棉花异性纤维图像分割方法  

Toward Image Segmentation of Foreign Fibers in Lint

作　　者：杨文柱[1,2] 李道亮[1] 魏新华[3] 康玉国 李付堂 

机构地区：[1]中国农业大学信息与电气工程学院,北京100083 [2]河北大学数学与计算机学院,保定071002 [3]江苏大学现代农业装备与技术省部共建教育部重点实验室,镇江212013 [4]北京中棉机械成套设备有限公司,北京100089

出　　处：《农业机械学报》2009年第3期156-160,171,共6页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家科技支撑计划资助项目(2006BAD11A14-1)

摘　　要：通过直方图分析,建立了适用于棉花异性纤维图像的增强模型;提出了改进Otsu法,并对Otsu方法的分割效果进行了深入分析。实验结果表明,所有棉花异性纤维图像的灰度直方图基本都呈单峰特性;所建的增强模型可以显著提高异性纤维目标与皮棉背景之间的对比度;改进Otsu法将最佳阈值的搜索范围从0255缩减到150230,使此环节的计算速度提高了2倍多。In computer-vision-based systems for detecting foreign fibers,an efficient segmentation method is vital for improving the speed and precision of detection.An image enhancement model was presented based on histogram analysis,an improved Otsu＇s method was proposed for segmenting gray images of foreign fibers,and the segmentation results were analyzed in detail.The results indicate that all histograms of gray images of foreign fibers are of single peak,and the contrast between object and background can be remarkably enhanced by the enhancement model presented.The Otsu＇s method is effective for images of sheet foreign fibers,but does not work well for images of wirelike or villiform fibers,as some background pixels with gray values closed to the threshold are easily segmented as object pixels by mistake.But after the images were enhanced,the satisfactory segmentation results were achieved.The improved Otsu＇s method reduced the searching range for calculating optimal threshold from 0～255 to The calculating speed in this stage was improved more than twice.

关 键 词：棉花 异性纤维 直方图分析 图像分割 改进OTSU法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] S126[自动化与计算机技术—计算机科学与技术]