Statistical models for data that scale: multiscale domain summaries with applicati...
Signal processing techniques for beamforming and coding schemes in IoT communicati...
Grant number: | 11/13351-0 |
Support Opportunities: | Scholarships in Brazil - Doctorate |
Start date: | April 01, 2012 |
End date: | August 31, 2012 |
Field of knowledge: | Physical Sciences and Mathematics - Computer Science - Computational Mathematics |
Principal Investigator: | Hélio Pedrini |
Grantee: | Vitor de Lima |
Host Institution: | Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil |
Associated research grant: | 07/52015-0 - Approximation methods for visual computing, AP.TEM |
Abstract The transmission and storage of images and videos require techniques to reduce the needed amount of bits, even if it is necessary to degrade the quality of the data. The methods for representing data meeting such restrictions are called lossy encoders. This project proposes the study of compression methods based both on data interpolation and polynomial adjust to segments of the signal.The method based on interpolation discards rows and columns of each video frame, reconstructing them later by an interpolator. This approach introduces fewer artifacts compared to the visual patterns based on orthogonal transforms (e.g. H.264, MPEG-2 and JPEG-2000), but it is still not able to encode any video without the aid of another algorithm.The encoding based on the signal division into polynomial segments is capable of efficiently representing the geometrical structure of images, respecting the boundaries of the regions. The borders between the coded segments have severe visual artifacts and the generalization of this method for encoding video has not been explored yet.This project proposes the study of these two methods for creating two video encoders, one based on interpolation and another purely based on dividing the signal into polynomial segments.A mixed approach will represent each frame through a mesh, interpolating the data using methods similar to those of high order finite elements, splines, or NURBS. The motion compensation can be either disturbing the mesh structure or changing the control points.Finally, a proposal for texture encoding using wavelet transforms is presented, trying to avoid problems encountered in solutions based purely on a single orthogonal transform. | |
News published in Agência FAPESP Newsletter about the scholarship: | |
More itemsLess items | |
TITULO | |
Articles published in other media outlets ( ): | |
More itemsLess items | |
VEICULO: TITULO (DATA) | |
VEICULO: TITULO (DATA) | |