| Since most commercial imaging devices (digital still cameras and cell-phone cameras) are capable of capturing video sequences, the so-called print-from-video feature, which prints a still image from a user selected frame out of a sequence, has become common in commercial photo printers. One issue of concern to consumers when using print-from-video is the resulting low-quality print which arises due to the low resolution video captured by the devices (especially cell-phone cameras). To improve the quality of the final print, super-resolution techniques can be utilized. These techniques involve combining information from multiple low resolution images to generate a high resolution image. In this work, the implementation of a print-from-video method is carried out based on the shift-and-add super-resolution approach, which is the most suitable technique for embedded, real-time applications in terms of computational speed and accuracy. To address the outlier pixel problem inherent to the shift-and-add approach, a number of improvements including frame alignment, selection, and fusion are introduced. In addition, a post-processing step named pattern filtering is applied to reduce the visual artifacts in the printed images. The effectiveness of the developed solution is demonstrated by showing both quantitative and qualitative comparison results for video sequences captured by low-resolution cell-phone cameras. |
