王东升,杨斌.基于梯度局部一致性的自适应Bayer模式彩色图像恢复算法[J].南华大学学报(自然科学版),2019,33(2):40~48.[WANG Dongsheng,YANG Bin.Adaptive Bayer Pattern Color Image Restoration Algorithm Based on Gradient Local Consistency[J].Journal of University of South China(Science and Technology),2019,33(2):40~48.] |
基于梯度局部一致性的自适应Bayer模式彩色图像恢复算法 |
Adaptive Bayer Pattern Color Image Restoration Algorithm Based on Gradient Local Consistency |
投稿时间:2018-10-30 |
DOI: |
中文关键词: 彩色滤波阵列 Bayer模式阵列 去马赛克 自适应插值 |
英文关键词:color filter array Bayer-pattern array demosaicking adaptive interpolation |
基金项目:国家自然科学基金项目(61871210;61102108);湖南省自然科学基金项目(2016JJ3106);湖南省教育厅优秀青年项目(16B225;YB2013B039);南华大学青年英才支持计划、南华大学重点学科项目(NHXK04);南华大学研究生科学基金项目(2018KYY006) |
|
摘要点击次数: 1066 |
全文下载次数: 408 |
中文摘要: |
基于图像局部梯度具有明显一致性先验,提出了一种基于梯度局部一致性的自适应Bayer模式彩色图像恢复算法。算法在恢复绿色通道时首先从垂直和水平两个方向的进行插值,得到候选插值结果。然后根据彩色通道结构相似的特点,提出利用多通道梯度线性加权估计图像局部梯度,并利用引导滤波实现用时域图像边缘结构对图像局部梯度进行一致化校验,从而提高边缘估计的精度。然后将一致化校验后的梯度特征与候选插值进行自适应融合得到最终绿色通道插值。最后依据图像通道之间的结构相似性先验,利用绿色通道结构细节实现对红色和蓝色通道重构图像的修正,从而恢复完整的去马赛克彩色图像。通过在Kodak和IMAX两个数据库进行的有效性验证,并与其他最新方法进行的性能对比,结果表明恢复图像主观视觉特性和客观评价指标都优于传统方法。此外,由于不涉及迭代和网络训练,具有高效性。 |
英文摘要: |
Based on the image local gradient with obvious consistency a priori,this paper proposed an adaptive Bayer model color image restoration algorithm based on gradient local consistency.When the algorithm restores the green channel,it first interpolates from the vertical and horizontal directions to obtain candidate interpolation results.Then according to the similar characteristics of the color channel structure,the multi-channel gradient linear weighting is used to estimate the local gradient of the image,and the guided filtering is used to achieve the consistency check of the image local gradient with the edge structure of the time domain image,thus improving the accuracy of the edge estimation.Then,the gradient feature after the uniform verification is adaptively fused with the candidate interpolation to obtain the final green channel interpolation.Finally,according to the structural similarity prior to the image channel,the green channel structure details are used to correct the red and blue channel reconstructed images,thereby restoring the complete demosaicking color image.Through the validation of the two databases in Kodak and IMAX,and the performance comparison with other recent methods,the results show that the subjective visual characteristics and objective evaluation indicators of the restored images are superior to the traditional methods.In addition,it is efficient because it does not involve iteration and network training. |
查看全文 查看/发表评论 下载PDF阅读器 |
关闭 |
|
|
|