Prediction In HEVC Research Articles

DECISION ALGORITHM FOR INTRA PREDICTION IN HIGH-EFFICIENCY VIDEO CODING (HEVC)

Prediction in HEVC exploits the redundant information in the fame to improve compression efficiency. The computational complexity of prediction is comparatively high as it recursively calculates the depth by comparing the rate-distortion optimization cost (RDO) exhaustively. The deep learning technology has shown a good mark in this area compared to traditional signal processing because of its content-based analysis and learning ability. This paper proposes a deep depth decision algorithm to predict the depth of the coding tree unit (CTU) and store it as a 16-element vector, and this model is pipelined to the HEVC encoder to compare the time taken and bit rate of encoding. The comparison chart clearly shows the reduction in computational time and enhancement in bitrate while encoding. The dataset used here is generated for the model with 110000 frames of the various resolutions, split into test, training, and validation, and trained on a depth decision model. The trained model interfaced with the HEVC encoder is compared with the normal encoder. The evaluation is done for quality check for the proposed model with BD-PSNR and BD-Bitrate shows a dip of 0.6 in BD-PSNR and increment of 6.7 in BD-Bitrate. When pipelined with the original HEVC, the RDO cost shows an improvement over existing techniques. The average encoding time is reduced by about 72% by the pipelined deep depth decision algorithm that points to the reduction in computational complexity. An average time saving of 88.49% is achieved with a deep depth decision algorithm-based encoder compared to the existing techniques.

Multi-stage all-zero block detection for HEVC coding using machine learning

Compared with deadzone hard-decision quantization (HDQ), rate-distortion optimized quantization (RDOQ) in HEVC brings non-negligible coding gain, however consumes considerable computations caused by exhaustive search over multiple candidates to determine optimal output level. Benefiting from efficient prediction in HEVC, transform blocks are frequently quantized to all zero, especially in small-size blocks. It is worthwhile to detect all zero block (AZB) for transform blocks to bypass subsequent computation-intensive RDOQ. Traditional thresholding based AZB detection algorithms are well-suited for deadzone quantized blocks, however miss partial optimal results in RDOQ and suffer from more or less accuracy degradation in RDOQ. This paper proposes a novel multi-stage AZB detection algorithm for RDOQ blocks with good tradeoff between complexity and accuracy. At the first stage, genuine all zero blocks (G_AZB) which are quantized to all zero both in HDQ and RDOQ are prejudged by comparison with conservative threshold determined by mathematical derivation for deadzone HDQ. At the second stage, an adaptive threshold model is built using adaptive deadzone offset by simulating the behavior patterns existing in RDOQ, aiming to further detect the pseudo AZB (P_AZB) which are quantized to all zero in RDOQ however not all zero in HDQ. At the final stage, machine learning based detection is proposed to classify the remaining “cunning” all zero blocks using eight distinguished RDO-related features, by which subtle working mechanism in RDOQ is leveraged. The experimental results demonstrate that the proposed algorithm achieves up to 7.471% total coding computation saving with 0.064% BD-RATE increment compared with RDOQ on average. Moreover, the average FNR and FPR detection accuracies are 6.3% and 6.5% respectively.

A Fast CU Size Decision Optimal Algorithm Based on Neighborhood Prediction for HEVC

A Fast CU Size Decision Optimal Algorithm Based on Neighborhood Prediction for HEVC

HEVC의 고속 화면내 예측 모드 결정 기법

HEVC의 화면내 예측 부호화는 예측 모드를 기존 H.264/AVC에 비해서 최대 35개까지 확장함으로써 향상된 부호화 효율을 갖는다. 반면 화면내 부호화의 복잡도 또한 크게 증가하여 복잡도 감소를 위한 고속 부호화 기법이 요구된다. 본 논문에서는 HEVC의 참조모델인 HM에 화면내 고속 부호화를 위하여 채택된 RMD(Rough Mode Decision)에 추가적으로 복잡도 감소를 위하여 고속 화면내 예측 모드 결정 기법을 제시한다. 본 기법은 후보 모드를 결정하기 위한 RMD 과정에서의 탐색 모드 수와 선택된 후보 모드로부터 최종 모드를 결정하기 위한 탐색 모드 수를 제한하는 방법을 결합하여 모드 결정을 위한 복잡도를 감소한다. 본 제안 기법은 실험결과 HM 12.0에서 다양한 테스트시퀀스에서 HM 대비 1.0%의 비트 증가로 13.2%의 복잡도를 감소시킬 수 있었다. Intra prediction in HEVC is also significantly improved from H.264/AVC with the increased prediction modes up to 35 that results in increase of the complexity cost. Hence, a fast intra mode decision algorithm is required for real-time applications. A fast intra mode decision is proposed in this paper for further complexity reduction in addition to the RMD (Rough Mode Decision) that has been adopted into the HEVC reference software, referred to as HM, for fast intra prediction. The proposed method reduces the complexity of intra mode decision by limiting the number of search modes in both steps of RMD and the final mode decision. Experimental results show that the proposed method provide about 13.2% encoding time reduction with 1.0% BD-rate increase on average over test sequences in HM 12.0.

Texture Direction Based Optimization for Intra Prediction in HEVC

Texture Direction Based Optimization for Intra Prediction in HEVC