A wavelet packet-based noise reduction algorithm of NTSC images using CVBS characteristics

This paper proposes a wavelet packet-based noise reduction algorithm for national television system committee (NTSC) images, in which the characteristics of a composite video burst signal (CVBS) are utilized. Most of conventional noise reduction algorithms apply spatial or spatio-temporal filtering only to the luminance signal, noting that the human eye is less sensitive to color than to luminance. Such noise reduction algorithms do not consider the realworld situation where a TV signal is transmitted over a noisy channel and decoded at a receiver. In this paper, it is assumed that an NTSVC signal is transmitted as a CVBS and corrupted with white Gaussian noise (WGN) by the channel. A CVBS has characteristics different from those of speech or image signals in a sense that encoded color information is modulated onto a high-frequency color subcarrier. The wavelet packet-based approach is suitable for noise reduction of the CVBS because decomposing a one-dimensional CVBS into eight subbands provides a chance to process each subband separately. In the proposed wavelet packet-based noise reduction algorithm, wavelet packet filtering is employed in subbands containing the color information whereas Wiener filtering is used in the other subbands. The separate treatment of each wavelet subband depending on the characteristics of CVBS leads to effective color and edge preserving noise reduction. The performance of the proposed method is validated by experiments with generated and corrupted CVBS images. Experimental results with various test images show that the proposed algorithm is effective in terms of the noise reduction efficiency and edge and color preservation.