Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14] [ 15] [ 16] [ 17] [ 18] [ 19] [ 20] [ 21] [ 22] [ 23]

@

Journal of Information Science and Engineering, Vol. 26 No. 5, pp. 1741-1753 (September 2010)

Spike Defect Detection and Removal on Aged Films*

RONG-CHI CHANG AND TIMOTHY K. SHIH+
Department of Digital Media Design
Asia University
Taichung, 413 Taiwan
+Department of Computer Science and Information Engineering
National Central University
Taoyuan, 320 Taiwan
E-mail: roger@asia.edu.tw; +timothykshih@gmail.com

Aged films contain valuable historical information. However, some films may have defects due to dirt, scratches or for other reasons. Usually, there is no efficient prediction model to precisely identify these defects. This study proposes a set of new algorithms based on the analysis of features among frames in aged films. Both spatial and temporal characteristics are used. The algorithms further use filters with adjustable thresholds to precisely identify defects, such as spikes and scratches. Finally, a temporal inpainting method and a multiple resolution image inpainting algorithm are used to repair the defects. Since no efficient automatic quantitative mechanism is available due to the loss of original films, the experiments use subjective evaluation on more than ten films. The results show that most defects are repaired with a pleasant visual quality.

Keywords: video inpainting, defect detection, motion estimation, image completion, inpainting

Full Text () Retrieve PDF document (201009_11.pdf)

Received September 11, 2008; revised November 25, 2008, July 3 & October 13, 2009; accepted October 29, 2009.
Communicated by Tong-Yee Lee.
* This paper was partially supported by the Asia University of Taiwan, R.O.C. under contract No. 97-D-05.