Methods of Entropy Coding

Kodiranje predstavlja transformaciju neke poruke ili općenito nekog objekta na način da se simboli (slova, brojevi, pikseli, ...) tog objekta zamjene simbolima neke druge abecede (npr. slova se zamjene brojevima). Metode entropijskog kodiranja vrše kodiranje bez gubitka, a temelje se izravno na teoriji informacije. Osim toga, neke od metoda postižu vrlo efikasan omjer kompresije što ih uz svojstvo kodiranja bez gubitka čini idealnim metodama kada se mora sačuvati originalan oblik sadržaja. U ovom radu izloženo je pet metoda entropijskog kodiranja: Shannon-Fanovo kodiranje, Huffmanovo kodiranje, aritmetičko kodiranje, metode rječnika i metode skraćivanja niza.Coding represents the transformation of some message or generally an object in the way that symbols (letters, numbers, pixels, ....) of that object being replaced with symbols of some other alphabet (for example, letters being replaced with numbers). Methods of entropy coding perform coding without loss, i.e. lossless compression. They are based directly on information theory. In addition, some of the methods achieve very efficient compression ratio which makes them ideal for coding when it is important to keep the original form of the content. In this thesis, five methods of entropy coding are presented: Shannon-Fano coding algorithm, Huffman coding algorithm, arithmetic coding algorithm, dictionary methods, and simple repetition suppression