Lines of Research

• Biometric analysis

  Biometrics is the study of automated methods for recognizing humans based onunique one or more intrinsic physical or behavioral traits.
  

  The "biometric information" is the application of mathematical and statistical techniques on physical or behavioral traits of an individual, to "verify" identities or to "identify" individuals.

   

• Image boundary analysis and applications

  Implementation of an Edge Detection Method with Sub-pixel Accuracy in a Graphic Environment and Comparative Study for Applications in 2D and 3D Images.

  
  

   

• Medical image analysis

  We developed a suite of techniques and processes to create images of the human body or parts thereof, for clinical purposes (medical procedures seeking to reveal, diagnose or examine disease) or medical science (including the study of normal anatomy and function).

   

• Partial differential equations in image processing

  The digital image processing is the set of techniques applied to digital images in order to improve quality or facilitate the search for information.

   

• Mathematical analysis in computer vision

  Computer vision is one of the major challenges in the field of signal process, we try to add improvements to these tasks.

   

• Optical flow

  The computation of the optical flow in a sequence of images is a basic topic in the field of computer vision. It is the basis for a wide range of applications, such as: stereoscopic vision and three-dimensional scene reconstruction, motion analysis of objects in image sequences, medical image analysis, robot guidance and collision detection, among others. It consists in the estimation of the displacement experienced by objects in a scene from an image sequence. Our research center has spent many years working on this subject and has specialized in variational techniques. Such techniques are especially useful for optical flow estimation. These are currently among the most precise techniques that are known.

  
  

   

• Terrain point cloud processing

  LIDAR technology, mounted on an airplane or helicopter, is capable of scanning millions of points over a territory. These points need to be filtered beforehand, to remove the noise captured during the acquisition process. After filtering the datasets, each point has an associated intensity value, and it is necessary to perform segmentation and classification of the entire cloud, distinguishing between vegetation, buildings, power lines, water, ground, etc.
  The Imaging Technology Center works in this research line within different fields: powerlines classification and anomalies detection, reconstruction of buildings in urban areas for later use in a 3D scene, displaying of visual data collected from sensors (solar radiation, shadows, etc.). In addition, we also work processing these datasets to obtain hierarchical structures that be efficient for a realistic display with good performance.

  
  

  

   

• Virtual Reality and Augmented Reality on GIS

  Virtual Reality (VR) is the technique by which we present completely immersive 3D experiences to the user through the use of binocular devices that cover the entire visual field. These headsets offer a stereoscopic view of the scenario, adding a sense of depth to the experience. In combination with odometric systems (inertial or visual), the system is able to calculate in real time the position of the user, thus allowing the recreation of his point of view.

  
  

  The Augmented Reality (AR) combines the synthetic elements generated by virtual reality with real elements that surround the user. In the case of augmented reality based on image combination, real images taken by a camera are combined in real time with the digital objects displayed from a point of view calculated by means of odometry. In this way, the user is able to interact with virtual objects in their environment.

  
  

  Our research center has developed several projects in which large 3D scenarios (many of them based on geographic information) are combined with these new visualization techniques. These developments have allowed us to explore geographic information, relegated to its visualization on maps, in the first person.

  
  

   

• Remote sensing

  Remote sensing is an active area of research in the CTIM (AMI division)  with several published publications, both in JCR Journals (IEEE-GRSL, IEEE-TGRS, IEEE-JSTARS, Remote Sensing, Philosophical Transactions) and in international conferences (IGARSS, EUSAR, IWOBI). The main topics pursued are: image denoising, image enhancement and image classification for SAR (synthetic aperture radar) and for PolSAR (polarimetric SAR). At present, there is also a funded research Project (MTM2016-75339-P (AEI/FEDER, UE) (Ministerio de Economía y Competitividad, Spain)), where mathematical analysis of SAR/PolSAR images is shared with mathematical analysis of medical images.
  In this research area participates also profesor Alejandro C. Frery, from Universidade Federal de Alagoas (Brazil).

  
  

  

  PolSAR image (San Joaquín Valley, California, USA, UAVSAR 18-11-2010)
  

   

• Stereoscopic and three-dimensional reconstruction of scenes

  Stereoscopic vision is the system that can capture the depth of a scene through a pair of related cameras. It has received much attention from the scientific community and has a special relevance, since it is equivalent to the human visual system. When we have two calibrated cameras that obtain images of a scene, then it is possible to know the 3D position of objects in space. It covers different topics such as: camera calibration, finding correspondences between images (similar to the problem of optical flow estimation) and three-dimensional reconstruction.