Video signal compression using prenonlinear filtering and a novel error correction technique

Abstract

In this paper, a nonlinear filter with a suitable transfer function is used at the front of the video compression transmitter. This filter reduces the range of relative amplitudes between the input sample amplitudes which results in increasing the discrete cosine transform (DCT) and discrete wavelet transform (DWT) abilities to concentrate the signal information in the lower frequency band and hence increasing the image compression ratio. In the receiver, an inverse filter is used after the inverse discrete cosine transform (IDCT) or inverse discrete wavelet transform (IDWT) to regain the signal information. The error due to neglecting high frequency bands was found to be approximately equal for each of the samples. Because the error is either positive or negative, using one bit of the quantization word will be very helpful to add the transmitted average error correctly at the receiving end with a very small negative effect on the quantization error due to the small difference between the sample amplitudes.