基于改进Otsu和Niblack的图像二值化方法  

Algorithm for Image Binarization Based on Improved Otsu and Niblack

作　　者：石为人[1] 方莉[1] 陈舒涵[1] 

机构地区：[1]重庆大学自动化学院,重庆400030

出　　处：《世界科技研究与发展》2013年第3期395-398,共4页World Sci-Tech R&D

基　　金：国家科技重大专项:课题3(2009ZX07528-003-09);十二五国家科技支撑计划项目(2011BAK07B03)资助

摘　　要：针对常用的图像二值化方法——全局阈值方法和局部阈值方法存在的对输入图像含有噪声或不均匀光照等情况抵抗能力差以及处理速度慢的不足,提出了一种全局阈值与局部阈值相结合的图像二值化方法。通过在传统Otsu算法中引入类内方差和快速阈值搜索方法,并将其与Niblack算法结合,将图像的每一行看成一个子图,对每一子图使用改进的阈值进行二值化。实验证明,该方法抗噪声能力强,能准确获取图像的目标信息,同时能保持较快的处理速度。Aimed at the weaknesses that the global thresholding teehniques have poor resistance to the input image with noise or uneven illu- ruination and local methods are more time-consuming, a new method for image binariation with global and local threahold combined was pres- ented. The within-class variance and rapid search strategy are introduood into traditional Otsu, Combined with Nihlack, considering each row of the image as a suh-image and selecting an improved threshold for eaoh aub-image.The experiments show that the new method has advantages of certain anti-noise abilities and keeps a faster speed,while the target can be extracted more precisely,

关 键 词：数字图像处理 二值化 改进Otsu算法 Niblack算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]