Research On Image Zooming Algorithm Based On Stair Thinning

Digital image zooming, which finds important applications in many areas, is one of the important digital image processing techniques. Image zooming aims at increasing the resolutions of digital images. The basic method to zoom digital images is image interpolation. There are many interpolation algorithms such as nearest neighbor interpolation, Bilinear interpolation and so on. The two main problems in image zooming are that the zoomed images involve the well-known blurring distortion and Zigzagging artifacts in the edge areas.In view of the Zigzagging artifacts in image zooming, this thesis analyses the reasons for the artifacts and considers the edges of digital image as non-continuous and can be seen as containing one by one stairs. The main reason for Zigzagging artifacts is that the non-continuous edges will also be enlarged in image zooming. According to the above analyses, a hybrid algorithm for image zooming based on stair thinning is proposed to decrease the artifacts. The algorithm adopts Bilinear interpolation in smooth areas of digital images. And Canny egde detection algorithm is implemented. In the detected edge areas, stairs are detected and the horizontal and vertical edges are removed, since they produce no Zigzagging artifacts. In order to decrease Zigzagging artifacts, in the edge areas where contain many stairs, stair thinning interpolation algorithm is implemented to keep stairs from being enlarged.The experiment results and analyses show that when the thresholds are appropriate, to some degrees, this hybrid interpolation algorithm can remove more Zigzagging artifacts in image zooming.