结合不可分小波分析和曲线拟合方法的图像对比度增强  

Image Contrast Enhancement Combining Nonseparable Wavelet Analysis and Curve Fitting Method

作　　者：刘斌 辛迦楠 谌文江 Liu Bin;Xin Jianan;Chen Wenjiang(School of Computer and Information Engineering, Hubei University, Wuhan, Hubei 430062, Chin)

机构地区：[1]湖北大学计算机与信息工程学院,湖北武汉430062

出　　处：《激光与光电子学进展》2018年第2期223-232,共10页Laser & Optoelectronics Progress

基　　金：国家自然科学基金面上项目(61471160);湖北省自然科学基金重点项目(2012FFA053)

摘　　要：传统的图像对比度增强技术容易造成图像的局部区域过分增强或增强不足等问题,因此提出了一种结合不可分小波分析和曲线拟合的图像对比度增强方法。该方法以标准清晰图像的均值和标准差作为图像对比度增强的理想目标,首先对同一目标的标准清晰图像和低对比度图像进行四通道不可分小波分解,然后利用曲线拟合的预测能力得出分解后具有相同频率通道和相同方向子图像的均值之间及标准差之间的函数关系,通过该函数关系和直方图匹配公式即可得到理想的增强后的子图像,最后将这些新的子图像进行不可分小波重构,并采用清晰度、标准差指标对图像对比度增强结果进行评价。实验结果表明,所提方法所得低对比度图像的增强效果较好,且得到了高清晰度图像。Traditional image contrast enhancement technology can easily cause over-enhancement or under- enhancement in the local area of images. Thus, an image contrast enhancement method combining nonseparable Wavelet analysis and curve fitting is proposed. The method uses mean and standard deviation of the standard clear image as the ideal target for image contrast enhancement. Firstly, four-channel nonseparable wavelet is used to decompose standard clear image and low contrast image of the same target. Next, the predictive ability of curve fitting is used to get the relationships among mean values and the relationships among standard deviations of sub- images with same frequency channel and same direction. Then, the ideal enhanced sub-images can be obtained by the functional relationships and histogram matching formulas. Finally, these new sub-images are reconstructed by nonseparable wavelet. The performance of the method is evaluated with clarity and standard deviation. The experimental results show that the proposed method has a good enhancement effect for low contrast image, and can obtain high definition images.

关 键 词：图像处理 图像对比度增强 不可分小波 曲线拟合 直方图匹配 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]