透過您的圖書館登入
IP:3.135.216.174
  • 期刊

高品質和快速彩色影像內插演算法之設計

High-quality and High-speed Demosaicking Algorithm Design

摘要


數位相機已是消費產品主流之一,其影像品質、價錢、與體積是使用者考量之重要因素。在各種設計中,為了達到降低成本與體積的考量,數位相機採用單一感測裝置加上彩色濾波陣列的濾波片,其主要目的是減低成本與體積,但是也降低了影像的品質。在本論文中,我們提出一套演算法針對此彩色濾波陣列中所遺失的像素作內插,此內插過程稱為彩色影像內插(Demosaicking),使得經過此內插後的影像可以逼近原始影像的品質,同時,在不減損影像的品質下,我們降低濾波器的大小來減少運算時間。從PSNR值的結果分析可知我們的演算法和文獻[5][8][1][7]比較,分別提升5.6%,6.2%,9%和4.4%的效能;運算時間則較[5][8][1][7]分別增速了405%、64%、12%、179%。

並列摘要


Digital still cameras (DSCs) have already been as one mainstream of consumer products, users consider three major factors, quality of images, prices, and volume, to purchase them or not. Nowadays, although the Color Filter Array (CFA) design has been adopted in DSCs for reducing their cost and volume, the major one disadvantage of such CFA design will lower its quality of the image. In this paper, we propose a high-quality and higt-speed demosaicking algorithm design to improve this poor image quality by interpolating the lost pixels of CFA design. The ”demosaicking” is an interpolation process which is to interpolate the lost pixels into its original image on such CFA design. After demosaicking stage, the quality of this interpolated image can approach that of its original image. Simultaneously, without reducing the quality of the image, our proposed algorithm will also lower the filter dimensions to reduce the computation time. From our results of PSNR (Peak Signal to Noise ratio) analysis, the proposed novel design effectively lowers 5.6%, 6.2%, 9%, and 4.4% noise than the methods of the literals [5][8][1][7], respectively. Moreover, compared with the computation time, our proposed algorithm can effectively reduce about 405%、64%、12% and 179% computation time than of methods of the literals [5][8][1][7], respectively.

延伸閱讀