Up-scaling with Edge Preservation Up-scaling with Edge Preservation Up-scaling with Edge Preservation

• Main Authors: Leon Lin, 林聖峰
• Other Authors: 蔡清欉
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/77438216683855514976

碩士 === 東海大學 === 資訊工程與科學系 === 94 === Image up-scaling is a significant technique in the area of image processing and a lot of image up-scaling algorithms have been proposed. However, in order to obtain a high quality scaled image, these algorithms often result in high operation cost and reduce the ability in real time process. On the other hand, algorithms with low operation cost are unable to make a high quality scaled image. The Winscale algorithm which proposed by Kim can adjust image resolution directly in its 2D pixel coordinates and obtain a not bad scaled image by a so-called “area pixel model”. But the quality of scaled image created by the new method is not better than that by the most commonly used Bicubic interpolation. To improve the Winscale algorithm, we introduced a new method, called “pixel- tagging”, to create the relationship between source image and its scaled image. Applying the method, we proposed an algorithm which can make the scaled image smooth and preserve the edge distribution through the “step-scaling” process. By using the algorithm, we can obtain higher quality and less operation cost than that by Bicubic interpolation with some images.