后处理在数字抠像中的应用与解析

发布时间：2019-03-25 08:26

【摘要】：数字抠像是将一幅图像中的前景物体与背景进行分离的问题,它的关键在于Alpha通道的计算.以往通过采样方法求得的Alpha中,由于采用逐点计算的离散化方式,求解出的Alpha通常不连续,并且包含很多噪声,因而需要对Alpha进行后处理,这不仅会增强Alpha在视觉上的平滑性,而且能够进一步提高Alpha的精确度.在目前国际上,有关数字抠像后处理领域已经进行了许多研究,但缺少相关的综述性文献,并且对后处理后的Alpha如何进行定量的评价也仍未系统解决.本文首先将数字抠像中的后处理方法分为2类:与仿射类方法相结合的方式及自平滑方式,其次,对两类方法进行了全面的总结和梳理,并对方法的优缺点进行了分析,对将来研究方向提出了建议,最后,针对后处理后的Alpha结果进行了全面的量化比较,弥补了传统方法基本上仅在视觉层面上进行比较的缺陷.
[Abstract]:Digital matting is a problem of separating foreground object from background in an image. The key of digital matting is the calculation of Alpha channel. In the former Alpha obtained by sampling method, the Alpha obtained by point-by-point calculation is usually discontinuous and contains a lot of noise, so it is necessary to post-process the Alpha, which will not only enhance the visual smoothness of Alpha, but also improve the visual smoothness of Alpha. Moreover, it can further improve the accuracy of Alpha. At present, a lot of research has been done in the field of digital matting post-processing, but there is a lack of relevant comprehensive literature, and the quantitative evaluation of Alpha after post-processing has not been systematically solved. In this paper, the post-processing methods of digital matting are divided into two categories: the combination of affine method and self-smoothing method. Secondly, the two methods are summarized and combed comprehensively, and the advantages and disadvantages of the method are analyzed. Suggestions for future research direction are put forward. Finally, a comprehensive quantitative comparison of Alpha results after post-processing is made, which makes up for the defects of traditional methods which are basically only compared at the visual level.
【作者单位】：哈尔滨商业大学计算机与信息工程学院;哈尔滨工业大学计算机科学与技术学院;
【基金】：国家重点基础研究发展计划项目(No.2015CB351804) 国家自然科学基金(No.61472103);国家自然科学基金重点项目(No.61133003) 哈尔滨商业大学博士科研启动项目(No.15KJ06)
【分类号】：TP391.41

【相似文献】