Wide-baseline stereo matching is a common problem of computer vision. By the explosion of smartphones equipped with camera modules, many classical computer vision solutions have been adapted to such platforms. Considering the widespread use of various networking options for mobile phones, one can consider a set of smart phones as an ad-hoc camera network, where each camera is equipped with a more and more powerful computing engine in addition to a limited bandwidth communication with other devices. Therefore the performance of classical vision algorithms in a collaborative mobile environment is of particular interest. In such a scenario we expect that the images are taken almost simultaneously but from different viewpoints, implying that the camera poses are significantly different but lighting conditions are the same. In this work, we provide quantitative comparison of the most important keypoint detectors and descriptors in the context of wide baseline stereo matching. We found that for resolution of 2 megapixels images the current mobile hardware is capable of providing results efficiently.

%B Proceedings of the International Symposium on Image and Signal Processing and Analysis (ISPA) %I IEEE %C Trieste %P 806 - 811 %8 Sep 2013 %G eng %9 Conference paper %R 10.1109/ISPA.2013.6703848 %0 Book Section %B Proceedings of International Symposium on Image and Signal Processing and Analysis (ISPA) %D 2013 %T On Topology Preservation in Triangular, Square, and Hexagonal Grids %A Péter Kardos %A Kálmán Palágyi %E Giovanni Ramponi %E Sven Lončarić %E Alberto Carini %E Karen Egiazarian %X

There are three possible partitionings of the continuous plane into regular polygons that leads to triangular, square, and hexagonal grids. The topology of the square grid is fairly well-understood, but it cannot be said of the remaining two regular sampling schemes. This paper presents a general characterization of simple pixels and some simplified sufficient conditions for topology-preserving operators in all the three types of regular grids.

%B Proceedings of International Symposium on Image and Signal Processing and Analysis (ISPA) %I IEEE %C Trieste %P 782 - 787 %8 Sep 2013 %G eng %9 Conference paper %M 14027933 %R 10.1109/ISPA.2013.6703844 %0 Conference Paper %B Proceedings of International Symposium on Image and Signal Processing and Analysis (ISPA) %D 2013 %T A uniqueness result for reconstructing hv-convex polyominoes from horizontal and vertical projections and morphological skeleton %A Norbert Hantos %A Péter Balázs %E Giovanni Ramponi %E Sven Lončarić %E Alberto Carini %E Karen Egiazarian %XIn this article we study the uniqueness of the reconstruction in a special class of 4-connected hv-convex images, using two projections and the so-called morphological skeleton. Generally, if just the two projections are given, there can be exponentially many hv-convex 4-connected images satisfying them. Knowing the morphological skeleton in addition, we can reduce the number of solutions. In the studied class, the images are defined by two parameters. We show that the uniqueness of their reconstruction depends only on the values of those parameters.

%B Proceedings of International Symposium on Image and Signal Processing and Analysis (ISPA) %I IEEE %C Trieste %P 788 - 793 %8 Sep 2013 %G eng %9 Conference paper %M 14027951 %R 10.1109/ISPA.2013.6703845 %0 Conference Paper %B Proceedings of the International Symposium on Image and Signal Processing and Analysis (ISPA) %D 2011 %T 2D Parallel Thinning Algorithms Based on Isthmus-Preservation %A Gábor Németh %A Kálmán Palágyi %E Sven Lončarić %E Giovanni Ramponi %E D. Sersic %X

Skeletons are widely used shape descriptors which summarize the general form of binary objects. A technique to obtain skeletons is the thinning, that is an iterative layer-by-layer erosion in a topology-preserving way. Conventional thinning algorithms preserve line endpoints to provide important geometric information relative to the object to be represented. Bertrand and Couprie proposed an alternative strategy by accumulating isthmus points that are line interior points. In this paper we present six new 2D parallel thinning algorithms that are derived from some sufficient conditions for topology preserving reductions and based on isthmus-preservation.

%B Proceedings of the International Symposium on Image and Signal Processing and Analysis (ISPA) %I IEEE %C Dubrovnik, Croatia %P 585 - 590 %8 Sep 2011 %@ 978-1-4577-0841-1 %G eng %9 Conference paper %M 12307467 %0 Book Section %B Proceedings of the 6th International Symposium on Image and Signal Processing and Analysis %D 2009 %T Object subsampling strategies to improve computational performance %A András Hajdú %A Péter Veres %A Attila Tanacs %A Rorland Harangozó %E P Zinterhof %E Sven Lončarić %E A Uhl %E Alberto Carini %XWe investigate object simplification methods based on Centroidal Voronoi Tesselation (CVT) that share the possibility of great speed-ups in various scenarios. We propose Constrained CVT to sample points from the object boundary and the Region-based CVT to be able to subsample lower dimensional objects, as well. Moreover we introduce custom weight functions based on object properties. Thus, wecan be more specific on what are the important parts of the subsampled object. We also list several novel applications corresponding to the theoretical achivements presented. The advantages of applying the subsampling strategies are presented for registration, human detection, and the segmentation of the retinal vascular system, respectively. Quantitative results are shown to check the deterioration of the accuracy with the level of subsampling, and the computational gain. We also make comparisons with other naive (e.g. random) subsampling methods.

%B Proceedings of the 6th International Symposium on Image and Signal Processing and Analysis %I IEEE Press %C Salzburg %P 448 - 453 %8 Sep 2009 %G eng %9 Conference paper %0 Book Section %B Proccedings of the 5th International Symposium on Image and Signal Processing and Analysis %D 2007 %T Reconstructing some hv-convex binary images from three or four projections %A Péter Balázs %E M Petrou %E T Saramaki %E Aytul Ercil %E Sven Lončarić %XThe reconstruction of binary images from their projections is animportant problem in discrete tomography. The main challenge in this task is that in certain cases the projections do not uniquely determine the binary image. This can yield an extremely large number of (sometimes very different) solutions. Moreover, under certain circumstances the reconstruction becomes NP-hard. A commonly used technique to reduce ambiguity and to avoid intractability is to suppose that the image to be reconstructed arises from a certain class of images having some geometrical properties. This paper studies the reconstruction problem in the class of hv-convex images having their components in so-called decomposable configurations. First, we give a negative result showing that there can be exponentially many images of the above class having the same three projections. Then, we present a heuristic that uses four projections to reconstruct an hv-convex image with decomposable configuration. We also analyze the performance of our heuristic from the viewpoints of accuracy and running time.

%B Proccedings of the 5th International Symposium on Image and Signal Processing and Analysis %I IEEE %C Istanbul, Turkey %P 136 - 140 %8 Sep 2007 %@ 978-953-184-116-0 %G eng %9 Conference paper %R 10.1109/ISPA.2007.4383678 %0 Book Section %B 3D Image Processing %D 2002 %T Virtual Dissection of the Colon %A Erich Sorantin %A Emese Balogh %A Anna Vilanova Bartroli %A Kálmán Palágyi %A László Gábor Nyúl %A Sven Lončarić %A Marco Subasic %A Domagoj Kovacevic %E Davide Caramella %E Carlo Bartolozzi %B 3D Image Processing %I Springer-Verlag %C New York %P 197 - 209 %8 2002/// %G eng %0 Book Section %B Proceedings of the 2nd International Symposium on Image and Signal Processing and Analysis, ISPA 2001, Pula, Croatia %D 2001 %T Virtual Dissection of the Colon Based on Helical CT Data - Can It Be Done? %A Erich Sorantin %A Emese Balogh %A Anna Vilanova Bartroli %A Kálmán Palágyi %A László Gábor Nyúl %E Sven Lončarić %E Hrvoje Babic %X Colorectal cancer is the third most commonly diagnosed cancer;and colonic polyps are known precursors of that particular cancer. Virtual dissection refers to a display technique for polyp detection based on helical CT data, where the colon is dissected and flattened as on the pathologist's table. The approach and image processing as well as the early experience are described in this paper. %B Proceedings of the 2nd International Symposium on Image and Signal Processing and Analysis, ISPA 2001, Pula, Croatia %I University of Zagreb %C Zagreb %P 224 - 229 %8 2001/// %G eng