has been tested, both qualitatively and quantitatively, by considering a representative set of images labeled with corresponding salient objects (i.e., ground-truth maps). Experimental results demonstrate that our method preserves crucial salient regions better than other state-of-the-art algorithms.

Computer Vision enables mobile devices to extract the meaning of the observed scene from the information acquired with the onboard sensor cameras. Nowadays, there is a growing interest in Computer Vision algorithms able to work on mobile platform (e.g., phone camera, point-and-shot-camera, etc.). Indeed, bringing Computer Vision capabilities on mobile devices open new opportunities in different application contexts. The implementation of vision algorithms on mobile devices is still a challenging task since these devices have poor image sensors and optics as well as limited processing power. In this paper we have considered different algorithms covering classic Computer Vision tasks: keypoint extraction, face detection, image segmentation. Several tests have been done to compare the performances of the involved mobile platforms: Nokia N900, LG Optimus One, Samsung Galaxy SII.

Scene recognition is extremely useful to improve different tasks involved in the Image Generation Pipeline of single sensor mobile devices (e.g., white balancing, autoexposure, etc). This demo showcases our scene recognition engine implemented on a Nokia N900 smartphone. The engine exploits an image representation directly obtainable in the IGP of mobile devices. The demo works in realtime and it is able to discriminate among different classes of scenes. The framework is built by employing the FCam API to have an easy and precise control of the mobile digital camera. Each acquired image (or frame of a video) is holistically represented starting from the statistics collected on DCT domain. This allow instant and "free of charge" feature extraction process since the DCT is always computed into the IGP of a mobile for storage purposes (i.e., JPEG or MPEG format). A SVM classifier is used to perform the final inference about the context of the scene.

Image retrieval from large databases, such as popular social networks, collections of surveillance images and videos, or digital investigation archives, is a very important task for a number of applications. In digital investigation, hashing techniques are commonly used to index large quantities of images to detect copies from different archives. In the last few years, a number of image hashing techniques based on the Bags of Visual Words paradigm have been proposed. Recently, this paradigm has been augmented by using multiple descriptors (Bags of Visual Phrases) to exploit the coherence between different feature spaces. In this paper we propose to further improve the Bags of Visual Phrases approach exploiting the coherence between feature spaces not only in the image representation, but also in the codebooks generation. Experiments performed on real and synthetic near duplicate image datasets show the effectiveness of the proposed approach, which outperforms the original Bags of Visual Phrases approach.

Red-eye artifact is a well-known problem in digital photography. Since the large diffusion of mobile devices with embedded camera and flashgun, automatic detection and correction of red-eyes have become an important task. In this paper we describe a technique that makes use of three steps to identify and correct red-eyes. First, red-eye candidates are extracted from the input image by using simple color segmentation coupled with geometrical constraints. A set of linear discriminant classifiers is then learned on the clustered patches space, and hence employed to distinguish between eyes and non-eyes patches. The proposed cluster-based Linear Discriminant Analysis is used to deal with the multi-modally nature of the input space. The third step of the pipeline is devoted to artifacts correction through de-saturation and brightness reduction. Experimental results on a large dataset of images demonstrate the effectiveness of the pro- posed pipeline that outperforms other existing solutions in terms of hit rates maximization, false positives reduction and ad-hoc quality measure.

Since the large diffusion of digital camera and mobile devices with embedded camera and flashgun, the red-eyes artifacts have de-facto become a critical problem. The technique herein described makes use of three main steps to identify and remove red-eyes. First, red eyes candidates are extracted from the input image by using an image filtering pipeline. A set of classifiers is then learned on gray code features extracted in the clustered patches space, and hence employed to distinguish between eyes and non-eyes patches. Once red-eyes are detected, artifacts are removed through desaturation and brightness reduction. The proposed method has been tested on large dataset of images achieving effective results in terms of hit rates maximization, false positives reduction and quality measure.

This paper proposes a method to recognize scene categories using bags of visual words obtained by hierarchically partitioning into subregion the input images. Specifically, for each subregion the Textons distribution and the extension of the corresponding subregion are taken into account. The bags of visual words computed on the subregions are weighted and used to represent the whole scene. The classification of scenes is carried out by discriminative methods (i.e., SVM, KNN). A similarity measure based on Bhattacharyya coefficient is proposed to establish similarities between images, represented as hierarchy of bags of visual words. Experimental tests, using fifteen different scene categories, show that the proposed approach achieves good performances with respect to the state-of-the-art methods.

The proposed paper describes a compression test analysis of JBIG standard algorithm. The aim of such work is to proof the effectiveness of this standard for images acquired through scanners and processed into a printer pipeline. The main issue of printer pipelines is the necessity to use a memory buffer to store scanned images for multiple prints. This work demonstrates that for very large scales the buffer can be fixed using medium compression case, using multiple scans in case of uncommon random patterns.

This paper proposes a method to recognize scene categories using bags of visual words obtained hierarchically partitioning into subregion the input images. Specifically, for each subregion the Textons distribution and the extension of the corresponding subregion are taken into account. The bags of visual words computed on the subregions are weighted and used to represent the whole scene. The classification of scenes is carried out by a Support Vector Machine. A k-nearest neighbor algorithm and a similarity measure based on Bhattacharyya coefficient are used to retrieve from the scene database those that contain similar visual content to a given a scene used as query. Experimental tests using fifteen different scene categories show that the proposed approach achieves good performances with respect to the state of the art methods.

This paper proposes a method to recognize scene categories using bags of visual words obtained hierarchically partitioning into subregion the input images. Specifically, for each subregions the texton histogram and the extension of the sub-region is taken into account. The bags of visual words, obtained in this way, are weighted and used in a similarity measure during the categorization. Experimental tests using ten different scene categories show that the proposed approach achieves good performances with respect to the state of the art methods.

Since the large diffusion of digital camera and mobile devices with embedded camera and flashgun, the redeyes artifacts have de facto become a critical problem. The technique herein described makes use of three main steps to identify and remove red eyes. First, red-eye candidates are extracted from the input image by using an image filtering pipeline. A set of classifiers is then learned on gray code features extracted in the clustered patches space and hence employed to distinguish between eyes and non-eyes patches. Specifically, for each cluster the gray code of the red-eyes candidate is computed and some discriminative gray code bits are selected employing a boosting approach. The selected gray code bits are used during the classification to discriminate between eye versus non-eye patches. Once red-eyes are detected, artifacts are removed through desaturation and brightness reduction. Experimental results on a large dataset of images demonstrate the effectiveness of the proposed pipeline that outperforms other existing solutions in terms of hit rates maximization, false positives reduction, and quality measure.