基于Mean-shift的棉花异性纤维图像分割  

IMAGE SEGMENTAION OF THE COTTON FOREIGN FIBERS BASED ON MENA-SHIFT

作　　者：郑文秀[1] 刘双喜[1] 魏新华[1] 康玉国 李付堂 王金星[1] 

机构地区：[1]山东农业大学机械与电子工程学院,山东泰安271018 [2]北京中棉机械成套设备有限公司,北京100089

出　　处：《山东农业大学学报（自然科学版）》2009年第2期224-228,共5页Journal of Shandong Agricultural University：Natural Science Edition

基　　金：“十一五”国家科技支撑计划子课题(2006BAD11A14-3)

摘　　要：提出了一种基于均值平移(Mean-shift)算法和膨胀、滤波算法相结合的棉花异性纤维图像分割方法。选择具有代表意义的头发、麻绳和塑料布的灰度图像进行处理,首先对棉花异性纤维图像采用Mean-shift算法进行图像分割,然后对分割后的图像元素进行膨胀处理和滤波处理,实现棉花异性纤维图像的精确分割。实验结果表明,本文采用的图像分割方法具有较强的抗噪声能力,达到了比较满意的分割效果。A method of foreign fibers image segmentation based on Mean shift, dilation and filtering algorithm is presented. For the representative gray images of hair,hemp rope and plastic cloth, the Mean shift algorithm was used to carry on image segmentation; then dilation and filtering process was carried on to the divided image element. In this way the precise image segmentation of foreign fibers was realized. It＇s proved by experiments that the image segmentation method proposed by this article can suppress the noise well, and the segmentation results were satisfied for all kinds of foreign fibers image.

关 键 词：异性纤维 直方图分析 图像分割 Mean SHIFT 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]