摘要
针对现有图像增强算法边缘保持性能不佳、抗噪性弱的问题,提出了一种改进的引导滤波图像增强算法——ABCO-NSST-GF。通过非下采样Shearlet变换(NSST)将图像分解成低频和高频2部分,利用引导滤波来增强低频系数,避免了高频噪声的放大;对图像的高频系数进行非线性增益函数变换,在增强边缘及细节的同时抑制噪声。最后,对处理后的低频和高频系数实施NSST反变换,重构出最终的增强图像。由于引导滤波中的盒滤波半径与正则化参数对增强结果有较大影响,采用了混沌蜂群算法搜索其最佳值,确保增强结果达到最优。针对约70幅实际工程图像进行了实验,结果表明,ABCO-NSST-GF算法能够明显改善图像视觉效果,与NSCT自适应阈值法等4种算法相比,所得图像清晰度、对比度和信息熵平均提高25.2%,与空域引导滤波算法相比,P峰值信噪比平均提高20.9%。
An improved image enhancement algorithm with guided filtering— ABCO-NSST-GF is proposed to solve the shortcomings of existing image enhancement algorithms in edge preservation and anti-noise performance.The NSST decomposes an input image into a lowfrequency component and several high-frequency components,and then the guided filtering is utilized to enhance the low-frequency coefficients to avoid amplifying noises in the process of image enhancement.The high-frequency coefficients are transformed by a nonlinear gain function so that the edges and details are enhanced while the noise is suppressed.Finally,the resultant image is reconstructed by applying the inverse NSST to the processed low-frequency coefficientsand high-frequency coefficients.Since the box filter radius and regularization parameter of guided filtering have significant influences on enhancement effects,the chaotic bee colony optimization algorithm is adopted to find their optimal values for best enhancement effects.Experiments on about 70 practical engineering images show that the ABCO-NSST-GF algorithm significantly improves visual effects.Comparisons with 4 existing algorithms such as adaptive threshold algorithm based on NSCT show that the quantitative evaluation indicators of the ABCO-NSSTGF algorithm such as definition,contrast and entropy get about 25.2% average improvement,while a comparison with the spatial guided filtering enhancement algorithm shows that the proposed algorithm has a 20.9%improvement in PSNR.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2015年第6期39-45,共7页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(60872065)
数字制造装备与技术国家重点实验室开放基金项目(DMETKF2014010)
国土资源部地质信息技术重点实验室开放基金项目(217)
江西省数字国土重点实验室开放基金项目(DLLJ201412)