结合沃尔什变换与列率截断的图像局部模糊抗噪检测  

A Noise-Robust Partial Blur Detection Algorithm Combining Walsh Transform and Sequency Truncation

作　　者：梁晓 王雪玮 郭京波[1,2] 韩彦军 郑津津 郭文武[1,2,3] Liang Xiao;Wang Xuewei;Guo Jingbo;Han Yanjun;Zheng Jinjin;Guo Wenwu(State Key Laboratory of Mechanical Behavior and System Safety of Traffic Engineering Structures,Shijiazhuang Tiedao University,Shijiazhuang 050043;School of Mechanical Engineering,Shijiazhuang Tiedao University,Shijiazhuang 050043;Hebei Advanced Manufacturing Technology Innovation Center for Concrete Components,Shijiazhuang 050043;Department of Precision Machinery and Precision Instrumentation,University of Science and Technology of China,Hefei 230027)

机构地区：[1]石家庄铁道大学省部共建交通工程结构力学行为与系统安全国家重点实验室,石家庄050043 [2]石家庄铁道大学机械工程学院,石家庄050043 [3]河北省混凝土构件先进制造技术创新中心,石家庄050043 [4]中国科学技术大学精密机械与精密仪器系,合肥230027

出　　处：《计算机辅助设计与图形学学报》2022年第1期94-103,共10页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(62003227,52102467);河北省自然科学基金(F2021210016);省部共建交通工程结构力学行为与系统安全国家重点实验室自主课题(ZZ2021-12);河北省高等学校科学技术研究项目(QN2019066,QN2021135).

摘　　要：图像局部模糊的有效检测是计算机视觉领域的一项挑战性任务,现有模糊检测算法多难以兼顾准确性和实时性,并且在检测噪声混叠的模糊图像时有较大局限.为此,提出一种无监督且抗噪的局部模糊快速检测算法.首先利用主动模糊策略和沃尔什变换对待测图像进行列率域解析,并自适应截断列率谱的低列率区域以消除噪声干扰;在此基础上进一步构造并求解各像素点的局部模糊度量,得到待测图像的模糊分布;最终在聚类引导下采用多尺度修正生长实现局部模糊区域的分割.在CUHK,DUT等代表性数据集上的实验结果表明,所提算法可在无监督情况下快速、有效地检测图像模糊并准确分割局部模糊区域,在精确率、召回率、F_(1)测度、平均绝对误差、平均处理时间等多个评估指标上均接近或超过同类算法的最优水平,尤其在噪声情况下具有显著优于同类算法的检测性能.Blur detection is an important yet challenging task in computer vision.The previous algorithms are mostly difficult to achieve a cost-benefit balance and their performance is largely limited when faced with the blur image polluted by noise.To address these issues,a fast and unsupervised blur detection algorithm is proposed,which is robust to noise.First,a re-blur strategy and Walsh transform are utilized to analyze the input image in sequency domain.Meanwhile,the low-sequency zone of sequency spectrum is adaptively truncated to eliminate the noise interference.Then,a noise-robust local blur metric is constructed and pixel-wise blurriness is calculated to obtain the blur map.Finally,the blur region is segmented using the clustering-guided multi-scale growth framework.Experimental results on CUHK and DUT datasets demon strate that the proposed algorithm can detect the image blur effectively and efficiently,and achieves the state-of-the-art performance on multiple indicators like precision,recall,F1-measure,mean absolute error,and mean runtime.Especially on noise-polluted conditions,the proposed algorithm significantly surpasses other competitive algorithms.

关 键 词：局部模糊 列率谱 模糊检测 自适应截断 噪声鲁棒性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]