基于Haar特征和级联结构的CISH细胞检出  

CISH cell detection based on Harr feature and cascade structure

作　　者：王晖[1] 刘秉瀚[1] 

机构地区：[1]福州大学数学与计算机科学学院,福州350108

出　　处：《中国体视学与图像分析》2012年第1期20-26,共7页Chinese Journal of Stereology and Image Analysis

基　　金：福建省自然科学基金(2009J01283);福建省科技重点项目(2011Y0040)

摘　　要：目的提出一种基于Haar特征和Gentle Adaboost级联结构的CISH细胞自动检出方法。方法根据乳腺癌CISH图像染色导致的细胞边缘模糊,不利于细胞的检出的问题。首先,对待检测图像进行光照补偿,减小光照对细胞检测造成的影响;然后,利用针对CISH细胞的Haar特征训练的级联强分类器快速检测细胞位置;最后,利用CISH细胞的结构特征对图像进行后处理,定位细胞的具体轮廓。结果实现乳腺癌CISH细胞的检出。结论实验结果表明,该算法检测速度快,并且具有较低误检率和漏检率,能够较好的应用于CISH细胞的快速检出。Objective To present an automatic detection method for chromogenic in situ hybridization （CISH） cells based on Harr feature and Gentle AdaBoost cascade structure. Methods CISH staining of breast cancer cells causing cell edge blur in the CISH images is unfavorable for cell image detection. In order to overcome this obstacle, at first is to make illumination compensation of the test image, and to reduce the disadvantageous influence of light illumination on the cell detection. Secondly, based on the Haar features of the CISH cells to train a strong classifier cascade, and use it to rapidly detect the possible cell location, which may include false detection area. Finally, use the CISH cell structure to remove false detection area and locate the concrete contour of the cells. Results The CISH breast cancer cells were well detected. Conclusions The experiment demonstrate that the speed of detection of CISH breast cancer cells is increased and the false detection rate and missing rate are reduced, indicating that this method is suitable for rapid detection of CISH cells.

关 键 词：CISH细胞 细胞检测 HAAR特征 级联结构 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]