摘要
针对传统图像处理技术存在的问题,比如压缩比例较低、图像重构精确度较差等,基于IPMI算法提出了一种基于信息瓶颈的图像模式处理算法,旨在利用相邻区域的重叠程度,尽可能地保留相邻区域之间的互信息,提高模式处理的质量。实验结果表明,IPMI算法可有效识别目标图像且复原图像的准确度高。
The traditional image processing technology has many problems, such as the lower compression ratio and the weaker accuracy of image reconstruction. The IPMI algorithm based on information bottlenecks improves tile quality of model processing by retaining mutual information between adjacent regions as possible. The simulation re- sult shows that IPMI algorithm can not only effectively identify the target image, but also has high accuracy of the restored image.
作者
韦佳佳
任海鹏
孙宇
WEI Jiajia;REN Haipeng;SUN Yu(Department of Information Engineering, Anhui Technical College of Mechanical and Electrical Engineering, Wuhu Anhui 241002, China)
出处
《重庆科技学院学报（自然科学版）》
CAS
2018年第3期57-59,共3页
Journal of Chongqing University of Science and Technology：Natural Sciences Edition
基金
安徽省高校自然科学研究重点项目"基于移动终端的智慧校园一站式服务平台的研究与实现"(KJ2016A133)
安徽省质量工程教学研究项目"基于虚拟现实的MOOC教学交互模式的研究与实践"(2016JYXM0194)
关键词
图像分割
互信息
阈值
区域
image segmentation
mutual information
threshoht
region
