基于背景约束机制的目标识别方法及图像语义抽取  

Method of object recognition based on background restraint mechanism and image semantic extraction

作　　者：陈文杰[1] 周海英[1] 

机构地区：[1]中北大学计算机与控制工程学院,山西太原030051

出　　处：《计算机工程与设计》2016年第4期993-998,共6页Computer Engineering and Design

基　　金：山西省自然科学基金项目(2013011017-6)

摘　　要：多数图像目标识别过程只对主要目标物进行提取,再分类识别,造成图像背景信息丢失,为此提出一种背景约束机制(background restraint mechanism)下的目标识别方法。通过视觉注意模型分别提取图像的前景目标物和背景信息,实现图像的前景目标物与背景分离,通过对背景图像信息的提取识别形成对前景目标物的概率约束。将此约束机制引入分类器中形成一种BRM_GAM(background-restraint-mechanism_Gaussian ARTMAP)分类模型,对前景目标物进行分类识别。实验结果表明,该方法有较好识别效率和时效性,符合人类认知。此外,提出一种利用GAM模型提取图像语义字典直方图,进行图像语义抽取的方法。Aiming at the problem of losing image background information,which is caused by only extracting the main object and then recognizing and classifying in processes of most object recognitions,an object recognition method based on background restraint mechanism was proposed.The foreground object and the background in the image were extracted and separated through visual attention model,and the background of image was recognized to form the probability restraint of foreground object.The object was recognized and classified using the BRM-GAM（background-restraint-mechanism_Gaussian ARTMAP）model which formed by introducing the restraint mechanism into the classifier.Experimental results show that the proposed method has better recognition accuracy,timeliness and it compliances with human cognition.In addition,a method which extracted semantic dictionary histogram using GAM model to extract image semantic was also proposed.

关 键 词：目标识别 背景信息 视觉注意模型 背景约束机制 图像语义 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]