Image Processing

Image Processing, Analysis and Understanding is a major field of specialization in Computer Vision that allows processing 2D images or pictures (isolated or possibly connected in motion or time-dependent sequences).

It is a form of signal processing for taking pictures or video frames and their low-level processing for extracting certain characteristics or parameters, or synthesizing new processed images as an output. In a broader sense, Computer Vision can be seen as a more generic form of image processing, closer to Image Understanding, in which the goal is to process, understand and decipher characteristics or track objects in an image or sequence of video images, with a clear goal in mind.
Vicomtech-IK4 is focusing its research on important aspects of Image Processing, such as:

  • Classifiers and other Pattern Recognition technologies allow classifying information according to predefined criteria. The information may vary in nature: visual, textual, acoustic, etc.
  • Object Tracking is used to follow the position of an object using Computer Vision or other techniques, to precisely determine its relative position and motion.
  • Camera/Multicamera Setups specifically designed for accurate Image Processing and Computer Vision applications, adjusting critical technical parameters.
  • Image Fusion and Registration provides the alignment and correlation of different images and image types to synthesize new combined images with enriched information.
  • Reconstruction is an important technology in Image Processing, including 2D and 3D reconstruction of objects, with the result being, for instance, the 3D reconstruction of scanned points data.
  • Segmentation provides a way to identify and isolate regions of interest or specific structures in images in order to focus the analysis on these areas, as for instance in medical imaging.
  • Feature Extraction allows reducing the complexity (dimensionality) of an image by transforming the input data into a set of features that display relevant characteristics of the image.
  • Visual 3D Model Fitting matches a 3D subset of acquired points in an object to its complete 3D model.
  • Simultaneous Location and Mapping (SLAM) matches a 3D subset of acquired points in an object to its complete 3D model.

Gallery

Outstanding projects