Research On Rate Control Of HEVC For High-resolution Aerial Video

The resolution of airborne optical sensor is much higher via the evolution of aerial survey technology. Greater demands to resolution and frame frequency make a challenging task for the storage and transmission of video data, which consequently restricts the application of unmanned aerial vehicles(UAV). On the basis of maintaining a certain level of reconstructed video quality, the limitations on transmission bandwidth and storage capacity gives rise to the requirment of impactful compressed coding for massive video data.High Efficiency Video Coding(HEVC) is a video compression standard developed by the JCT-VC organization, it offers about double the data compression ratio at the same level of video quality, or substantially improved video quality at the same bit rate. By virtue of rate control, to be a guarantee of encoder taking full advantage of determinatus memory space as well as transmission bandwidth, is a critical technology of video codec, it has been widely employed to obtain the optimal coding quality.Firstly, we introduce the principle and basic theory of rate control in video coding. After summarizing and analyzing the related works of HEVC rate control algorithm, achievements of our study are proposed combined with the demand of high resolution aerial video. The contributions of this work over previous approaches are summarized as follows:(1) Rate control plays an important role in video coding and transmission. The initial quantization parameter of the first Intra frame is vital for the rate control performance of an HEVC encoder. In order to improve the performance of HEVC rate control, we present a novel rate control initialization algorithm for HEVC rate control by video features. First, we analyze video features which have influence on the determination of initial quantization parameter, such as bpp(bit per pixel) and the video complexities. Then, we exploit to find a suitable initial quantization parameter efficiently and adaptively according to bpp and the video complexity. The video complexity is evaluated by gradient value and the sum of absolute transformed difference(SATD) for 8×8 pixel. Experimental results show that the proposed algorithm can increase the average PSNR(peak signal-to-noise ratio) of reconstructed video and the rate control error can be decreased.That is to say, we can get a high quality with smooth and accurate reconstructed video by the proposed method which is adaptive for various types of video sequences.(2) A new low-delay buffer status and gradient based rate control algorithm is proposed for airborne reconnaissance of the HEVC, to alleviate the buffer overflow or underflow and improve the poor video quality caused by the state-of-the-art HEVC rate control scheme. First, both the buffer state and the remainder of bits are considered in frame layer bit allocation. Then, allocate Largest Coding Unit(LCU)layer bits hierarchically with the reconnaissance video features evaluated by gradient value. Finally, encode LCU with Quantization Parameter(QP) derived from R-λmodel in LCU. Model parameter updating will be done after encoding. Experimental results show that, compared with the rate control scheme in HM16.0, the proposed algorithm reduces mismatch ratio to 2/5 and PSNR fluctuations within 1/4 under the8 Mbps bandwidth. Moreover, the proposed algorithm provides an improvement in subjective quality evaluation and has less bit fluctuation and lower, steadier buffer occupancy, which enables a lower delay of aerial photography applications.(3) In order to meet the application requirement of high-quality transmission of videos captured by UAV under low bandwidth, a novel rate control scheme based onregion-of-interest(ROI) is proposed, avoiding inaccurate bit allocation produced by camera movement. First, the ROIs information is sent to the encoder of the latest HEVC standard to generate a ROI map. Then, by using the ROI map, a new frame level bit allocation method is developed. In addition, different numbers of bits are dynamically allocated at LCU level. At last, by using a better robustness R-λ model,the QP for each LCU is calculated. The experimental results showed that the proposed rate control method can get a lower bitrate error and a higher quality for reconstructed video by choosing appropriate pixel weight on the HEVC platform.