Comparison of entropy coders for lossless grayscale image compression

In this paper, a comparison between three lossless entropy coders, interpolative coding combined with FELICS, arithmetic coding, and ANS coding, in terms of grayscale image compression is presented. The encoders were compared on sequences of values, produced by applying the JPEG-LS prediction to the...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2023 International Conference on Data, Information and Computing Science (CDICS) s. 1 - 6
Hlavní autoři: Jeromel, Aljaz, Zalik, Borut
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 08.12.2023
Témata:
Arithmetic
arithmetic coding
asymmetric numeral systems
Data compression
Entropy
Gray-scale
Image coding
image compression
interpolative coding
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In this paper, a comparison between three lossless entropy coders, interpolative coding combined with FELICS, arithmetic coding, and ANS coding, in terms of grayscale image compression is presented. The encoders were compared on sequences of values, produced by applying the JPEG-LS prediction to the image, and calculating the differences between the actual values and predicted values. These values were then transformed to obtain positive numbers, which can be entropy coded by the three encoders. Experimental results show that the ANS coding is the most efficient of the three, closely followed by arithmetic coding, and interpolative coding combined with FELICS. The first two entropy coders, the ANS coder and arithmetic coder, also outperformed the PNG algorithm by a small margin.
DOI:10.1109/CDICS61497.2023.00011