Image Retargeting Based on Sparse Multi-seam carving

• Main Author: 張隆陞
• Other Authors: 陳玲慧
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/g98a22

碩士 === 國立交通大學 === 多媒體工程研究所 === 105 === With the development of science and technology, the varieties of display device are more and more produced, such as smartphones, tablets, electronic billboards, and so on. The images should be adjusted to the adequate size before demonstrated on the screen. Seam carving is one of well-known methods. However, some objects look unnatural due to uneven or compact seams be deleted or inserted.To treat this disadvantage, sparse seams finding algorithm is proposed in this thesis. Besides, we propose a system which provides four retargeting methods, and these four methods must keep the shape of each important areas.Moreover, we also define three criteria based on image contents to suggest the suitable method.