摘要
由于图像分类标准的可靠性较低,导致在具体的分类阶段,错误分类的情况较为普遍,提出基于深度字典学习的图像分类系统设计研究。结合图像分类的实际计算需求,在硬件构架中设置了3个数字低压差线性稳压器(Low Dropout Regulator,LDO)和旁路调节场效应晶体管(Field Effect Transistor,FET)结构,并将ET200SP的SIMATIC ET 200SP模块作为系统主体构架,从而实现图像分类标准输出模块和字典输出模块的集中控制。在软件运行逻辑的设计上,构建了具有分层特征的学习网络结构,分析得到图像稀疏度字典库,将其作为图像分类的标准,实现对图像的分类处理。测试结果表明,设计系统可以实现对图像的准确分类。
Due to the low reliability of image classification criteria,which leads to more common cases of misclassification in the specific classification stage,the design study of image classification system based on deep dictionary learning is proposed.Combined with the actual computational requirements of image classification,three digital LowDropout Regulator(LDO)and bypass regulation Field Effect Transistor(FET)structures are set in the hardware architecture,and the SIMATIC ET 200SP module of ET200SP is used as the main system architecture,so as to realize image classification.The standard output module and the dictionary output module are centrally controlled.In the design of software operation logic,a learning network structure with hierarchical features is constructed,and the image sparsity dictionary library is analyzed and obtained,which is used as the criteria for image classification to realize the classification processing of images.The test results show that the designed system can realize the accurate classification of images.
作者
毛颖颖
MAO Yingying(School of Information Engineering,Henan Polytechnic,Henan ZhengZhou 450046,China)
出处
《信息与电脑》
2022年第21期157-159,共3页
Information & Computer
基金
河南省高等学校重点科研项目“基于PHOG的目标分类深度字典学习方法研究”(项目编号:22B520017)。
关键词
深度字典学习
图像分类
分层特征
deep dictionary learning
image classification
hierarchical features
