基于卷积神经网络的分形图像编码研究  

Research on Fractal Image Encoding Based on Convolutional Neural Network

作　　者：许才顼 贺杰[1,2] 庞家豪 XU Cai-xu;HE Jie;PANG Jia-hao(Guangxi Key Laboratory of Machine Vision and Intelligent Control,Wuzhou University,Wuzhou Guangxi 543002,China;College of Computer Science and Electronic Engineering,Hunan University,Hunan Changsha 410082,China;Faculty of Innovation Engineering,Macao University of Science and Technology,Macao 999078,China)

机构地区：[1]梧州学院广西机器视觉与智能控制重点实验室,广西梧州543002 [2]湖南大学信息科学与工程学院,湖南长沙410082 [3]澳门科技大学创新工程学院,中国澳门999078

出　　处：《计算机仿真》2024年第8期210-214,共5页Computer Simulation

基　　金：国家自然科学基金地区基金(61961036,62162054);广西高校中青年教师基础能力提升项目(2021KY0673);广西自然科学基金项目(2020JJA170007);梧州学院校级青年基金项目(2020C001)。

摘　　要：为了解决图像压缩时间较长的问题,提出了一种基于卷积神经网络的分形图像编码方法。方法将分形图像编码中的Range块和Domain块进行相似度分类,通过采用卷积神经网络来设计新的图像块分类方法,实现编码过程中的并行化图像块分类,编码时每个类在各自类别中进行搜索,减少一些不必要的匹配过程,以此达到减少图像块匹配的计算量,减少图像压缩的时间。实验结果表明,上述方法的图像还原质量更高,且图像压缩编码时间较短,说明其具有较高的应用价值。To solve the problem that the image compression time is too long,this paper presented a fractal image encoding method based on convolutional neural network.In this method,we classified the similarities of Range blocks and Domain blocks in fractal image encoding at first,and then used the convolutional neural network to design a new image block classification method,thus achieving the parallel image block classification.Each category searched within its own category during the encoding process,thus reducing some unnecessary matching processes.Finally,the computational complexity of image block matching and the time of image compression were reduced.The experimental results show that the method has higher image restoration quality and shorter compression encoding time,so its application value is high.

关 键 词：分形图像压缩 卷积神经网络 图像块匹配 相似度分类 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]