通道相关自适应去马赛克研究

In digital color imaging, a raw image is typically obtained via a single sensor covered by a color filter array (CFA), which allows only one color component to be measured at each pixel. The procedure to reconstruct a full color image from the raw image is called demosaicking. The demosaicking algorithm is crucial to the quality of demosaicked images. Researchers find that the different relationships among channels make the demosaicking algorithms perform different. Nowadays, a few work consider the relationship among channels, but the results are not satisfying. Hence, this proposal aims to study “what is the relationship among channels, how it affects demosaicking, and how to eliminate the effect”. We first propose the concept of channel-correlation and analyze its effect to demosaicking. Then we design corresponding dictionaries to raw images with different channel-correlations to get adaptive demosaicking. Finally, we solve the adaptive demosaicking from the perspective of the meaning of the channel-correlation. This proposal will make the demosaicking procedure adaptively to different channel-correlations and improve the quality of color imaging. In addition, this research will develop the channel-correlation adaptive demosaicking theory.

在数字彩色成像中，原始图像由滤色阵列(Color Filter Array, CFA)覆盖的单个传感器获取，原始图像的每个像素仅记录一种颜色的值，从原始图像重建彩色图像的过程称为去马赛克(Demosaicking)。去马赛克决定了彩色图像的质量，研究发现由于图像的颜色通道之间的关系不同，导致现有去马赛克方法在不同图像上的性能差别很大。目前少数工作考虑了颜色通道之间的关系，然而效果并不理想，本项目拟针对此问题研究“什么是通道之间的关系、其如何影响去马赛克，如何解决其对去马赛克的影响”，首先提出通道相关性的概念，分析其对去马赛克的影响，在此基础上对不同原始图像的通道相关性设计对应的字典，初步实现自适应去马赛克。最后本项目拟从通道相关性本身的意义出发解决自适应去马赛克问题。本项目的研究将实现去马赛克根据不同的通道相关性自适应地完成，提高彩色成像的质量，极大地推动通道相关自适应去马赛克的理论发展。

在数字彩色成像中，原始图像由滤色阵列(Color Filter Array, CFA)覆盖的单个传感器获取，原始图像的每个像素仅记录一种颜色的值，从原始图像重建彩色图像的过程称为去马赛克(Demosaicking)。去马赛克决定了彩色图像的质量，由于图像的颜色通道之间的关系不同，导致现有去马赛克方法在不同图像上的性能差别很大。本课题提出通道相关性的概念，分析其对去马赛克的影响，在此基础上对不同原始图像的通道相关性设计对应的字典，实现自适应去马赛克。本课题的研究实现去马赛克根据不同的通道相关性自适应地完成，提高彩色成像的质量，极大地推动通道相关自适应去马赛克的理论发展。相关的成果包括发表国际期刊论文3篇，国际会议论文1篇，培养已毕业硕士研究生2名。