Improved 3D experience through depth-map image compression methodsM.Sc. Ionut Schiopu’s doctoral dissertation tackles the problem of depth-map image compression by providing more efﬁcient methods for the lossy and lossless compression of depth-map images.
Depth-map images used for view synthesis are compressed more efﬁciently and without any information loss by representing them in terms of regions and contours. In a depth-map image, the regions are defined through segmentation by way of grouping pixels that have similar properties and separating them using (region) contours. A depth-map image is encoded by the contours and the auxiliary information needed to reconstruct the depth values for each region. The contours are encoded either by describing them using two matrices of horizontal and vertical contour edges, or by describing them as a sequence of contour segments, where each such segment is deﬁned by an anchor (starting) point and a string of contour edges. The regions are reconstructed with a decoder using predictive coding or the piecewise constant model representation.
The doctoral dissertation offers new solutions for applications where some information loss is needed for achieving a high compression. The lossy compression was achieved using algorithms based on image segmentation and greedy slope optimization using region merging or splitting. The study includes algorithms with progressive coding functionalities, which allows the encoder to broadcast a bitstream for multiple decoders, each having different rate-distortion requirements. The best performance is achieved when the regions are reconstructed using planar model parameterization based on three heights, where the positions of the three heights are optimally selected.
Compared with the state-of-the-art coders, the algorithms have shown the best results for lossless compression with images with either high or low density of contour edges. For lossy compression, the results show signiﬁcant improvement compared to the state-of-the-art methods, and the algorithms offer important functionalities that improve the 3D image quality for a better 3D experience.
Public defence of a doctoral dissertation on Friday, 29 January
M.Sc. Ionut Schiopu’s doctoral dissertation in the ﬁeld of signal processing entitled ‘Depth-Map Image Compression Based on Region and Contour Modeling’ will be publicly examined at the Faculty of Computing and Electrical Engineering of Tampere University of Technology (TUT) in auditorium TB109 in the Tietotalo building (address: Korkeakoulunkatu 1, Tampere, Finland) at 12 noon on Friday, 29 January 2016.
The opponent will be Professor Søren Forchhammer (Technical University of Denmark, Department of Photonics Engineering, Denmark). Professor Ioan Tabus from the Department of Signal Processing of TUT will act as Chairman.
Ionut Schiopu (29) comes from Slatina, Romania, and works as a researcher at the Department of Signal Processing of TUT.
Further information: Ionut Schiopu, tel. +358 41 719 3289, ionut.schiopu<at>tut.ﬁ
The dissertation is available online at: http://URN.fi/URN:ISBN:978-952-15-3680-9