MASTER THESIS Video Watermarking - Computer Graphics Group ...

More documents

Recommendations

Info

3.3 Decoding The decoding process is reversal process to encoding resulting in visual video data, as depicted in Figure 7. Incoming slices are decoded, using the same entropy coding as in the encoding process, up to intra prediction modes or motion vectors and quantized transform coefficients. Macroblock by macroblock, block by block, the quantized transform coefficients are scaled to the former range, i.e. multiplied by dequantization factors, and transformed by inverse frequency transform. Hereby, the prediction residual is obtained. Figure 7: Decoding process scheme The prediction process is invoked using the intra prediction mode in case of intra prediction, or the motion vector in case of inter prediction. Predicted samples are added to the residual. Such decoded blocks and macroblocks are joined together to form the visible picture that is stored in the buffer of reference pictures for the inter prediction process in next pictures. In both encoding and decoding processes, deblocking filter process is invoked over decoded pictures to increase final visual quality. The process eliminates blocking artefacts on block borders, as may be seen in video sequences compressed according to many of previous video coding standards at lower bit-rates. 18
Chapter 4 Implementation Details This chapter deals with implementation details of the software framework for watermark embedding and detection. A partial decoder / encoder (codec), as mentioned in Section 2.2, of H.264 video streams has been implemented in order to obtain transform coefficients for direct watermark embedding in frequency domain and further processing for embedding in spatial domain. Further, a generic framework for watermarking of H.264 streams in both spatial and frequency domains has been designed and implemented. In this framework, three different watermarking methods have been written. The practical comparison of these methods is presented in Chapter 5. 4.1 The H.264 Codec The partially decoding part of the codec is implemented according to the H.264 standard [9]. Of course, there are free implementations of the standard but writing an own helps to deeply understand the standard and video compression at all. Moreover, the transform coefficients are needed only, not fully decoded visible pictures. The decoder produces all syntactical elements of a stream which are relevant for watermarking. In particular, it decodes and parses SPSs, PPSs, slices and macroblocks up to transform coefficients. During the process, it checks whether the decoded elements and the whole stream are correct. When an error occurs, it is reported by a message of particular severity level (critical error, error, warning etc.) and decoding of the current NAL unit ends immediately. The encoding part of the codec is written as an inverse process to decoding because the standard contains only fragment information about the process. The 19
Page 1 and 2: Charles University in Prague Facult
Page 3 and 4: Table of Contents Abstract 5 1 Intr
Page 5 and 6: Název práce: Watermarking videa A
Page 7 and 8: information about copyright owner a
Page 9 and 10: Common geometric distortions - the
Page 11 and 12: watermark, i.e. values of the water
Page 13 and 14: Chapter 3 The H.264 Standard The H.
Page 15 and 16: There are three types of slices: I,
Page 17: frequency transform. The choice is
Page 21 and 22: auxiliary coded pictures), Supplem
Page 23 and 24: an intra coded slice, content ID -
Page 25 and 26: samples is subtracted from the resi
Page 27 and 28: At the beginning of the entire proc
Page 29 and 30: file in order to provide detailed r
Page 31 and 32: as a part of the whole process. The
Page 33 and 34: coefficients are hardly set to non-
Page 35 and 36: Table 1 outlines characteristics of
Page 37 and 38: The third and the fourth row sets c
Page 39 and 40: Using block method, we cannot consi
Page 41 and 42: probabilities of watermark detectio
Page 43 and 44: 5.4.3 Cropping In the cropping test
Page 45 and 46: 5.4.6 Blurring Blurring can be cons
Page 47 and 48: 5.4.8 Multiple Watermark Embedding
Page 49 and 50: Despite the expectations, all the m
Page 51 and 52: ecompression and noising, and to so
Page 53 and 54: Bibliography [1] Ingemar J. Cox, Jo
Page 55 and 56: Appendix A Enclosed CD & Installati
Page 57 and 58: Appendix B The plugin usage The wat
Page 59: length is the length of the list of

MASTER THESIS Video Watermarking - Computer Graphics Group ...

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?