摘要
针对传统超分辨率重建算法中存在的不足,提出一种基于非下采样轮廓波变换(NSCT)和插值的单幅图像超分辨率重建算法。首先对低分辨率图像进行NSCT变换处理,将得到的低频子带及高频子带系数分别进行软判决自适应插值和三次样条插值,再对三次样条插值后的高频系数进行形态学增强。对插值后的低频子带进行稀疏表示,通过香农熵取大的融合规则进行融合;对于高频子带,采用一种运用子带系数的空间频率、梯度指标信息,并与高斯隶属函数相结合的自适应融合规则进行融合。最后对融合后的系数进行NSCT逆变换得到重建图像。实验结果表明,该算法无论在主观视觉效果和峰值信噪比、结构相似性等客观指标上均优于其他经典的重建算法,进而验证了该算法的有效性。
To overcome the shortages of traditional super-resolution algorithms,a novel single-image super-resolution algorithm based on non-subsampled contourlet transform( NSCT) and interpolation is proposed. Firstly,the coefficients of low frequency sub-band and high frequency sub-band are obtained through NSCT,and these coefficients are interpolated separately by soft-decision adaptive interpolation( SAI) and cubic splines interpolation( CSI). Then,the high frequency sub-band interpolated by CSI is enhanced. Next,for the low frequency sub-band coefficient,a fusion method based on sparse representation is presented,and the fused coefficient is obtained through the rule of "max-Shannon entropy". For the high frequency sub-band,the fused coefficients are obtained by combining its spatial frequency information and gradient information with Gaussian function. Finally,the high-resolution image is obtained by performing the inverse NSCT on the fused coefficients. The experimental results show that the proposed algorithm can improve the quality of the high-resolution image,and outperforms other classical fusion algorithms in terms of both visual quality and objective evaluation,and the effectiveness of the algorithm is verified.
出处
《电子测量与仪器学报》
CSCD
北大核心
2018年第1期19-25,共7页
Journal of Electronic Measurement and Instrumentation
基金
国家自然基金(11301131)资助项目
关键词
超分辨率重建
形态学增强
稀疏表示
图像融合
自适应加权
super-resolution reconstruction
morphological enhancement
sparse representation
image fusion
adaptive weighted
