|
RESEARCH |
RESEARCH INTERESTS |
Dealing with rendering from real imagery is generally called image-based rendering (IBR): The problem is the generation of novel images starting from a sample set which can be either a sparse set of images depicting a certain scene or a video-shot.
Taking the more liberal view of the term, IBR can be simplified as a black box process whose input is a set of real images and whose output is given by new images. IBR techniques can be classified in techniques for view synthesis and techniques for space-time data manipulation. The former focus on the generation of physically valid novel views as taken from a virtual camera placed in a different location with respect to all the acquired images while the latter have the main goal to obtain new images (that could not be necessarily physically valid, e.g., video inpainting) without changing the point of view.
After some keywords regarding my interests, some details about my research work follows...
Keywords: Video sequence analysis, Robust background initialization, Content-based representation, Image Based Modeling and Rendering, Uncalibrated view synthesis plane+parallax.
|
Robust Background InitializationBackground initialization is crucial for all those algorithms relying on background subtraction, a technique which detects changes from a model of the background scene. The background model has to be first initialized and then maintained and maintenance requires a correct initial background model. The presented algorithm can even deal with sequences where occluding objects persist in the same position for a considerable portion of time. As pixel-level heuristic are not sufficient in this case, we introduce spatial support. First the sequence is subdivided in patches that are clustered along the time-line in order to narrow down the number of background candidates. Then a tessellation is grown incrementally by selecting at each step the best continuation of the current background, according to the principles of visual grouping. The method rests on sound principles in all its stages, and only few, intelligible parameters are needed. More information? Click here. |
|
Content-based representationWe have developed a technique that produces a content-based representation of a video shot composed by a background (still) mosaic and one or more foreground moving objects. Segmentation of moving objects is based on ego-motion compensation and on background modelling using tools from robust statistics. Region matching is carried out by an algorithm that operates on the Mahalanobis distance between region descriptors in two subsequent frames and uses singular value decomposition to compute a set of correspondences satisfying both the principle of proximity and the principle of exclusion. The sequence is represented as a layered graph, and specific techniques are introduced to cope with crossing and occlusion. More information? Click here. |
|
Uncalibrated view synthesis plane+parallaxWe have proposed a generic framework for novel view synthesis from two uncalibrated reference views that allows to move a virtual camera along a curve that is obtained starting from the epipolar geometry of the reference views. The scene is described by its relative affine structure from which novel views are extrapolated and interpolated. The main contribution of this work is an automatic method for specifying virtual camera locations in an uncalibrated setting. More information? Click here. |
SEMINARS AND PRESENTATIONS |
|
September, 11th 2007: "Video Objects Segmentation by Robust Background Modeling" in the 14th International Conference on Image Analysis and Processing, Modena, Italy (Oral presentation). |
|
May, 9th 2006: "Robust Background Estimation with Stationary or Moving Cameras" at Department of Computer Science, University of Verona (Seminar). |
|
April, 21th 2006: "Image-Based Techniques for View Synthesis and Video Editing" at Department of Computer Science, University of Verona (Seminar). |
|
November, 30th - December, 1st 2005: "Continuous Parallax Adjustment for 3D-TV" in the 2nd IEE European Conference on Visual Media Production London, UK (Oral presentation). |
|
September, 11th-14th 2005: "Uncalibrated Interpolation of Rigid Displacements for View Synthesis" in the IEEE International Conference on Image Processing, "Magazzini del cotone", Genova, Italy (Poster presentation). |
|
November, 28th 2005: "Un metodo automatico per la calibrazione di
sistemi multi-telecamera"
at Department of Information Engineering, University of Padova (Seminar). |
|
September, 8th 2004: "Plenoptic Modelling and Analysis"
at Department of Electrical, Electronic & Computer Engineering,
Heriot-Watt University, Edinburgh (Seminar). |
|
September, 25th-26th 2003: "On-line registration of acoustic images" in the Secondo convegno della sezione italiana EUROGRAPHICS at Politecnico di Milano - Bovisa, Italy (Oral presentation). |
4.16 MB]
2.69 MB]