结合边缘模糊的景深显著性在图像缩放中的研究  

Image Retargeting Using Blur Based Depth Saliency Descriptor

作　　者：岳朋飞 王化雨[1,2] 郑元杰[1,2,3] 赵彦会 崔嘉 Yue Pengfei;Wang Huayu;Zheng Yuanjie;Zhao Yanhui;Cui Jia(Institute of Information Science and Engineering,Shandong Normal University,Ji’nan 250358;Key Laboratory of Intelligent Information Processing,Shandong Normal University,Ji’nan 250358;Institute of Life Sciences,Shandong Normal University,Ji’nan 250358)

机构地区：[1]山东师范大学信息科学与工程学院,济南250358 [2]山东师范大学智能信息计算与安全实验室,济南250358 [3]山东师范大学生命科学学院,济南250358

出　　处：《计算机辅助设计与图形学学报》2018年第3期415-423,共9页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(61502285);山东省青年基金(ZR2014FQ013)

摘　　要：针对内容敏感图像缩放技术调整图像尺寸时会出现重要信息丢失、物体边缘扭曲、非显著区域结构信息不完整等问题,提出基于边缘模糊的景深显著性算法.根据心理学原理,引入非显著区域结构信息保持的概念.采用边缘模糊特性进行景深估算,通过学习得到模糊字典提出模糊景深描述子,并以模糊景深描述子辅助进行显著性计算;为了达到在图像缩放时保持结构均衡,通过改进传统的区域型显著计算,提出显著性边缘分布特征,能够在保持显著信息、边缘完整的同时,有效减少结构信息的丢失或损坏,使缩放后的图像更符合人们的视觉体验.通过与经典算法和最新算法的对比实验,验证了文中算法的有效性.Image retargeting technique becomes research hot with the widely used digital images on different display screens,such as TV,iPhone,iPad and other wearable equipment.The objectives of image retargeting are important information preservation,less edge distortion during increasing/decreasing image size.The major existed content-aware methods perform well.However,there are two problems should be improved:the slight distortion appeared at the object edges and the structure distortion in the non-salient area.In order to solve the problems above,we propose a novel algorithm to calculate depth-based saliency based on the edge blur.According to psychological theories,people evaluate image quality based on multi-level judgments and comparison between different areas,both image content and structure.The paper proposes a new standard:the structure preserving in non-salient area.After observation and image analysis,blur(slight blur)is generally existed at the edge of objects.The blur feature is used to estimate the depth cue,named blur depth descriptor.It can be used in the process of saliency computation for balanced image retargeting results.In order to keep the structure information in non-salient area,the salient edge map is presented in Seam Carving process,instead of field-based saliency computation.The derivative blur-based depth from x-and y-direction can avoid the redundant energy seam around salient objects causing structure distortion.There are three comparison experiments between the proposed algorithm and selected classic algorithms:salient edge preservation experiment,large scale retargeting on high-resolution images experiment and average energy comparison.The promising results prove the feasibility of the proposed algorithm.

关 键 词：图像缩放 SEAM CARVING 显著性计算 模糊景深描述子 结构信息 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]