clear all
clc
format long;
symbol = ['abcd'];
pr = [0.1 0.4 0.2 0.3];
seqin = ['cadacdb'];
codeword = arencode(symbol, pr, seqin)
seqout = ardecode(symbol, pr, codeword, 7)
function symseq = ardecode(symbol, pr, codeword, symlen)
%给定字符概率的算术编码
%输出：symse：字符串
%输入：symbol：由字符组成的行向量
% pr：字符出现的概率
% codeword：码字
% symlen：待解码字符串长度
format long
high_range = [];
for k = 1 : length(pr),
high_range = [high_range sum(pr(1 : k))];
end
low_range = [0 high_range(1 : length(pr) - 1)];
prmin = min(pr);
symseq = [];
symseq = [];
for i = 1 : symlen,
index = max(find(low_range <= codeword));
codeword = codeword - low_range(index);
%duo to numerical error, sometimes the encoded number
%will be slightly smaller than the current lower bound.
%If this happens, a correction is required.
if abs(codeword - pr(index)) < 0.01 * prmin,
index = index + 1;
codeword = 0;
end
symseq = [symseq symbol(index)];
codeword = codeword/pr(index);
if abs(codeword) < 0.01 * prmin,
i = symlen + 1; %break the for loop immediately
end
end
function arcode = arencode(symbol, pr, seqin)
%算术编码
%输出：码串
%输入：symbol：字符行向量
% pr：字符出现概率
% seqin：待编码字符串
high_range = [];
for k = 1: length(pr),
high_range = [high_range sum(pr(1: k))];
end
low_range = [0 high_range(1: length(pr) - 1)];
sbidx = zeros(size(seqin));
for i = 1: length(seqin),
sbidx(i) = find(symbol == seqin(i));
end
low = 0; high = 1;
for i = 1: length(seqin),
range = high - low;
high = low + range * high_range(sbidx(i));
low = low + range * low_range(sbidx(i));
end
arcode = low;
