Algorithmic and architectural design for real-time and power-efficient Retinex image/video processing

This paper presents novel algorithmic and architectural solutions for real-time and power-efficient enhancement of images and video sequences. A programmable class of Retinex-like filters, based on the separation of the illumination and reflectance components, is proposed. The dynamic range of the input image is controlled by applying a suitable non-linear function to the illumination, while the details are enhanced by processing the reflectance. An innovative spatially recursive rational filter is used to estimate the illumination. Moreover, to improve the visual quality results of two-branch Retinex operators when applied to videos, a novel three-branch technique is proposed which exploits both spatial and temporal filtering. Real-time implementation is obtained by designing an Application Specific Instruction-set Processor (ASIP). Optimizations are addressed at algorithmic and architectural levels. The former involves arithmetic accuracy definition and linearization of non-linear operators; the latter includes customized instruction set, dedicated memory structure, adapted pipeline, bypasses, custom address generator, and special looping structures. The ASIP is synthesized in standard-cells CMOS technology and its performances are compared to known Digital signal processor (DSP) implementations of real-time Retinex filters. As a result of the comparison, the proposed algorithmic/architectural design outperforms state-of-art Retinex-like operators achieving the best trade-off between power consumption, flexibility, and visual quality.

Determination of the Fe content of embedded Cu-rich particles in ferritic alloys using energy-filtered TEM

A novel technique for the quantification of the iron content of copper precipitates in ferritic steels is presented. Energy-filtered (EF) imaging has been used to extract elemental maps with high spatial resolution. These maps contain enough information to attempt the quantification of the signal produced by the precipitates when either a line profile is measured across them or the whole image signal is integrated. Assumptions such as sphericity of the precipitates and composition variations are discussed. Special attention to the assessment of drift on the information extracted from EF images has been taken. Minimum detectability and optimum acquisition conditions are discussed.     

The Social Sustainable Aquifer Yield: An Indicator for the Analysis and Assessment of the Integrated Aquifers Management

Integrated Aquifers Management (IAM) demands innovative tools and methods that are able to consider as much perspectives as possible. This research is aimed to design, apply and provide an indicator named Social Sustainable Aquifer Yield (SSAY), expressed in units of time that includes pure hydrological variables as well as social ones. The indicator is defined as the relation between the Residence Time, which is the relation between aquifer Storage (S) and Recharge (R) (S/R), and the relation between the aquifer Pumping (P) and the new variable named Aquifer Social Yield (ASY). The whole indicator is defined by this formula: (S/R)/(P/ASY). The assessment of the residence time is essential in aquifers with at least one of the following features: i) high hydraulic diffusivity, and ii) small volume of reserves. Finally, the variable ASY is defined as the average perception from the stakeholders about the maximum acceptable aquifer exploitation. This indicator has been successfully applied in the aquifers located in southern Jaen province (South Spain) belonging to the Water System SE4-Jaén Water Supply. The results probe the high utility of the indicator, especially in regards to the public participation processes.     

GPGPU implementation of growing neural gas: Application to 3D scene reconstruction

Self-organising neural models have the ability to provide a good representation of the input space. In particular the Growing Neural Gas (GNG) is a suitable model because of its flexibility, rapid adaptation and excellent quality of representation. However, this type of learning is time-consuming, especially for high-dimensional input data. Since real applications often work under time constraints, it is necessary to adapt the learning process in order to complete it in a predefined time. This paper proposes a Graphics Processing Unit (GPU) parallel implementation of the GNG with Compute Unified Device Architecture (CUDA). In contrast to existing algorithms, the proposed GPU implementation allows the acceleration of the learning process keeping a good quality of representation. Comparative experiments using iterative, parallel and hybrid implementations are carried out to demonstrate the effectiveness of CUDA implementation. The results show that GNG learning with the proposed implementation achieves a speed-up of 6×$6\times$ compared with the single-threaded CPU implementation. GPU implementation has also been applied to a real application with time constraints: acceleration of 3D scene reconstruction for egomotion, in order to validate the proposal.     

Generalized Bayesian inference in a fuzzy context: From theory to a virtual reality application

A generalized Bayesian inference framework in order to embed fuzzy sets and partial probabilistic information is provided. The general framework of reference is that of coherent conditional probabilities, which allows giving a rigorous interpretation of membership function as a conditional probability, regarded as a function of the conditioning event. The inferential problem needs to be studied in situations where the prior can be partial; moreover, membership and prior can be given on different classes of events. This inferential model is applied for the virtual representation of a female avatar.     

A platform for the development of patient applications in the domain of personalized health

Personalized health (p-health) systems can contribute significantly to the sustainability of healthcare systems, though their feasibility is yet to be proven. One of the problems related to their development is the lack of well-established development tools for this domain. As the p-health paradigm is focused on patient self-management, big challenges arise around the design and implementation of patient systems. This paper presents a reference platform created for the development of these applications, and shows the advantages of its adoption in a complex project dealing with cardio-vascular diseases.     

Prediction of parking space availability in real time

Intelligent parking reservation (IPR) systems allow customers to select a parking facility according to their preferences, rapidly park their vehicle without searching for a free stall, and pay their reservation in advance avoiding queues. Some IPR systems interact with in-vehicle navigation systems and provide users with information in real time such as capacity, parking fee, and current parking utilization. However, few of these systems provide information on the forecast utilization at specific hours - a process that requires the study of the competition between parking alternatives for the market share. This paper proposes a methodology for predicting real-time parking space availability in IPR architectures. This methodology consists of three subroutines to allocate simulated parking requests, estimate future departures, and forecast parking availability. Parking requests are allocated iteratively using an aggregated approach as a function of simulated drivers’ preferences, and parking availability. This approach is based on a calibrated discrete choice model for selecting parking alternatives. A numerical comparison between a one-by-one simulation-based forecast and the proposed aggregated approach indicates that no significant discrepancies exists, validating and suggesting the use of the less time consuming proposed aggregated methodology. Results obtained from contrasting predictions with real data yielded small average error availabilities. The forecast improves as the system registers arrivals and departures. Thus, the forecast is adequate for potential distribution in real-time using different media such as Internet, navigation systems, cell phones or GIS.     

Instrumental Learning Within the Rat Spinal Cord: Localization of the Essential Neural Circuit.

Following spinal transection of the upper thoracic spinal cord, male Sprague-Dawley rats given legshock whenever a hindlimb is extended learn to maintain the leg in a flexed position. The region of the cord that mediates this instrumental learning was isolated using neuroanatomical tracing, localized infusion of lidocaine, and surgical transections. DiI and Fluoro-Gold microinjection at the site of shock application labeled motor neuron bodies of lamina IX in the lower lumbar region. Local application of the Na++ channel blocker lidocaine disrupted learning when it was applied over a region extending from the lower lumbar (L3) to upper sacral (S2) cord. The drug had no effect rostral or caudal to this region. Surgical transections as low as L4 had no effect on learning. Learning also survived a dual transection at L4 and S3, but not L4 and S2. The results suggest that the essential neural circuit lies between L4 and S3. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Synthesis of MCM-41 frameworks incorporating oxygen-sulphur derivates of molybdenum and tungsten

The first examples of mesostructured materials containing Mo- and W-oxo-sulphur species incorporated into a poorly ordered MCM-41 framework have been prepared under hydrothermal condition and alkaline medium. The incorporation of oxygen-sulphur derivates of molybdenum and tungsten not only increases significantly the pore diameter, but also improves the thermal stability of the MCM-41-related mesoporous material.