基于改进的Otsu方法的钢渣彩色图像分割  

Segmentation of steel slag color image based on improved Otsu method

作　　者：严晨曦 熊凌 YAN Chen-xi;XIONG Ling(Institute of Robotics and Intelligent Systems,Wuhan University of Science and Technology,Wuhan 430081,China;School of Information Science and Engineering,Wuhan University of Science and Technology,Wuhan 430081,China)

机构地区：[1]武汉科技大学机器人与智能系统研究院,湖北武汉430081 [2]武汉科技大学信息科学与工程学院,湖北武汉430081

出　　处：《计算机工程与设计》2020年第9期2585-2591,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(51475340);湖北省科技厅湖北省技术创新专项重点基金项目(2017AFB264、2018AFB363)。

摘　　要：针对彩色钢渣图像目标与背景区域颜色存在较大差异以及钢渣亮光被误分割为钢渣的问题,采用基于Lab颜色空间的K均值聚类算法对彩色图像进行预处理,提出一种改进的Otsu方法对钢渣彩色图像进行分割。将RGB颜色空间通过各种变换转到均匀的Lab颜色空间;利用k均值聚类算法对图像进行预处理,预处理后的图像基本实现钢渣与背景的分离;对传统的Otsu方法进行改进,将背景像素点占整幅图像的比例作为权重引入到目标方程式中,用改进的Otsu方法对预处理后的图像进行阈值分割。通过对采集的彩色钢渣图像进行分割实验,将改进的Otsu方法与其它方法进行对比分析,实验结果表明,该方法有效地改善了分割效果,降低了分割误差。To solve the problem of color difference between target and background area of color steel slag image and the problem that the segmentation of bright light of steel slag is mis-segmented into steel slag,K-means clustering algorithm based on Lab color space was used to preprocess the color image,and an improved Otsu method was proposed to segment the steel slag color image.RGB color space was transformed into uniform Lab color space through various transformations.The image was prep-rocessed using K-means clustering algorithm,and the preprocessed image basically achieved the separation of steel slag and background.The traditional Otsu method was improved by introducing the proportion of background pixels to the whole image as a weight into the objective equation,and the improved Otsu method was used to segment the pre-processed image.By segmenting the collected color steel slag image,the improved Otsu method was compared with other methods.Experimental results show that the proposed method effectively improves the segmentation effect and reduces the segmentation error.

关 键 词：Lab颜色空间 K均值聚类 OTSU方法 彩色图像分割 阈值分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]