基于改进ITTI模型及粒子群优化算法的白细胞区域提取  

Extraction of Leukocyte Areas Based on Improved ITTI and Particle Swarm Optimization Algorithm

作　　者：纪滨[1] 杨盼盼[1] 申元霞[1] 

机构地区：[1]安徽工业大学计算机科学与技术学院,安徽马鞍山243032

出　　处：《安徽工业大学学报（自然科学版）》2016年第3期284-288,共5页Journal of Anhui University of Technology（Natural Science）

基　　金：国家自然科学基金项目(61300059)

摘　　要：白细胞显微图像病理分析中,人眼关注的白细胞是感兴趣的区域。ITTI视觉模型是提取图像感兴趣区域(ROI)的有效办法。为了进一步改善其提取的准确性,提出了基于改进的ITTI视觉模型与粒子群优化算法相结合的目标控制方法,并将其应用于医学骨髓细胞图像中的白细胞区域提取。首先利用高斯滤波和多尺度归一化的方法分别提取原始图像的方向、亮度、颜色显著性特征,再根据人眼的视觉对不同显著性特征敏感程度不同的特性对3种显著性特征采用自适应系数相融合的方式得到显著图,最后利用基于改进的粒子群优化算法的Otsu法对显著图进行ROI的提取,并采用数字形态学的方法对其进行后续处理。结果表明,本文算法可以较好地提取完整的白细胞区域,有助于提高病理分析的效率。In the process of pathological analysis of the Leukocyte microscopic image, the leukocyte areas areregions of interest（ROI）. The ITTI visual model is an effective method of extracting the ROI from image. Forimproving the extracting accuracy, an object extracting method combining the improved ITTI visual model withparticle swarm optimization algorithm is proposed and used to extract the ROI from the bone marrow cell image.Firstly, based on Gaussian filter and multi- scale normalization, features of the orientation, brightness, and colorare computed from the original image. And then, according the fact that the sensitivity of eyes is not the same asdifferent features the saliency map is obtained with adaptive coefficient from three significant characteristics.Finally, by using Otsu method based on the improved particle swarm optimization（PSO） algorithm, the ROIs areextracted and subsequent processed with method of morphology. Experimental results show that this method canextract white blood cell areas perfectly, which is helpful for improving the efficiency of pathological analysis.

关 键 词：感兴趣区域 显著图 粒子群优化算法 ITTI模型 白细胞 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]