Detail-preserving image information restoration guided by SVM based noise mapping
Digital Signal Processing
| Hi-index | 0.43 |
An adaptive method for suppressing scratch (impulsive) noise from audio signals is proposed. The algorithm has been developed primarily for the restoration of old phonograph records, but it can also be used for other applications as well. Impulses are suppressed by selectively applying variable length median filtering to contaminated signal regions only, thus minimizing distortion of clean passages and loss of high musical frequencies. The median window length is adaptively adjusted depending on the number of samples in each contaminated signal region. For off-line processing, the algorithm is simple enough to be implemented on a personal computer equipped with an inexpensive sound board. Experiments with old recordings produced restored signals of dramatically improved sound quality. The restored audio signal can be played through the sound board and stored onto an analog recording media such as high-quality cassette tapes