High Fidelity Compression and Fast Reconstruction Technology for Film and Television Recording Based on Compressed Sensing Algorithm

With the widespread application of film and television recordings in the multimedia field, traditional audio compression methods cannot effectively balance compression rate and restoration quality under high fidelity requirements. To this end, this paper proposes a high-fidelity compression and fast reconstruction technology for film and television recordings based on the Orthogonal Matching Pursuit (OMP) algorithm. First, the compressed sensing theory is used to sparsely represent the film and television recording signal, and the OMP algorithm is used for efficient signal reconstruction. Then, in order to further improve the reconstruction speed, the OMP algorithm is improved, and an accelerated OMP (Accelerated Orthogonal Matching Pursuit, A-OMP) algorithm is introduced to accelerate the reconstruction process. Finally, combined with the adaptive dictionary learning technology, the compressed sensing dictionary can be dynamically adjusted according to the characteristics of different film and television recording signals. The results show that, under different compression rates, the A-OMP algorithm can significantly improve the sound quality fidelity compared with the traditional OMP algorithm. Especially when the compression rate is 16, the MSE (Mean Squared Error) value is reduced from 0.25 of the traditional OMP to 0.18, and the PESQ (Perceptual Evaluation of Speech Quality) score is increased from 2.8 to 3.0. In addition, compared with the fixed dictionary method, the adaptive dictionary learning method can effectively improve the sound quality restoration effect at a high compression rate. In the above data conclusions, the high-fidelity compression and fast reconstruction technology of film and television recordings based on the OMP algorithm proposed in this paper has shown good performance in sound quality restoration, computational efficiency and compression rate, and has high application value.