Towards Practical Print-from-video: Improved Pattern Filtering

Print-from-video can be achieved by using super-resolution (SR) techniques. These techniques involve combining information from multiple low resolution images to generate and print a high resolution image. Among the existing SR techniques, the most suitable technique for deployment on portable consumer electronics products having limited processing and memory resources is the shift-and-add method. However, this technique is known to generate artifacts when applied to real video sequences. This paper introduces a number of improvements made to the shift-and-add technique to reduce the level of artifacts so that it can be deployed on portable camera devices. These improvements, named improved pattern filtering, consist of three computationally efficient steps of frame alignment, frame selection, and frame fusion. In these steps, motion vector field and DCT coefficient information are derived from an MPEG codec and utilized to improve the quality of printed images. The effectiveness of the developed solution is demonstrated by showing both quantitative and qualitative comparison results.