The two-dimensional discrete wavelet transform (DWT) can be applied in the heart of many image-processing algorithms.
Until recently, several studies have compared the performance of such transform on parallel architectures, for example, on graphics
processing units (GPUs). All these studies however considered only separable calculation schedules.
The two-dimensional discrete wavelet transform (DWT) can be applied in the heart of many image-processing algorithms.
Until recently, several studies have compared the performance of such transform on parallel architectures, for example, on graphics
processing units (GPUs). All these studies however considered only separable calculation schedules.
Lifting Scheme Cores for Wavelet TransformDavid Bařina
The thesis focuses on efficient computation of the two-dimensional discrete wavelet transform. The state-of-the-art methods are extended in several ways to perform the transform in a single loop, possibly in multi-scale fashion, using a compact streaming core. This core can further be appropriately reorganized to target the minimization of certain platform resources. The approach presented here nicely fits into common SIMD extensions, exploits the cache hierarchy of modern general-purpose processors, and is suitable for parallel evaluation. Finally, the approach presented is incorporated into the JPEG 2000 compression chain, in which it has proved to be fundamentally faster than widely used implementations.
Lifting Scheme Cores for Wavelet TransformDavid Bařina
The thesis focuses on efficient computation of the two-dimensional discrete wavelet transform. The state-of-the-art methods are extended in several ways to perform the transform in a single loop, possibly in multi-scale fashion, using a compact streaming core. This core can further be appropriately reorganized to target the minimization of certain platform resources. The approach presented here nicely fits into common SIMD extensions, exploits the cache hierarchy of modern general-purpose processors, and is suitable for parallel evaluation. Finally, the approach presented is incorporated into the JPEG 2000 compression chain, in which it has proved to be fundamentally faster than widely used implementations.
13. kódování koeficientů
blokový efekt
linearizace zig-zag, nulové více ke konci, EOB
DC zapsán rozdílově
AC pomocí RLE předcházejících nul
následně arit. nebo Huff. kódování (5–10 %)
Huffman: tabulky pro DC a AC koeficienty
14. kontejnery
JIF, JFIF, Exif, SPIFF
datový tok, big endian
rozdělen do segmentů, max. 65 535 B
segment uvozen markerem 0xff + 0x01 až 0xfe
markery: APP0 (hlavička JFIF), APP1 (Exif), SOF0 (začátek
obrázku v základním režimu), SOS (začátek komprimovaných
dat), RSTm (restartovací značky), DHT (definice Huffmanových
tabulek), DQT (definice kvantizačních tabulek)
JIF rozšířen na JFIF a Exif (nekompatibilní), JFIF+Exif
náhledy, metadata (model fotoaparátu)
Exif: TIFF (náhled JFIF)
SPIFF se neujal
15. zajímavosti
lapped DCT
dekomprese v 1/8..7/8 rozlišení
rozdílná kvantizace, možnost ROI
bezeztrátové transformace (jpegtran) – ořez, rotace, překlopení
postprocessingové filtry (deblocking)