一种光照不均匀图像的二值化方法  

BINARISATION METHOD FOR IMAGES ACQUIRED UNDER NON-UNIFORM ILLUMINATION

作　　者：郭佳[1] 刘晓玉[1] 吴冰[1] 付晓薇[1] 

机构地区：[1]武汉科技大学信息科学与工程学院,湖北武汉430081

出　　处：《计算机应用与软件》2014年第3期183-186,202,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(61201423)

摘　　要：在图像处理与分析中,二值化方法能把特征或目标从图像中分割出来,以便识别。在实际拍摄的图像中,图像光照不均匀使得图像二值化阈值确定困难。针对一类光照不均匀图像的二值化问题,提出一种单列处理及改进OSTU的二值化方法,通过分析该类图像上不均匀光照的分布特点,首先应用单列二值化处理方案,在省去专门光照均匀化处理步骤的同时,解决统一阈值二值化方法所造成的信息丢失问题。在逐列二值化处理过程中,首先采用传统OSTU方法获取图像各列的初始二值化阈值,然后根据各列的光照强度计算出该列二值化阈值的自适应微调量,实现对初始二值化阈值的优化。一组缺陷钢坯图片处理结果证实了所述方法的有效性。In image processing and analysing, binarisation method can be used to extract the features or targets from images for identification. But in actual shooting images, it is difficult to determine the binarisation threshold due to the non-uniform illumination on the image. To solve the binarisation problem of a class of non-uriiform illumination, we propose a binarisation method which is based on single- column processing and improved OSTU. By analysing the distribution character of non-uniform illumination in such class of images, the scheme of single-column binarisation processing is firstly applied, while eliminating the step of specialised illumination uniformity processing, the losing information problem caused by the unified threshold binarisation method is overcome. During column by column binarisation processing, classical OSTU method is firstly used to obtain initial binarisation threshold values of each column of the image, then according to the illumination intensity of each column, the adaptive fine-tune value of the binarisation threshold of this column is calculated, thus we implement the optimisation on initial binarisation threshold. The method has been proved to be effective by the image processing results of a group of defective billet images.

关 键 词：二值化阈值 光照不均匀 OSTU法 自适应微调量 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]