基于多子窗口快速归一化互相关的肿瘤识别  

Tumor recognition based on fast multi-subwindow normalized cross correlation

作　　者：胡顺波[1] 王广泰[1] 

机构地区：[1]临沂师范学院物理系,山东临沂276005

出　　处：《光电子．激光》2009年第3期422-424,共3页Journal of Optoelectronics·Laser

基　　金：国家"863"计划资助项目(2006AA02Z4D9)

摘　　要：对肿瘤进行快速准确地精细分割和生长监控是当前医学图像实时处理需要解决的一个问题,本文通过使用基于加和表的快速归一化互相关(FNCC)方法提高肿瘤识别速度,通过大子窗口技术提高肿瘤识别的正确度,通过多子窗口技术实现肿瘤的精细分割。对人体脑部核磁共振(MR)图像的肿瘤进行识别实验表明,本文方法可以很好地实现肿瘤的快速准确精细分割和生长监控。In order to ensure the speed and accuracy of tumor segmentation and tumor growth monitoring.firstly fast normalized cross correlation based on sum tables is utilized to increase the tumor recognition speed,and then large subwindow skill is used to identify tumor rightly,at last the multi-subwindow technique is used to achieve accurate tumor segmentation.The proposed tumor recognition method is tested by human brain magnetic resonance images which are gotten at different time.The results of tests show that t...

关 键 词：肿瘤识别 快速归一化互相关 多子窗口技术 加和表 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]