BACKGROUND: A Virtual Screening algorithm has to adapt to the different stages of this process. Early screening needs to ensure that all bioactive compounds are ranked in the first positions despite of the number of false positives, while a second screening round is aimed at increasing the prediction accuracy. RESULTS: A novel CNN architecture is presented to this aim, which predicts bioactivity of candidate compounds on CDK1 using a combination of molecular fingerprints as their vector representation, and has been trained suitably to achieve good results as regards both enrichment factor and accuracy in different screening modes (98.55% accuracy in active-only selection, and 98.88% in high precision discrimination). CONCLUSION: The proposed architecture outperforms state-of-the-art ML approaches, and some interesting insights on molecular fingerprints are devised.

Copy-move forgery is one of the most common types of tampering for digital images. Detection methods generally use block-matching approaches, which first divide the image into overlapping blocks and then extract and compare features to find similar ones, or point-based approaches, in which relevant keypoints are extracted and matched to each other to find similar areas. In this paper, we present a very novel hybrid approach, which compares triangles rather than blocks, or single points. Interest points are extracted from the image, and objects are modeled as a set of connected triangles built onto these points. Triangles are matched according to their shapes (inner angles), their content (color information), and the local feature vectors extracted onto the vertices of the triangles. Our methods are designed to be robust to geometric transformations. Results are compared with a state-of-the-art block matching method and a point-based method. Furthermore, our data set is available for use by academic researchers.

In this paper we present the prototype of a teledentistry system to perform the remote diagnosis of oral diseases. It makes use of a particular device called intra-oral (or dental) camera properly designed to shoot video and take pictures of the inner part of the mouth. The intra-oral cameras can be connected via USB to a common PC and they are very cheap, unlike the intra-oral photography kit for DSLR cameras. Usually this kind of devices are used in dentistry studies for local visualization by means of specialized software. The novelty of our system is that the real-time video produced by this device is canalized into a video streming by means of Video LAN client server (VLC) and pictures can be sent by a current File Transfer Protocol (FTP) service to realize a Second Opinion System.

We here introduce an image/video retargeting method that operates arbitrary aspect ratios resizing achieved in real-time performances. Most of the literature retargeting approaches sacrifice real-time performances in behalf of quality. On the other hand, existing fast methods provide arguable results. We can obtain a valuable trade-off between effectiveness and efficiency. The method named Spring Simulation Retargeting (SSR) is mainly based on a physical springs-based simulation. The media are assumed as flexible objects composed of particles and springs with different local stiffness properties, related to the visual importance of the content. The variation of the object size generates elastic forces which determine a new arrangement of the particles, according to the elongation of their connected springs. The deformations are mostly introduced in regions where low importance content is present, while high saliency regions are preserved as desired. The proposed method is evaluated and compared both for images and videos, against several state of the art methods and a user study is taken to assess the results, showing the value of the approach.

The immersive panoramas are widely used to provide virtual tours of real scene. Their use covers a wide field of applications: art, industry, space research, topography, forensic investigation and all those systems requiring the exploration of a virtual environment which simulates a real one. Often sophisticated devices are used to perform the panorama acquisition. In this paper, we present an image based immersive panorama requiring low cost devices for the acquisition task and provides an innovative human-computer interaction approach. Many panoramic images of the same location are captured. The visualization system changes the panorama in a transparent way with respect to the user interaction in function of the position of the virtual point of view. Indeed, the new and more detailed panorama is loaded each time the current point of view position corresponds to a point of acquisition. Our solution is particularly suited to tele-immersive software environments, because it is lighter than a full 3D scene, but the user can lives a full exploration experience of a scene.

In this paper an efficient method for image retargeting is proposed. It relies onto a mechanical model based on springs network. Each pixel displacement (compression or expansion) is given by the network response, according to the springs stiffness. The properties of the springs are determined as function of the visual relevance of the pixels. Such model does not require any optimization, since its solution is obtained simply from a linear system of equations, allowing real-time application even for large images. The approach is fully automatic, though can be improved by interactively providing cues such as geometric constraints and/or manual relevant object labeling. The results prove that the presented method achieves results comparable or superior to reference methods, while improving efficiency.

This paper presents a study conducted for evaluating different interpolation schemes, similarity metrics and optimization algorithms for the purpose of volumetric medical image registration. Each technique has been implemented to be plugged in a modular system. Rotation, translation and scale error has been measured to obtain a performance evaluation for all of the combinations of the considered techniques. Several experimental tests were conducted for validation both on synthetic and real datasets providing an exhaustive overview of the various strategies used.

In this paper an automatic texture based volumetric region growing method for liver segmentation is proposed. 3D seeded region growing is based on texture features with the automatic selection of the seed voxel inside the liver organ and the automatic threshold value computation for the region growing stop condition. Co-occurrence 3D texture features are extracted from CT abdominal volumes and the seeded region growing algorithm is based on statistics in the features space. Each CT volume is composed by 230 slices, having 512 x 512 pixels as spatial resolution, and 12-bit gray level resolution. In this initial feasible study, 5 healthy volunteer acquisitions has been used. Tests have been performed on both basal phase and arterial phase images. Segmentation result shows the effectiveness of the proposed method: liver organ is correctly recognized and segmented, leaving out liver vessels form the segmented area and overcoming the “organ-splitting” problem. The goodness of the proposed method has been confirmed by manual liver segmentation results, having analogous and super-imposable behavior.

Copy-move forgeries are parts of the image that are duplicated elsewhere into the same image, often after being modified by geometrical transformations. In this paper we present a method to detect these image alterations, using a SIFT-based approach. First we describe a state of the art SIFT-point matching method, which inspired our algorithm, then we compare it with our SIFT-based approach, which consists of three parts: keypoint clustering, cluster matching, and texture analysis. The goal is to find copies of the same object, i.e. clusters of points, rather than points that match. Cluster matching proves to give better results than single point matching, since it returns a complete and coherent comparison between copied objects. At last, textures of matching areas are analyzed and compared to validate results and to eliminate false positives.

In this paper is presented an effective and efficient method for content-aware image resizing. It is based on the solution of a linear system where each pixel displacement (compression or expansion) is determined in dependence of the visual relevance of the pixel itself. The linear nature of the model allows real-time application of the method even for large images. This fully automatic approach can be also improved by interactively providing cues such as geometric constraints and/or manual relevant object labeling. The results have proven that the presented method achieves results comparable to existent strategies, while improving efficiency.

Old photos may be affected by several types of defects. Manual restorers use their own taxonomy to classify damages by which a photo is affected, in order to apply the proper restoration techniques for a specific defect. Once a photo is digitally acquired, defects become part of the image, and their aspect change. This paper wants to be a first attempt to correlate real defects of printed photos, and digital defects of their digitized versions. A dual taxonomy is proposed, for real and digital defects, and used to classify an image dataset, for a posteriori comparative study. Furthermore, a set of digital features is analyzed for digitized images, to identify which of them could be useful for an automatic inspection method.

This paper presents a study aimed to the realization of a novel multiresolution registration framework. The transformation function is computed iteratively as a composition of local deformations determined by the maximization of mutual information. At each iteration, local transformations are joint together using fuzzy kernel regression. This technique represents the core of the method and it's formally described from a probabilistic perspective. It avoids blocking artifacts and allows to keep the final deformation spatially congruent and smooth. Both qualitative and quantitative experimental results show that this approach is equally effective for registering datasets acquired from both single and multiple diagnostic modalities.

Bias artifact corrupts MRIs in such a way that the image is afflicted by illumination variations. Some of the authors proposed the exponential entropy-driven homomorphic unsharp masking ( E(2)D-HUM) algorithm that corrects this artifact without any a priori hypothesis about the tissues or the MRI modality. Moreover, E(2)D-HUM does not care about the body part under examination and does not require any particular training task. People who want to use this algorithm, which is Matlab-based, have to set their own computers in order to execute it. Furthermore, they have to be Matlab-skilled to exploit all the features of the algorithm. In this paper, we propose to make such algorithm available as a service on a grid infrastructure, so that people can use it almost from everywhere, in a pervasive fashion, by means of a suitable user interface running on smartphones. The proposed solution allows physicians to use the E(2)D-HUM algorithm (or any other kind of algorithm, given that it is available as a service on the grid), being it remotely executed somewhere in the grid, and the results are sent back to the user's device. This way, physicians do not need to be aware of how to use Matlab to process their images. The pervasive service provision for medical image enhancement is presented, along with some experimental results obtained using smartphones connected to an existing Globus-based grid infrastructure.

This paper presents a method aimed to noise removal in MRI (magnetic resonance imaging). We propose an improvement of Perona and Malik's anisotropic diffusion filter. In our schema, the diffusion equation of the filter has been modified to take into account the edges direction. This allows the filter to blur uniform areas, while it better preserves the edges. Both quantitative and qualitative evaluation is presented and the results are compared with other methods.

In this paper we propose a probabilistic approach for the automatic organization of pictures in personal photo album. Images are analyzed in term of faces and low-level visual features of the background. The description of the background is based on RGB color histogram and on Gabor filter energy accounting for texture information. The face descriptor is obtained by projection of detected and rectified faces on a common low dimensional eigenspace. Vectors representing faces and background are clustered in an unsupervised fashion exploiting a mean shift clustering technique. We observed that, given the peculiarity of the domain of personal photo libraries where most of the pictures contain faces of a relatively small number of different individuals, clusters tend to be not only visually but also semantically significant. Experimental results are reported

Digital image restoration aims to recover damaged zones of a digital image, using surrounding information. In this paper we propose a novel approach, based on bit-plane slicing decomposition, with the purpose to make information analysis and reconstruction process easy, fast and effective. Tests have been made on digitized damaged old photos to restore several classes of typical defects in old photographic prints.

A novel approach to the detection of multiple sclerosis (MS) lesions in T2and PD-weighted MR images is presented. The core of the proposed method is the use of the two channels fuzzy c-means (FCM) segmentation of data, where the classical FCM approach runs, at first, on the two separate spectra. Then, the one-dimensional distributions of the cluster centers obtained by FCM, are composed in the two-dimensional one, which is a-priori imposed to the two-spectra segmentation procedure. Images are preprocessed to expand their grey level dynamics, and to allow clustering of noise and soft brain tissues. The description of the whole system is reported, along with several comparative experiments where a. pool of physicians judged the outcomes of the presented approach with the T2-only, the PD-only, and the standard two spectra FCM segmentation.

The field of human-computer interaction has been widely investigated in the last years, resulting in a variety of systems used in different application fields like virtual reality simulation environments, software user interfaces, and digital library systems.A very crucial part of all these systems is the input module which is devoted to recognize the human operator in terms of tracking and/or recognition of human face, arms position, hand gestures, and so on.In this work a software architecture is presented, for the automatic recognition of human arms poses. Our research has been carried on in the robotics framework. A mobile robot that has to find its path to the goal in a partially structured environment can be trained by a human operator to follow particular routes in order to perform its task quickly. The system is able to recognize and classify some different poses of the operator's arms as direction commands like “turn-left”, “turn-right”, “go-straight”, and so on.A binary image of the operator silhouette is obtained from the gray-level input. Next, a slice centered on the silhouette itself is processed in order to compute the eigenvalues vector of the pixels co-variance matrix. This kind of information is strictly related to the shape of the contour of the operator figure, and can be usefully employed in order to assess the arms' position. Finally, a support vector machine (SVM) is trained in order to classify different poses, using the eigenvalues array.A detailed description of the system is presented along with some remarks on the statistical analysis we used, and on SVM. The experimental results, and an outline of the usability of the system as a generic shape classification tool are also reported.

In this work a software architecture is presented for the automatic recognition of human arm poses. Our research has been carried on in the robotics framework. A mobile robot that has to find its path to the goal in a partially structured environment can be trained by a human operator to follow particular routes in order to perform its task quickly. The system is able to recognize and classify some different poses of the operator's arms as direction commands like "turn-left", "turn-right", "go-straight", and so on. A binary image of the operator silhouette is obtained from the gray-level input. Next, a slice centered on the silhouette itself is processed in order to compute the eigenvalues vector of the pixels covariance matrix. Finally, a support vector machine is trained to classify different poses using the eigenvalues array. A detailed description of the system is presented. Experimental results and an outline of the usability of the system as a generic shape classification tool are reported.

Describes a knowledge-based approach for modelling and retrieving video data. Selected objects of interest in a video sequence are described and stored in a database. This database forms the object layer. On top of this layer, we define the schema layer that is used to capture the structured abstractions of the objects stored in the object layer. We propose two abstract languages on the basis of description logics: one for describing the contents of these layers and the other, which is more expressive, for making queries. The query language provides possibilities for navigation of the schema through forward and backward traversal of links, sub-setting of attributes and constraints on links.

The increasing development of advanced multimedia applications requires new technologies for organizing and retrieving by content databases of still digital images or digital video sequences. The authors describe JACOB, a prototypal system allowing content-based browsing and querying in video databases. The JACOB system automatically splits a video into a sequence of shots, extracts a few representative frames (said r-frames) from each shot and computes r-frame descriptors based on features like color and texture. No user action is required during the database population step. Queries exploit this image content description and may be direct or by example.

The increasing development of advanced multimedia applications requires new technologies for organizing and retrieving by content databases of digital video. Several content based features (color, texture, motion, etc.) are needed to perform a reliable content based retrieval. We present a method for automatic motion based video indexing and retrieval. A prototypal system has been developed to prove the validity of our approach. Our system automatically splits a video into a sequence of shots, extracts a few representative frames (said r-frames) from each shot and computes some motion based features related to the optical flow field. Motion based queries are then performed either in a qualitative or quantitative way. The results obtained with our system proved that motion based query can play a central role in content based video retrieval.

Indexing and retrieval methods based on the image content are required to effectively use information from the large repositories of digital images and videos currently available. Both global (colour, texture, motion, etc.) and local (object shape, etc.) features are needed to perform a reliable content based retrieval. We present a method for automatic extraction of global image features, like colour and motion parameters, and their use for data restriction in video database querying. Further retrieval is therefore accomplished, in a restricted set of images, by shape feature (skeleton, local symmetry moments, correlation, etc.) local search. The proposed indexing methodology has been developed and tested inside JACOB, a prototypal system for content-based video database querying.

In this paper we present a method for automatic motion and color based video indexing and retrieval. Our system automatically splits a video into a sequence of shots and extracts a few representative frames (r-frames) from each shot. For each r-frame we compute the optical flow field; motion features are then derived from the flow field. Color features are related to the three-dimensional RGB color histogram. Queries (direct or by example) are based on these features. Obtained results proved that motion and color based querying can play a central role in content based video retrieval.

The proposed architecture is aimed to recover 3-D shape information from gray-level images of a scene: to build a geometric representation of the scene in terms of geometric primitives; and to reason about the scene. The novelty of the architecture is in fact the integration of different approaches: symbolic reasoning techniques typical of knowledge representation in artificial intelligence, algorithmic capabilities typical of artificial vision schemes, and analogue techniques typical of artificial neural networks. Experimental results obtained by means of an implemented version of the proposed architecture acting on real scene images are reported to illustrate the system capabilities. © 1996 John Wiley & Sons, Inc.

Strong deficiencies are present in symbolic models for action representation and planning, regarding mainly the difficulty of coping with real, complex environments. These deficiencies can be attributed to several problems, such as the inadequacy in coping with incompletely structured situations, the difficulty of interacting with visual and motorial aspects, the difficulty in representing low-level knowledge, the need to specify the problem at a high level of detail, and so on. Besides the purely symbolic approaches, several nonsymbolic models have been developed, such as the recent class of subsym-bolic techniques. A promising paradigm for the modeling of reasoning, which combines features of both symbolic and analogical approaches, is based on the construction of analogical models of the reference for the internal representations, as introduced by Johnson-Laird. In this work, we propose a similar approach to the problem of knowledge representation and reasoning about actions and plans. We propose a hybrid approach, symbolic and analogical, in which the inferences are partially devolved to measurements on analogical models generated starting from the symbolic representation. the interaction between the symbolic and the analogical level is due to the fact that procedures are connected to some symbols, allowing generating, updating, and verifying the mental model. the hybrid model utilizes, for the symbolic component, a representation system based on the distinction between terminological and assertional knowledge. the terminological component adopts a SI-Net formalism, extended by temporal primitives. the assertional component is a subset of first-order logics. the analogical representation is a set of concurrent procedures modeling parts of the world, action processes, simulations, and metaphors based on force fields concepts. A particular case study, regarding the problem of the assembly of a complex object from parts, is taken as an experimental paradigm. © 1993 John Wiley Sons, Inc.