Anil C. Kokaram
Motion Picture Restoration
Digital Algoritms for artefact suppression in degradated motion picture film and video
Springer Verlag
1998
ISBN 3-540-76040-7
Springer Verlag
1998
ISBN 3-540-76040-7
Manual de restauro automático de filme e vídeo mais reconhecido no meio académico a nível mundial
This book looks at the design of automatic motion picture restoration tools that can be used for enhancing archive material, and reducing the storage requirements necessary for acceptable viewing quality so that data can be recorded on relatively low-cost media. An increasing amount of research is being carried out in the area of visual manipulation, aspects of which include Motion Estimation, Data Reconstruction and Human Visual Perception. Motion Estimation is relevant to everything in video processing, from reconstruction to robotics. Research is also being carried out in the area of post processing decoded video streams to improve the image quality of the film. This research also involves reconstruction. This book addresses both Motion Estimation and Reconstruction, and as such will be of great interest to researchers in both academia and industry.
Inclui CD-ROM
A PORTION OF THE CD-ROM SUPPLEMENT FOR
"MOTION PICTURE RESTORATION"
Anil C. Kokaram
(Presented on the WWW by permission of Springer Verlag)
- Chapter 3: The Juggler sequence is in the 3DAR subdirectory, 64 images of 256x256.
- Chapter 6: Heuristics for Reconstructing Missing Data
- Chapter 8: Large Area Reconstruction
- Chapter 10: Noise Reduction for Image Sequences
The manipulation of pictures and video in digital form has been an established research activity for more than twenty years. It is only recently, however, that digital image and video processing equipment has been accessible to the general public. This is due in part to the rapidly growing economy of the home computer. A major contributing factor has been the marked rise in the presence of the non-academic user on the internet, particularly due to the World Wide Web (WWW). Manipulating digital imagery has become synonymous with the WWW. It is the drive to present audio and visual media to the home user in an interactive form and to increase the available range of choices, which has encouraged agreements to begin digital video television broadcasting before the turn of the century. With the increased demand for video material, there is a perceived increase in demand for material from archive sources and this has fuelled commercial interest in automatic digital restoration processes. Furthermore there is a continuing effort to design techniques for correcting errors in received compressed video bit streams for the purposes of live communications links over noisy channels e.g. mobile telephones and the internet.
The book introduces the reader to a range of digital restoration activities beyond the well traversed areas of noise reduction and deblurring. It describes a number of problems associated with archived film and video. Particular attention is given to missing data since that applies equally well to correcting packet loss in coded video bit streams. The book should appeal both to those in industry and to academic researchers.
The purpose of this CD-ROM is to give the reader a better feel for the issues addressed in the book. There are subdirectories corresponding to the Chapters 5 to 11, which contain the more useful visual information in the book. In each subdirectory is an HTML document of the same name which dscribes the content of the directory and also presents some additional remarks on restoration quality. In particular, the document on Line Removal contains information not discussed in the book since the observations need to be backed up with a clearer visual presentation.
I would like to acknowledge the assistance of Jean-Hugues Chenot of INA, Paris, France in acquiring permission to use the FACE , CAR and GREENHOUSE image sequences and João Sequeira of RTP, Lisbon, Portugal for supplying the SITDOWN , STAR and PULL sequences.
The HTML documents were arranged to have optimal impact when the viewing screen is of size 1024 x 768.
The reader is encouraged to switch to that or higher resolutions if possible. In assembling a document like this, there is always the risk that the displayed artefacts/restorations do not have the impact they should have because of viewing area size as well as quality of image reproduction inside the browser. Raw image data is therefore included for viewing on the reader's own equipment. In particular, because of the JPEG format used for image display inside Netscape (which was used to test the document), the noise reduction examples are not reproduced very well and the reader is encouraged to resort to viewing the .PGM files included. Furthermore, in many cases no attempt was made to implement algorithms so that they have useful behaviour at the extremities of each image frame, the reader is asked to ignore any artefacts in these areas.
In all cases, raw image data is stored as 8 bits per pixel in a binary file which contains the rowwise scanned image from left to right and top to bottom. For image sequences each such frame of data is consecutively written to a file which therefore consists of all the raw image data for all the frames. Colour sequences are stored in planes of Red, Green and Blue data consecutively written to the files. The size of each image frame is given as part of the description contained in the HTML documents.
Links to the documents are listed below. (In the full CD, all chapters have supporting material.)
Muito Bom estado. Como novo (nunca foi manuseado)
Sem comentários:
Enviar um comentário