Study Of Old Film Restoration Algorithms And System Design

With the development of digital technology, it has become an effective meansfor protecting film archives by scanning and saving them as digital format. However,these film archives usually have many defects due to storage conditions andfrequently playing, such as blotches, scratches, noises, and film unsteadiness. Underthis condition, using the advanced digital image technology to detect and restoredefects in old films has become essential.Old film restoration consists of two processes: defects detection and defectsrestoration. Old films usually contain different types of defects, this thesis focused onblotch detection and blotch removal algorithms. Besides that, different digital imagerestoration algorithms were integrated into a man-machine interactive imagerestoration system to improve the practicability of the image completion algorithms.Firstly, this thesis presented the research background and significance ofblotch detection and restoration algorithm for the old films, and summarized theresearch status of different algorithms. Then the thesis deeply discussed thetraditional blotch detection algorithms, and on this basis proposed an adaptivethreshold blotch detection method, which can get the optimal threshold with iterativethreshold and statistical properties between each frame, so the algorithm’s adaptivitywas improved. The simulation results show that blotches detection effect by thismethod is more accurate compared with the traditional algorithms, as well as theadaptivity.Secondly, the thesis briefly described traditional blotch removal algorithmwhich is based on interpolation. Then the thesis elaborated PDE image inpaintingalgorithm which is for restoring small-scale damaged area, and the texture synthesisalgorithm which is for large-area damage. The simulation results show that PDEalgorithm can realize perfect effect for images with small-scale damaged area orscratches, and when the damaged area is large or contains texture information, thetexture synthesis algorithm can achieve better effect.In the end, the thesis described an image restoration system, which integrateddifferent image completion algorithms and scratch detection algorithm. By the way of human-computer interaction, we can select appropriate restoration algorithm fordifferent damaged images, which improved the practicability of the algorithm.Especially for the images with structure information, it automatically realized edgesfitting and structure propagation by manual intervention. The simulation results showthat the system can achieve the optimal effect under conditions of less humanintervention.