Resolution enhancement of video sequences by using discrete wavelet transform and illumination compensation

Abstract

This research paper proposes a new technique for video resolution enhancement that employees an illumination compensation procedure before the registration process. After the illumination compensation process, the respective frames are registered using the Irani and Peleg technique. In parallel, the corresponding frame is decomposed into high-frequency (low-high, high-low, and high-high) and low-frequency (low-low) subbands using discrete wavelet transform (DWT). The high-frequency subbands are superresolved using bicubic interpolation. Afterwards, the interpolated high-frequency subbands and superresolved low-frequency subband obtained by registration are used to construct the high-resolution frame using inverse DWT. The superiority of the proposed resolution enhancement method over well-known video superresolution techniques is shown with quantitative experimental results. For the Akiyo video sequence, there are improvements of 2.26 dB when compared to the average peak signal-to-noise ratio obtained by the state-of-the-art resolution technique proposed by Vandewalle.