Motion blur detection based on lowest directional high-frequency energy

Publication Type:
Conference Proceeding
Proceedings - International Conference on Image Processing, ICIP, 2010, pp. 2533 - 2536
Issue Date:
Filename Description Size
Thumbnail2010000767OK.pdf1.15 MB
Adobe PDF
Full metadata record
Motion blur detection and the relevant blurring parameter estimation are important for many computer vision tasks. The contribution of this paper is in two folds. First, we propose a closed-form solution for motion direction estimation on blurred image. Secondly, a novel method is proposed for motion blurred region detection. The proposed direction estimation is based on measurement of lowest directional high-frequency energy. Compared with traditional methods, it will improve accuracy with less computational cost. Moreover, the proposed motion blurred region detection can efficiently estimate blurred regions without Point Spread Function estimation. Encouraging results are shown by experiments. © 2010 IEEE.
Please use this identifier to cite or link to this item: