The High Efficiency Video Coding (H.265/HEVC) has better coding efficiency, but the encoding performance must be improved to meet the growing multimedia applications. This paper deals with the entropy encoding methods and algorithms in video encoder H.265/HEVC. Based on an analysis of their advantages and disadvantages, a method called the entropy coding algorithm using the enumerative coding of the hierarchical approach is proposed. The proposed algorithm consists of the Context-Adaptive Binary Arithmetic Coding (CABAC) algorithm and the enumerative coding algorithm with a hierarchical approach. The proposed algorithm is tested in the Visual C++ development environment on various test video sequences. The results of the experiments shows a greater efficiency of coding of multimedia data. Proposed method reduces up to 38.13% of the storage volume compared to the traditional CABAC method in HEVC. However, this method requires a longer coding time. The proposed method can be recommended for use in telecommunication systems for storage, transmission and processing of multimedia data, where a high compression ratio firstly required.
Entropy coding is performed at the last stage of video encoding (and first
stage of video decoding), after the video signal has been reduced to a series of
syntax elements. The entropy encoder converts binary sequences representing
elements of a video sequence into a compressed bit stream that can be stored
in a file or transmitted over communication networks [
An adaptive binary arithmetic coder with low memory consumption is
proposed in [
HEVC uses CABAC as its single entropy coding method. HEVC CABAC was redesigned to offer higher throughput than its AVC predecessor, while still maintaining a higher compression ratio. This paper will focus on HEVC CABAC, and his aim is to propose a new method with higher efficiency compared to the traditional one. 2
The CABAC method is based on arithmetic coding with several changes to
adapt it to the needs of video encoding standards [
At the first stage, this non-binary syntactic element is uniquely mapped into a binary sequence. When a binary element is given, this step is skipped. Depending on the encoding mode, one or two consecutive steps may follow for each element.
In the basic coding mode, the binary element enters the context modeling stage, where the probabilistic model is chosen so that the corresponding choice depends on the coded elements. Then, after assigning the model value, the binary element with its model goes to the basic coding mode, where the final stage of arithmetic coding with the subsequent modification of the model (for example, if the value was "1", the unit counter increases).
Alternatively, encoding with a probability of 0.5 is selected for the selected binary elements using a simplified coding mechanism (fig. 1). In H.264, an arithmetic coder is a set of procedures of low complexity in which there are no multiplication operations. Procedures include shifts and referrals to tables. Using CABAC allows you to reduce the bitrate by an average of 10-15%. The greatest gain is usually obtained when processing interlaced TV signals. 3
After analyzing the advantages and disadvantages of existing methods, a new one was proposed. The general scheme of the proposed algorithm is shown in fig. 2.
The input data of the entropy encoder is the parameters of the temporal model (motion vectors) and the spatial model (conversion coefficients), as well as markers (codes denoting synchronization points in the video sequence) and headers (macroblock, image, sequence and other object headers). After the execution of the CABAC algorithm, the input data is converted into binary sequences that are input to the enumerative coding algorithm with a hierarchical approach. After it, the final output of the entropy encoder is obtained, or, in other words, the compressed data of the video codec.
The enumerative coding algorithm with using hierarchies [
Let f0; 1gn - the vector n of binary numbers and x = (x1; x2; : : : ; xn) - the elements of this vector, S - the set of vectors n of binary numbers. Define ns the number of elements in and denote ns(x1; x2; : : : ; xn) the number of elements in S, the first k elements were defined. We suggest the vector w has elements 1 or Pn
j=1 xj = w.
Coding by enumerative coding method using hierarchical approach includes the following steps: 1. Construct a sum tree (fig. 3). 2. Encode wpmax;1, using [log2 (N + 1)] bits. 3. Produce cycle for p from pmax 1 down to 1 with a step 1: – Produce cycle for i from 1 up to [ QpN ],
j=1 nj – Encode vector {wp;(i 1)np+1; : : : ; wp;inp } by enumerative encoding method of bounded integers. 4. Produce cycle for i from 1 up to [ nN1 ], – Encode {wp;(i 1)np+1; : : : ; wp;inp } by Lynch-Davison method to obtain an output sequence
The decoding process by enumerative coding method using hierarchical approach is performed in the same order as the encoding process. 3.1
The Lynch-Davisson method
The coding process consists in calculating the lexicographic index (the usual dictionary ordering in interpreting 0 < 1) of the vector x 2 S f0; 1; 2; : : : ; M gn is determined by the following formula:
n xj 1 is(x) = X X xj ns(x1; x2; : : : ; xn)
The approach proposed above was implemented in the C++ development environment and tested with video sequences of different information character and different dimensions.
In the experiment we will compare the proposed algorithm with the algorithm CABAC, which are used in HM 16.0 (H.265/HEVC Test Model) with configuration “intra main profile” by the following criteria: the quality of the encoding (by the volume of the output stream) and the encoding time (here, QP is the quantization parameter). The experiment is carried on Window 10 OS platform with Intel(R) Core(TM) i5-4210U CPU @1.70GHz 2.40 GHz and 6GB RAM.
The quality of the encoding is estimated by comparing the volume of the output stream (shown in Table 1). The experimental results showed that the proposed algorithm reduces the output file up to 38.13%.
The coding time is estimated by comparing the average time required to encode each video stream shown in Table 2. It can be seen that the proposed algorithm increases the coding time by an average of 15.4% compared to the CABAC H.265 / HEVC method.
Thus, it can be argued that the proposed method provides a greater compression ratio (up to 38.13%), while it requires more time to work compared to other algorithms.
In this paper, we propose to apply the entropy encoding algorithm to improve the efficiency of multimedia data compression. The experimental results show that proposed method gives a good compression ratio (the output file size is reduced up to 38.13%), but the encoding procedure time is increased (approximately 15.4%) compared to the traditional CABAC method in the HEVC standard.
Fig. 4. Decoding process by numbering encoding of bounded integers