Automatic Compensation of Nonlinear Distortions

When a signal goes through a static nonlinear device, it will be distorted. Typical examples are the nonlinearly distorted movie soundtracks. At old movie soundtracks the sound was recorded by the variable density method[1]. This method is very sensitive to the exposition and development processes. If the recording light was too strong, or the development was produced by wrong parameters, the sound can be extremely distorted. This distortion sometimes makes the sound incomprehensible. If the distortion is unacceptable, the distorted sound should be post processed. This post process is difficult, since the exact nonlinear function is not known and the distorted signal is superimposed by wide-band noise, which can be amplified during the restoration process. The first step is to identify the nonlinear function. After that we can restore the sound by a proper compensation function, which avoids the strong amplification of the noise. In this article, in Section II, we will show a method that can identify the parameters of the nonlinearity. In Section III a method will be shown, based on Tikhonov regularization, which can automatically find a proper compensation function.