基于边缘检测的图像缩放算法  

Image Scaling Algorithm Based on Edge Detection

作　　者：王效灵[1] 汪颖[1] 陈涛[1] 顾伟康[1] 

机构地区：[1]浙江大学信息与电子工程学系,杭州310027

出　　处：《科技通报》2005年第5期584-588,共5页Bulletin of Science and Technology

基　　金：国家计委产业化专项基金资助(计高技[2001]2462号)

摘　　要：论文提出了一种保持边缘特征的图像缩放算法,算法首先检测图像中物体的边沿,并在传统边沿检测方法的基础上增加“不确定度”这个参数来提高边沿检测的鲁棒性;对边缘区域像素的插值沿边缘方向进行,对非边缘区域的像素的插值用三次卷积的方法生成。该算法能有效消除传统插值算法在物体边缘部分出现的锯齿,提高了缩放后图像的视觉质量。仿真结果表明,对人造图像和自然图像,本算法均能保持良好的边缘特征,取得较理想的图像缩放效果。This paper proposes an edge preserving scaling algorithm for digital images. The algorithm first detects edges in the picture, the boundary pixels interpolate along edge directions while pixels in fiat areas use Cubic Convolution interpolation kernel. Based on the traditional edge detecting algorithm the proposed method adds a control parameter named “ambiguity” to improve the robustness of the edge detection. The algorithm can effectively avoid jaggy edges and improve image quality. Simulation studies show that the proposed algorithm can preserve edges and not introducing spurious artifacts.

关 键 词：图像缩放 边沿检测 不确定度 三次卷积 

分 类 号：TN949.198[电子电信—信号与信息处理] TN919.8[电子电信—信息与通信工程]