Abstract—We knew that image rotation will result in jagged edge and image blurry, especially for dominant structure images. Here we present a novel content-aware image rotation approach based on gradient map: first, compute the gradient map for the image which describes object structure of the image. Second, take gradient map into consideration, use our proposed weighting bilinear interpolation method to rotate image that combine the gradient value and the distance together to regulate interpolation when the pixel distribute in the sharpness region. Besides, a coefficient was used to eliminate the blurry caused by average operation on local areas. Compared to classic methods, our method is faster and will achieve better result in rotating images.
Index Terms—Image rotation, gradient map, weighting bilinear interpolation, content-aware.
The authors are with the Dep. of Computer Science and Technology, East China Normal University (e-mail: email@example.com, firstname.lastname@example.org).
Cite: Fenghui Niu and Haisheng Li, "Content-Aware Image Rotation," International Journal of Future Computer and Communication vol. 2, no. 1 pp. 1-3, 2013.