摘要
针对当前检测体育场监控图像DCT域时,出现检测结果边缘模糊,显著性差的问题,提出一种新型的DCT域视觉显著性检测模型。通过构建图像库结构,获取满足检测需求的监控图像,采用灰度值标注的方式,实现DCT组织域的应用与转换,搭建总体检测框架,根据图像特征的分割与提取原则,计算具体的显著性特征权值,完成场馆监控图像DCT域视觉显著性检测模型的建立。仿真结果表明,与传统人类视觉监控系统相比,所提方法提高了图像区块显著值,检测结果图像清晰边缘锐利且图像DCT域视觉显著性较好,具有较强的应用性。
In this paper,a model of detecting DCT domain visual saliency was designed.By constructing the structure of image library,we obtained the monitoring images that met the detection needs.And then,we used the gray annotation method to implement the application and conversion for the DCT domain.Moreover,we built the o-verall detection framework.Based on the principle of segmentation and extraction of image features,we calculated specific weights of saliency features and thus to construct the model of visual saliency detection in DCT domain of venue monitoring image.Compared with the traditional human visual monitoring system,the proposed method im-proves the significant value of image block.According to the detection result,the image has clear edges and sharp-ness.Meanwhile,the visual saliency is good in image DCT domain.
作者
索岩
崔智勇
SUO Yan;CUI Zhi-yong(Xinlian College,Henan Normal University,Xinxiang Henan 453000,China;College of Computer and Information Engineering,Henan Normal University,Xinxiang Henan 453000,China)
出处
《计算机仿真》
北大核心
2020年第12期421-425,共5页
Computer Simulation
基金
河南省软科学研究计划项目(142400411151)。
关键词
图像灰度值
特征权值
模块化区域
区块显著值
Image gray value
Feature weight
Modularized area
Significant value of block
