基于多稳态特性与灰度差异熵的图像背景Python扩充方法  

Python Expansion of Image Background Based on Multi-Stability and Gray Difference Entropy

作　　者：汤恒[1] TANG Heng(School of Information and Artificial Intelligence,Wuhu Institute of Technology,Wuhu,Anhui 241006,China)

机构地区：[1]芜湖职业技术学院信息与人工智能学院,安徽芜湖241006

出　　处：《西昌学院学报（自然科学版）》2023年第2期81-85,122,共6页Journal of Xichang University(Natural Science Edition)

基　　金：安徽省教育厅线上课程(原MOOC)项目(2020mooc550)。

摘　　要：图像应用范围逐渐扩大,相关技术发展迅速,其发展的主要支撑是大规模的图像数据集合。现有公开资源以图像目标数据集合为主,而背景数据集合较少,并且内部数据体量较小,制约了图像相关技术的发展与应用,故提出基于多稳态特性与灰度差异熵的图像背景数据Python扩充方法。基于图像背景与目标的灰度熵差异,选取灰度熵阈值分割原始图像,获取图像背景区域,采用k近邻(k-Nearest Neighbor,KNN)算法深度挖掘背景数据,应用多稳定特性构造混沌序列,加密处理图像背景数据,通过Python制定图像背景数据并行扩充程序,执行制定程序即可实现图像背景数据的扩充。实验数据显示:提出方法获得的图像背景区域分割效果更好,图像背景数据扩充量最大值为1800MB,充分证实了提出方法应用性能更好。The application scope of image is gradually expanding,and related technologies are developing rapidly.The main support of its development is large-scale image data collection.The existing public resources are mainly image tar⁃get data sets,while the background data sets are few,and the internal data volume is small,which restricts the develop⁃ment and application of image related technologies.Therefore,a Python expansion method of image background based on the characteristics of multi-stability and gray difference entropy is proposed.Based on the gray entropy difference be⁃tween the image background and the target,the gray entropy threshold is selected to segment the original image;obtain the image background area;use the k-Nearest Neighbor(KNN)algorithm to deeply mine the background data;apply the multi-stability characteristics to construct chaotic sequences;encrypt and process the image background data;develop a parallel expansion program for image background data through Python;and implement the established program to achieve the expansion of image background data.Experimental data show that the proposed method achieves better image background region segmentation effect,and the maximum image background data expansion is 1800 MB,which fully confirms that the proposed method has better application performance.

关 键 词：图像背景数据 灰度差异熵 数据挖掘 PYTHON 多稳态特性 数据扩充 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]