Abstract:
A method for encoding and decoding media signals, includes the operations of generating at a transmitting side multiple descriptions associated to data of the media signals through a downsampling operation (DS) performed on the data, and decoding at a receiving side the multiple descriptions for reconstructing the data by merging (US) the multiple descriptions. The operation of generating multiple descriptions further includes the operations of obtaining a spectral representation (S) of the data, comprising bands (LS, MS HS) associated to different ranges ((LR, MR HR), the bands (LS, MS HS) being obtained by a suitable quantization operation and comprising at least one highly quantized band (HS), that is subjected to a higher degree of quantization. A scrambling operation (SC, SC') is performed on the spectral representation (S) by moving the at least one highly quantized band (HR) to a different range (LR, MR), the scrambling operation being performed prior the downsampling operation (DS). In decoding, a descrambling operation (DS, DS') is performed before the merging (US) operation on the multiple descriptions. Preferred application is in coding of video signals by using polyphase downsampling (PDMD).
Abstract:
A method for encoding/decoding a video signal sequence by generating therefrom multiple description subsequences wherein said subsequences are produced by a plurality of parallel video encoding processes (102) based on respective encoding parameters. The method includes the step of commonly controlling (103) said encoding/decoding parameters for said plurality of video encoding/decoding processes (102).
Abstract:
A method for estimating motion in a sequence of bidimensional arrays of sampled values (e.g. video frames), includes the step of generating (104; 114) estimated motion vectors having at least one component along at least one reference axis (x, y). Motion estimation is performed with a given resolution (110, 112), and the method includes at least one of:
dynamically varying (110, 112) the resolution between subsequent estimation steps (114), and if said estimated motion vectors have two components along two reference axes (x, y), selectively varying (110, 112) the resolution of motion estimation independently along said two axes (x, y).
Preferred application is in optical mouse arrangements.