一种改进的显著性区域提取模型  

Improved Model of Salient Regions Extraction

作　　者：于明[1] 邳艳芹[1] 

机构地区：[1]河北工业大学信息工程学院,天津300401

出　　处：《电视技术》2012年第19期167-169,192,共4页Video Engineering

基　　金：国家科技支撑计划项目(2009BAI71B02);河北省科技支撑项目(11213518D);河北省教育厅自然科学重点项目(ZD200911)

摘　　要：基于Itti模型,提出一种改进的模型来提取图像的显著区域,即分别提取原始图像的亮度、颜色、方向和形状4个低层特征,并且将这4个低层特征的多尺度显著图融合成一幅综合的显著图。在图像方向特征提取中采用倒数函数来代替Gabor特征函数,这样可完善方向特征图。该模型与经典的Itti模型相比较,提取出来的显著图更为明显,速度较快且易于实现。Image salient regions express the main content of image,Extraction of the regions plays an important role in searching image and compressing and identifying. Here a novel algorithm for salient region extraction is presented. First, four characteristics graphs of edging, brightness, color, the orientation of the original image is extracted respectively; then, multi-scale image features of the four characters are combined into one saliency map. In orientation feature extraction, reciprocal function is used to make the orientation feature map more complete than using Gabor function.The algorihm proposed in this paper is faster and easier to realize and can extract more obvious salient maps than Itti model.

关 键 词：显著区域 显著图 特征提取 视觉注意机制 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391.4[电子电信—信息与通信工程]