摘要
通过对经典插值算法放大后得到的高分辨率图像先进行边缘检测,然后根据多方向边缘梯度变化,对边缘邻接像素点进行自相关模糊消除修正插值;多倍放大时采取小倍率循环修正方式实现。实验证明,算法克服了经典插值后图像的边缘锯齿和模糊效应,使修正优化后的图像清晰。对于RGB彩色图像先进行各分量色融合,再边缘检测,对提高图像质量也取得了明显的效果。
The proposed algorithm first detects edges in the high-resolution image obtained by classical interpolation algorithm, Then the edge＇s neighbors are amended by a self-relativity interpolation algorithm based on multi-direction gradient variety. And small ratio cycle amendatory method is used in multi-time enlargement. Experimental results show that the algorithm can effectively avoid blurting and artifacts edges and generate high quality images with less data. If this image is RGB, color fusion is processed first. Then image edges are detected. And the visual quality of the color images more prominent.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第15期3636-3639,共4页
Computer Engineering and Design
关键词
边缘检测
图像插值
图像放大
消除模糊
色融合
edge detection
image interpolation
enlargement
deblurring
color fusion
