Super-resolution reconstruction of image sequences compressed with DWT-based techniques
Xiang Ping Zhu, Jie Yuan, Sidan Du · 2007
Spatial domain based super-resolution reconstruction techniques, which were designed for uncompressed video to produce high-resolution image or image sequences, may not work well when applied directly to compressed videos, especially to those with severe quantization errors. Here a reconstruction approach designed for videos or image sequences, which were compressed using DWT-based techniques, is presented. This method utilizes the theory of projection onto convex sets (POCS) with a new projection operator based on Discrete Wavelet Transformation (DWT) for reducing blurring artifacts. It also applies maximum a posteriori (MAP) estimation techniques to remove ringing noises from restored images. Experimental results show that such approach could effectively recover objects’ edges and details that were blurred during compressing process while, expanding image’s size simultaneously.