计算机应用 ›› 2011, Vol. 31 ›› Issue (04): 1040-1042.DOI: 10.3724/SP.J.1087.2011.01040

• 图形图像技术 • 上一篇    下一篇

基于轮廓波局部高斯模型与全变差的彩色滤波阵列插值

丁丽媛,练秋生   

  1. 燕山大学 信息科学与工程学院,河北 秦皇岛 066004
  • 收稿日期:2010-09-28 修回日期:2010-11-25 发布日期:2011-04-08 出版日期:2011-04-01
  • 通讯作者: 丁丽媛
  • 作者简介:丁丽媛(1985-),女,河北邢台人,硕士研究生,主要研究方向:数字图像处理;
    练秋生(1969-),男,江西遂川人,教授,博士,主要研究方向:数字图像处理、模式识别、压缩传感。
  • 基金资助:
    国家自然科学基金资助项目(60873058);河北省自然科学基金资助项目(F2010001294)

Color filter array interpolation based on contourlet local Gaussian model and total variation

Li-yuan DING,Qiu-sheng LIAN   

  1. Institute of Information Science and Engineering, Yanshan University, Qinghuangdao Hebei 066004, China
  • Received:2010-09-28 Revised:2010-11-25 Online:2011-04-08 Published:2011-04-01
  • Contact: Li-yuan DING

摘要: 彩色图像的彩色滤波阵列(CFA)插值是从单传感器数字相机通过CFA获得的采样图像中重构完整RGB图像的过程。针对Bayer格式图像提出了一种基于轮廓波局部高斯模型与全变差的彩色图像CFA插值算法。为进一步提高图像边缘插值质量,将图像梯度的稀疏性结合到图像插值过程中,并且图像梯度的稀疏性用彩色全变差(CTV)来衡量。实验结果表明,该算法比现有的图像插值算法在峰值信噪比与主观视觉效果两方面均有显著提高。

关键词: 彩色滤波阵列, 插值, 轮廓波, 局部高斯模型, 彩色全变差

Abstract: Single-chip digital cameras use Color Filter Array (CFA) to sample different color information; the color image CFA interpolation algorithm interpolates these data to produce an RGB image. A color image CFA interpolation algorithm was proposed based on contourlet local Gaussian model and Total Variation (TV). In order to improve the edge interpolation quality, the sparsity of image gradient was integrated in interpolation process, and the Color Total Variation (CTV) was introduced to measure the sparsity of the image gradient. The experimental results show that the proposed algorithm outperforms the classical algorithms in terms of both Peak Signal-to-Noise Ratio (PSNR) and visual quality.

Key words: Color Filter Array (CFA), interpolation, contourlet, local Gaussian model, Color Total Variation (CTV)

中图分类号: