| With the rapid development of network multimedia,digital video is widely favored by people because of its rich content.The H.264 video standard is widely used in real life,while providing convenience for people's information interaction,it also brings many hidden dangers such as privacy leakage and copyright disputes,causing serious losses.Information hiding technology can hide the secret facts and play an important role in solving security problems.Therefore,it is of great significance to study the efficient and non-perceptual H.264 video information hiding algorithms.The compression effect of the H.264 standard is very superior,and the intra-frame prediction technology can significantly improve the compression ratio.Nowadays,the research on hiding algorithms based on this technology has achieved certain results,but there are still shortcomings: 1.There are limitations in the evaluation of distortion when hiding information;2.There are defects in the robustness of secret messages.This thesis mainly focuses on the H.264 intra-frame prediction mode to research the optimized video information hiding algorithm.The main work content includes the following aspects:(1)Aiming at the I-frame luminance 4 ?4 block prediction mode,this thesis proposes a robust concealment algorithm that optimizes distortion.The algorithm first establishes high-fidelity mapping rules based on edge features.Then,combining the matrix coding and LSBMR algorithm to modulate modes and embed,the recoding efficiency is higher,and the distortion can be reduced.In order to reduce the location limitation and expand the scope of comparison,the elements of the carrier domain are grouped to distorted the contrast during embedding.And the design of distortion evaluation integrates the texture features of LOCP,which can quantify the distortion value and make the evaluation more accurate.Finally,design BCH code encoding and long division decoding for secrets,which are used to quickly detect and correct random errors.And block Arnold scrambling transformation is performed on the coded secrets to chaotic correlation and limit error propagation.The algorithm can embed 5-bit data in a 16 ?16 macro block.Multiple sets of experimental data can show that the algorithm has advantages over single-mode modulation and parity modulation.The encrypted video quality obtained by this algorithm is better,the degree of distortion is lower,the bit rate increase is less,and the robustness of the encrypted message is also significantly enhanced.(2)In order to further optimize the distortion of the modulation mode and improve the robust performance,this thesis proposes a hiding algorithm based on the theory of minimizing distortion and the idea of convolutional codes.The algorithm first adopts Syndrome Trellis Codes to consider the distortion situation globally,and adds a distortion value for each carrier domain element,which significantly expands the distortion comparison range.The core is to design the distortion function to generate the check matrix.The distortion function refines the distortion value evaluation based on fuzzy entropy and edge direction,and fully considers the influence of embedding on adjacent blocks,so as to minimize the distortion when the bit rate is fixed.Then,the algorithm adopts convolutional code encoding and Viterbi decoding for the secret message.The convolutional code has memory characteristics,which can shorten the redundant code length and improve the efficiency.Finally,chaos scrambling is also used to eliminate relevance.The algorithm can control parameters to embed 4-bit or 8-bit data in a 16 ?16macro block.Multiple sets of experimental data can show that the algorithm has a certain degree of advancement.Compared with the first solution,the embedding capacity is increased significantly,and it is difficult to find changes visually,and when the same amount of data is embedded,the distortion is better.And the error correction capability of secrets has also been enhanced. |
PDF Full Text Request