树突状细胞算法在图像分类中的应用  

APPLICATION OF DENDRITIC CELL ALGORITHM IN IMAGE CLASSIFICATION

作　　者：常文萃[1] 秦振吉[1] 

机构地区：[1]晋中学院,山西晋中030600

出　　处：《计算机应用与软件》2014年第4期250-253,共4页Computer Applications and Software

摘　　要：由于传统树突状细胞算法存在大量的随机参数,很难对算法的有效性进行评价与分析,提出一种用于图像分类的改进树突状细胞算法。针对图像的颜色及纹理特征,定义抗原和分类元素,根据抗原的危险度对目标进行聚类;设计更精确的异常程度指标,简化算法的输入信号,并对处理过程进行优化,使其具有较快的收敛速度和稳定的识别性能。实验证明,该算法能够提高图像的识别率以及识别过程的稳定性,并且减少了算法运行时间。Since traditional dendritic cell algorithm has massive random parameters which results in the difficulty of assessment and analysis on the effectiveness of the algorithm,we propose a modified dendritic cell algorithm used for image classification. It defines the antigen and classifies the elements in light of the colour and textural features of images,and clusters the objectives according to the degree of antigen＇s risk. More accurate anomaly intensity index is designed. It also simplifies the input signals and optimises the processing procedure, thus can achieve faster convergence speed and more stable recognition performance. Experimental result shows that this algorithm is able to improve the recognition rate of image and the stability in recognition process,and reduces the running time of the algorithm as well.

关 键 词：人工免疫系统 危险理论 树突状细胞算法 图像分类 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]