kdu_synthesis( kdu_resolution resolution, kdu_sample_allocator * allocator, bool use_shorts, float normalization=1.0F)
Java: Kdu_synthesis( Kdu_resolution resolution, Kdu_sample_allocator allocator, boolean use_shorts, float normalization)
[Declared in "../coresys/common/kdu_sample_processing.h"]
Constructing an instance of this class for the highest visible resolution level of a tile-component, will cause the constructor to recursively create instances of the class for each successive stage of DWT synthesis and also for the block decoding process.
Interface to the resolution level for which the synthesis object is being created. The synthesis stage reconstructs the image at that resolution from LL, LH, HL and HH subbands, where LH, HL and HH subbands are the resolution object's member subbands, as identified by resolution.access_subband and the LL subband is synthesized by the next lower DWT stage, if there is one, as identified by resolution.access_next.
A kdu_sample_allocator object whose finalize member function has not yet been called must be supplied for pre-allocation of the various sample buffering arrays. This same allocator will be shared by the entire DWT tree and by the kdu_decoder objects at its leaves.
Indicates whether 16-bit or 32-bit data representations are to be used. The same type of representation must be used throughput the DWT processing chain and line buffers pulled from the DWT synthesis engine must use this representation.
Ignored for reversibly transformed data. In the irreversible case, it indicates that the nominal range of data recovered from the kdu_pull_ifc::pull function will be from -0.5*R to 0.5*R, where R is the value of the normalization argument. This capability is provided primarily to allow normalization steps to be skipped or approximated with simple powers of 2 during lifting implementations of the DWT; the factors can be folded into quantization step sizes. The best way to use the normalization argument will generally depend upon the implementation of the DWT.