仿生模式识别在影像学肺炎判别算法的改进  

Improved Algorithm of Bionic Pattern Recognition in Imaging Pneumonia Discrimination

作　　者：邹倩颖 吴宝永 王小芳 ZOU Qianying;WU Baoyong;WANG Xiaofang(Department of Cloud Computing Science and Technology,Chengdu College of UESTC,Chengdu 611731,China;Computer Academy,China West Normal University,Nancong 637002,Sichuan,China)

机构地区：[1]电子科技大学成都学院,云计算科学与技术系,成都611731 [2]西华师范大学计算机学院,四川南充637002

出　　处：《实验室研究与探索》2020年第2期34-38,共5页Research and Exploration In Laboratory

基　　金：成都市科技局重点研发支撑计划技术创新研发项目(2018-YFYF-00191-SN)。

摘　　要：提出了一种结合卷积神经网络和仿生模式识别的改进判别算法,以仿生模式识别为基础,首先构建一个基于卷积神经网络的特征提取网络。将图像特征提取之后,利用仿生模式识别构建并训练一个分类网络用于图像的分类。为证明方法的有效性,进行了3组对比实验,第1组为算法在少量数据下的对比分析,其改进后算法平均准确率比传统算法高了10%;第2组为算法在稍多数据下的对比分析,本算法平均准确率达到92%,高于传统算法;第3组为算法在较多数据下的对比分析,其平均准确率达到88%,高出传统算法10%。The paper proposes a combined convolution. The improved discriminant algorithm for neural network and bionic pattern recognition is based on bionic pattern recognition. Firstly, a feature extraction network based on convolutional neural network is constructed. After extracting image features,bionic pattern recognition is used to construct and train a classification network for the classification of images. In order to prove the validity of the method,three sets of comparative experiments are carried out. The first group is the comparative analysis of the algorithm under a small amount of data. The average accuracy of the improved algorithm is 10% higher than the traditional algorithm;the second group uses slightly more data in algorithm. Under the comparative analysis of the data,the average accuracy of the algorithm is 92%,which is higher than the traditional algorithm;the third group is the comparative analysis of the algorithm under more data,the average accuracy rate is 88%,which is 10% higher than the traditional algorithm.

关 键 词：仿生模式识别 卷积神经网络 肺炎判别 医学影像学 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]