Seam Carving和显著性分析的图像缩放方法研究  

Image resizing based on combination of Seam Carving and salient region analysis

作　　者：王玺[1,2] 夏清国[1] 窦召虎 李群组[1] 

机构地区：[1]西北工业大学计算机学院 [2]解放军94153部队

出　　处：《计算机工程与应用》2011年第30期168-171,197,共5页Computer Engineering and Applications

摘　　要：提出一种基于Seam Carving和显著性分析相结合的图像缩放方法。通过不断移除(插入)缝合线来实现对图像尺寸的缩小(放大)。通过研究Seam Carving在图像缩放及目标保留移除中存在的问题,已有的基于像素点计算图像能量函数的方法会对某些图像失效,提出一种基于显著性区域求解图像能量函数的SDRS方法。实验结果表明,Seam Carving方法对自然图像的处理能够达到令人满意的效果。Seam Carving方法和SDRS方法的结合,也是一个创新点。A new image resizing based on Seam Carving and salient region analysis is presented.By repeatedly carving out or inserting seams in different directions the aspect ratio of an image can be changed.The original Seam Carving method fails when it meets some special images on resizing,object reserving,and object removing.With the analysis of the method called Seam Carving and the definition of the energy function,another method Salient Region Detection and Segmentation（SRDS） is introduced into the paper.The experiments on real image show that Seam Carving is acceptable in nature image processing.The combine of the Seam Carving and the SRDS is an innovation.

关 键 词：SEAM CARVING 尺寸缩放 能量函数 显著区域检测和分割(SRDS) 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]