自适应模糊集与分数阶微分相结合的股骨头DR影像增强  

DR IMAGE ENHANCEMENT OF FEMORAL HEAD BASED ON ADAPTIVE FUZZY SET AND FRACTIONAL DIFFERENTIAL

作　　者：孙福权[1,2] 孔超然 张琨[1,2] 姜玉山[1,2] 丛成龙 Sun Fuquan;Kong Chaoran;Zhang Kun;Jiang Yushan;Cong Chenglong(Northeastern University at Qinhuangdao,Qinhuangdao 066004,Hebei,China;Northeastern University,Shenyang 110819,Liaoning,China)

机构地区：[1]东北大学秦皇岛分校,河北秦皇岛066004 [2]东北大学,辽宁沈阳110819

出　　处：《计算机应用与软件》2020年第12期191-196,共6页Computer Applications and Software

基　　金：国家重点研发计划项目(2018YFB1402800);教育部科技发展中心科研创新项目(2018A03031);全国教育信息技术研究规划课题重点项目(16222874)。

摘　　要：针对部分DR影像对比度低、影像纹理细节模糊不清、含有噪声过高等问题,提出一种自适应阈值模糊集增强与分数阶微分增强相结合的图像增强算法。使用双正交小波变化将原始图像分解成多个频带分量;对于低频子带分量,通过自适应提取阈值并构建新的隶属函数,将影像映射到适应该影像的模糊空间并进行增强以提高图像整体的对比度;对于高频子带分量,通过构造分数阶微分掩模并与高频子带分量进行卷积达到增强股骨头影像高频带的目的。实验结果表明:与其他方法相比,该算法能够有效地提高图像对比度、抑制噪声,使医学影像更多细节展示出来,进而提高其可读性与医疗诊断的准确度。Aiming at the problems of low contrast,blurred texture details and high noise in some DR images,we propose an image enhancement algorithm combining adaptive threshold fuzzy set enhancement and fractional order differential.The original image was decomposed into several sub-band components by using biorthogonal wavelet transform.For the low-frequency sub-band components,the image was mapped to the fuzzy space by adaptive threshold extraction and constructing a new membership function to improve the overall contrast of the image.For the high frequency sub-band component,the fractional differential mask was constructed to convolute the high frequency sub-band component to enhance the high-frequency band of femoral head image.The experimental results show that compared with other methods,our algorithm can effectively improve the image contrast,suppress the noise,make the medical image more detailed,and improve the readability of the image and the accuracy of medical diagnosis.

关 键 词：图像增强 小波变换 分数阶微分 模糊增强 DR影像 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]