Increasing the standard of a raster-scanned video signal, and interlace-to-sequential conversion

摘要

An interlace-to-sequential video standards converter receives an interlaced raster-scanned input signal at an input standard. A deinterlacer (14) inserts zero values into the signal so as to increase the standard to a required output standard. The output standard signal is formed into data blocks which in turn are subjected to a Fourier transformation (22) to produce a transformed signal containing a plurality of components each having a value. The components of the transformed signal are assigned (30) as signal or alias components, the assignment being made in dependence upon the component values. The component values which are assigned as signal components are then subjected (38) to an inverse Fourier transformation to provide at least the additional signal values required to increase the input signal to the output standard. Other types of up-conversion are possible, such as increasing the line standard of an existing sequential signal. <??>The assignment operation proceeds by a cyclic process of growing signal and alias regions. As the assigner (30) assigns a component as 'signal', it assigns that component's alias as 'alias'. Initially the assigner assigns the "zero frequency" component (or a region around it) to 'signal' and the alias of that component (or region) to 'alias'. The 'signal' region is grown by considering all unassigned neighbour components and adding the one with greater magnitude. Its alias is added to the 'alias' region. Any component which is its own alias is assigned to a special 'border' class. When all samples have been assigned, those in the 'alias' region are set to zero, those in the 'border' class are attenuated by a factor of two and those in the 'signal' region are not modified. <IMAGE>