基于感兴趣深度的多视点图像压缩  

Multi-view image compressing based on depth of interest

作　　者：陈铖[1] 邓磊[1] 熊艳[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《信息技术》2014年第4期153-156,共4页Information Technology

摘　　要：多视点图像包含丰富的场景信息,但是相应的数据量也产生了急剧的扩张,给储存和传输带来巨大压力。提出了运用深度信息压缩图像的方法,该方法利用深度信息提取人眼感兴趣区域,再利用修改之后的最新编码标准HEVC(High Efficiency Video Coding)对图像进行编码,从而达到降低码率且符合人眼视觉需求的目的。实验证明,该方法相对HEVC原始方法,在同等比特率的前提下,能够提高感兴趣区域的峰值信噪比PSNR(Peak Signal to Noise Ratio),且有更好地主观视觉效果。Multi-view images contain rich scene information. However the dramatic expansion of data quantity brings enormous pressure to storage and transport. This paper uses depth map to encode image. This method extracts region of interest by depth standard HEVC （High Efficiency Video Coding information and then encodes images by the latest codec ）. It can meet the requirements of human eyes without much bitrate. The experimental results prove the method can increase the PSNR （Peak Signal to Noise Ratio） in the region of interest and the reconstructed image keep details better than the original method.

关 键 词：HEVC 图像编码 深度信息 

分 类 号：TN919.81[电子电信—通信与信息系统]