首页|基于感兴趣区域及LCU特征的码率分配算法

基于感兴趣区域及LCU特征的码率分配算法

扫码查看
针对高效视频编码码率分配策略不考虑视觉冗余,并且LCU层图像内容特征考虑单一的情况下导致人眼观看视频时视觉感受下降的问题,提出了一种基于感兴趣区域及LCU特征的码率控制算法.通过预训练的目标检测网络检测视频中的人眼感兴趣区域;利用纹理复杂度、失真程度、结构强度的图像特征构建出新的权重计算公式调整QP.实验结果表明,在低延迟配置下保持客观质量的同时,可提高视频的主观质量,平均获得0.64%的SSIM增益.
Rate Allocation Algorithm Based on Region of Interest and LCU Feature
In case of bitrate allocation strategy in HEVC does not consider visual redundancy,and LCU layer image content features are single,visual perception is reduced when viewing videos by eyes.A rate control algorithm based on region of interest and LCU features is proposed in this paper.A pre-trained target detection network detects region of in-terest in the video.A new weight calculation formula is constructed using the image features of texture complexity,distortion degree,and structural strength to adjust QP.Experimental results show that the subjective quality of the video can be im-proved with an average SSIM gain of 0.64%while maintaining objective quality under low-latency configuration.

HEVCrate controldeep learningimage features

崔士明、黄新彭、林兴斌、滕国伟

展开 >

上海大学通信与信息工程学院,上海 200444

上海华平信息股份有限公司,上海 200438

HEVC 码率控制 深度学习 图像特征

2024

工业控制计算机
中国计算机学会工业控制计算机专业委员会 江苏省计算技术研究所有限责任公司

工业控制计算机

影响因子:0.258
ISSN:1001-182X
年,卷(期):2024.37(3)
  • 10