Ghost removal method for image morphing using co-occurrence frequency image

Many image morphing methods require feature points meaning correspondence points between the input images by the hand. If user does not add feature points in detail, a noise called “ghost” appears in the output image. We propose a ghost removal method which extracts detailed feature points to modify a texture correspondence. The proposed method can remove the ghost at the textures having different gray values between the images.

[1]  George Wolberg,et al.  Image morphing: a survey , 1998, The Visual Computer.

[2]  Hajime Nagahara,et al.  Generation of high resolution video using morphing , 2005 .

[3]  Yasuyo Kita,et al.  Change detection using joint intensity histogram , 2006, 18th International Conference on Pattern Recognition (ICPR'06).

[4]  Dani Lischinski Incremental Delaunay Triangulation , 1994, Graphics Gems.

[5]  Hiroyasu Koshimizu,et al.  Proposals of Co-occurrence Frequency Image Based Filters , 2007, MVA.