Study On Blotch Detection And Restoration Algorithm For Digital Videos

Film archives were defected because film were aged and lost and scraped in the process of repeated playing, redeposit, and copying, so visual artifacts such as blotches, flickers, scratches, jitters, color fading are exist while playing. And artifacts such as black picture, green picture, mosaic and line-brace appear for some reasons in the process of program production today. All these defects are called video artifacts. This paper studies the most common defect exist in old film archives-blotch. This work study blotch detection and restoration in video defects through the comprehension use of digital image processing technology, such as image identification, image restoration and motion estimation and so on in move image sequences, provides feasible detection and restoration algorithm.On the basis of in-depth study and repeated experimenting to previous blotch detection algorithms, this work proposes an edge-based blotch detection algorithm. First, edges of current frame are extracted by edge detectors. Then, a non-blotch edge removal approach is proposed to remove partial non-blotch edge points according blotch edge characteristic. Finally, according to remaining edge pixels, all blotches are completely detected by the use of local motion compensation (MC)-based SROD blotch detection algorithm. In this edge-based blotch detection algorithm, blotch detection area is confined to blotch edge pixels and pixels nearby, greatly reduces time complexity. Non-blotch edge removal method further confines detection area to blotch edge pixels and pixels nearby further reduces time complexity. Being compared to original SROD algorithm, local motion compensation (MC)-based SROD blotch detection algorithm considers local motion, with higher motion compensation precision, higher detection accuracy and precision. Experimental results show that the proposed approach not only gives successful detection performances, but also provides the computational simplicity and outperforms classical algorithms. Edge-based blotch detection algorithm greatly improves the automation of blotch restoration and detection performance.On the basis of in-depth study and repeated experimenting to previous blotch restoration algorithms, especially image restoration algorithm based on non-local discrete regularization, this paper makes improvements to video restoration algorithm based on non-local discrete regularization, and proposes an improved blotch restoration algorithm. Improved algorithm is mainly doing post-processing to video restoration algorithm based on non-local discrete regularization. Being compared to original algorithm which uses similar pixels in current frame to reconstruct current pixel, this work also uses similar pixels in preceding and succeeding frames. Improved algorithm is not only superior to original algorithm in visual effect, but also in the evaluation of quality objective standard PSNR value and mean square error (MSE). Improved algorithm used in the restoration of real old film archives, which comprehensively combines time redundant information, provides great help to improve restoration effect.