基于亮度感知的前景提取  

Foreground extraction based on the brightness perception

作　　者：彭橦 何坤[2] 周激流[2] PENG Tong;HE Kun;ZHOU Ji-Liu(College of Electronic Information,Sichuan University,Chengdu 610065,China;College of Computer Science,Sichuan University,Chengdu 610065,China)

机构地区：[1]四川大学电子信息学院,成都610065 [2]四川大学计算机学院,成都610065

出　　处：《四川大学学报（自然科学版）》2021年第5期39-45,共7页Journal of Sichuan University(Natural Science Edition)

基　　金：国家重点研发计划(2018YFC0832301)。

摘　　要：前景提取是在图像整体认知基础上将感兴趣对象分离出来,本文联合图像亮度视觉感知和水平集方法提出了一种基于亮度感知的前景提取模型.该模型依据像素对的亮度视觉相关性,联合视觉区域内的相似性和区域间的差异性,设计了亮度感知能量泛函,运用瑞利熵求解能量泛函得到视觉区域,利用视觉区域特征驱使初始曲线演化至前景轮廓.相对于传统算法,该模型运用图像视觉特征有利于从图像的整体认知上提取前景,提高了水平集方法的前景提取质量.Foreground extraction is to separate the objects of interest based on the whole image cognition.In this paper,a foreground extraction model based on brightness perception is proposed by combining the visual brightness perception and level set method.Based on the brightness'visual correlation of pixel pairs,combining the similarity within the visual region and the difference among regions,the brightness perception energy functional is designed.The visual region is obtained by optimizing the energy function using Rayleigh entropy,and the initial curve evolves to the foreground contour by using the characteristics of the visual region.Compared with the traditional algorithm,the proposed model uses the image visual features to extract the foreground from the overall cognition of the image and improves the foreground extraction quality of the level set method.

关 键 词：前景提取 亮度感知 视觉区域 水平集方法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]