一种显著性检测及其自适应传输控制算法研究  

Research on saliency detection and adaptive transmission control algorithm

作　　者：刘文韬 李华 LIU Wen-tao;LI Hua(Zhongshan Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Zhongshan 528400,Guangdong Province,China)

机构地区：[1]广东电网有限责任公司中山供电局,广东中山528400

出　　处：《信息技术》2022年第12期189-194,共6页Information Technology

摘　　要：针对图像目标提出了基于离散余弦变换域的显著性检测算法,并在此基础上通过综合考虑图像空间特性,构建了基于空间分布和空间频率的算法模型,并通过自适应权值与高斯差分,最终整合了空间分布与空间频率的显著性。结果显示,显著性检测的ROC曲线下面积达到0.7745,较之于传统显著性检测优势明显;与此同时,在PR曲线比较上,该显著性检测及其自适应传输控制算法的准确率值达到90.82%且召回率值达到77.40%,同样具有显著优势。该研究提出的算法既在目标分割上较为准确,又在人工操作上较为简易。A saliency detection algorithm based on discrete cosine transform domain is proposed for image targets.On this basis,by comprehensively considering the spatial characteristics of images,an algorithm model based on spatial distribution and spatial frequency is constructed.Finally,the saliency of spatial distribution and spatial frequency is integrated through adaptive weight and Gaussian difference.The results show that the area under the ROC curve of saliency detection is 0.7745,which has obvious advantages over the traditional saliency detection.At the same time,compared with PR curves,the accuracy of the significance detection and its adaptive transmission control algorithm reaches 90.82%and the recall rate reaches 77.40%,also showing strong advantages.In short,the algorithm proposed in this study is not only more accurate in target segmentation,but also simpler in manual operation.

关 键 词：目标分割 显著性检测 自适应权值 空间分布 空间频率 

分 类 号：TP306[自动化与计算机技术—计算机系统结构]