CAJ | 학술논문

与传统的二维图像不同，三维视频系统中的深度图像具有明显的稀疏性特征，并且其边缘信息对于合成虚拟视点的质量非常重要。为此，本文提出一种基于压缩感知的自适应深度图像编码方法。该方法首先提取深度图像的边缘，然后根据边缘在各个图像块中所占的比例，自适应地选取不同的采样率进行压缩，从而能够更好地重构深度图像。实验结果表明：与采用均匀采样率的方法相比，所提出的方法在同样的采样率下，合成的虚拟视点图像具有更高的PSNR值和更好的主观视觉质量。
여전통적이유도상불동，삼유시빈계통중적심도도상구유명현적희소성특정，병차기변연신식대우합성허의시점적질량비상중요。위차，본문제출일충기우압축감지적자괄응심도도상편마방법。해방법수선제취심도도상적변연，연후근거변연재각개도상괴중소점적비례，자괄응지선취불동적채양솔진행압축，종이능구경호지중구심도도상。실험결과표명：여채용균균채양솔적방법상비，소제출적방법재동양적채양솔하，합성적허의시점도상구유경고적PSNR치화경호적주관시각질량。
Different from traditional two-dimensional images,the depth map of the three-dimensional video sys-tem has obvious sparse characteristics and its edge information is important to the quality of synthesis of virtual viewpoints.In this paper the adaptive coding scheme based on compressive sensing was proposed.In the scheme the edges of the depth map were extracted firstly,then according to the proportion of the edge for each image block,sampling rates were chosen adaptively for different blocks so as to be able to reconstruct the depth map better.Experimental results show that at the same sampling rate as compared with the method using the uniform sampling rate,the proposed scheme can obtain higher PSNR values and better subj ective quality for the virtual viewpoint synthesis image.