Video resolution enhancement by using discrete and stationary wavelet transforms with illumination compensation

Abstract

This paper proposes a new video resolution enhancement technique, in which a state-of-the-art illumination compensation procedure is applied to the respective frames before the registration process. After illumination compensation process, the respective frames are registered by using Vandewalle technique. In parallel, the corresponding frame is decomposed into its frequency subbands by using discrete wavelet transform (DWT) and stationary wavelet transform (SWT). Furthermore, the high-frequency subbands (LH, HL, and HH) have been super resolved by using Vandewalle super resolution technique. Afterward, the super resolved high-frequency subbands are being enhanced by the ones obtained through SWT as the latter ones contain more information. The enhanced high-frequency subbands and the output of the registration technique, which is regarded as the low-frequency subband, have been combined by using inverse DWT (IDWT) in order to construct the high-resolution frame. The quantitative (PSNR) results show the superiority of the proposed technique over the conventional and state-of-the-art video resolution enhancement techniques, in which for Akiyo video sequence there are 5.45 dB improvements over the average PSNR compared to Vandewalle registration technique with structure adaptive normalized convolution.