A fast HEVC intra-coding algorithm based on texture homogeneity and spatio-temporal correlation

Abstract

The high efficiency video coding (HEVC) standard supports a flexible coding tree unit (CTU) partitioning structure, and thus coding efficiency is improved significantly. However, the use of this technique inevitably results in greatly increased encoding complexity. In order to reduce the complexity of intra-coding, we propose a hybrid scheme consisting of fast coding unit (CU) size decision and fast prediction unit (PU) mode decision processes. An adaptive method is utilised to measure the homogeneity of video content thus avoiding unnecessary rate distortion (RD) evaluations. The depth range to be tested is narrowed based on the partitioning parameters of the spatially adjacent CUs and the temporally co-located CU. Furthermore, the mode correlation between neighbouring frames and between adjacent coding levels in the same quad-tree structure is taken into account to predict the most probable directional mode. The number of candidate PU modes is further decreased according to the Hadamard cost. Experimental results illustrate that our scheme achieves a significant reduction in computational complexity of HEVC intra-coding. Compared with the HM encoder, the encoding time is reduced by up to 71% with negligible degradation in coding efficiency.

Fast algorithm for HEVC intra‐coding implemented by preprocessing

“…It can be seen from Table 8 that compared with [24] and [27], this study has great advantages in rate penalty and coding time reduction. The authors of [23,26,28] have obtained a larger coding time reduction than this study, but at the cost of a larger rate increase. The authors of [5,16,25,26] and this study have obtained a better trade-off on the reduction of coding time and rate loss.…”

“…The authors of [23,26,28] have obtained a larger coding time reduction than this study, but at the cost of a larger rate increase. The authors of [5,16,25,26] and this study have obtained a better trade-off on the reduction of coding time and rate loss. However, in calculating the AC coefficient, the square calculation and more multiplication and division operations are used in [26], which increases the difficulty of hardware implementation.…”

“…The authors of [5,16,25,26] and this study have obtained a better trade-off on the reduction of coding time and rate loss. However, in calculating the AC coefficient, the square calculation and more multiplication and division operations are used in [26], which increases the difficulty of hardware implementation. In [5,16], a large amount of adjacent coded block information is used in the fast CU size decision, which increases the dependence of the data and is not conducive to the parallelism of the hardware preprocess algorithm, and the reduction of the computational complexity of [16] is low.…”

“…Tseng et al [5] 1.00 25.0 Liu et al [14] 0.60 29.5 Abdelrasoul et al [15] 1.75 40.8 Jamali et al [19] gradient analysis 0.62 11.4 Jamali et al [19] RDO cost modelling 0.81 29.5 Lu et al [26] 1.51 39.0 Yao et al [27] 1.86 36.3 Benhajyoussef et al [25] 0.90 31.8 proposed 0.47 20.1…”

Iterative training of neural networks for intra prediction

“…Instead, one can consider a class C of pairs of a block and its context defined by a known context-block relationship, then train a neural network on samples from C, hence limiting the blurriness of the neural network predictions for C. For example, in [12], C gathers the blocks provided by the H.265 partitioning [13] of images, each paired with its context. The context-block relationship is that a block of C is relatively well predicted from its reference samples by the H.265 intra prediction [14], see Figures 1 and 2, because the H.265 intra prediction drives the partitioning [15]. But, this has two drawbacks.…”