A stereoscopic shadowgraph system has been developed based on the conventional z-type schlieren configuration. The test volume is set at the intersection of two inclined converging beams formed by two pairs of parabolic mirrors. Two synchronized high speed cameras are applied to record the shadowgraph image pairs simultaneously. A precisely etched metal mesh plate is used to calibrate the stereoscopic shadowgraph system. By combining the calibration parameters and coordinates of the matching points in stereo images, the depth information and the 3D view of the shadowgraph images are obtained. A crystal block with internal 3D images created by laser etching is used as a model for static object reconstruction and validation. The 3D coordinates obtained by the digital 3D reconstruction are in good agreement with the real dimensions. The developed stereoscopic technique is then applied to investigate the bursting dynamics of a bubble. The time resolved bubble-bursting process has been reconstructed successfully. The quantitative velocity measurement reveals that the bubble collapses at a constant velocity of around 7 m s−1, which corresponds to a high Weber number. As a result, finger-like structures are observable around the rim of the collapsing bubble. The stereoscopic shadowgraph technique has been shown to be effective for 3D visualization and quantitative measurement.

To solve the problem of video repair, we propose a new optical flow guidance solution that uses parallel structured convolution and attention networks to jointly infer video missing regions. In the network, a parallel structural model based on convolution and attention networks guided by optical flow is used to extract feature information to integrate the spatial and temporal context of video frames. This method integrates information between the target frame and the reference frame. To enhance feature learning capabilities using convolution and attention mechanisms, the feature fusion module fuses local and global features in an interactive manner, maximizing the retention of local and global representations. Our model produces visually satisfactory and time consistent results, while demonstrating on two benchmark datasets that our method outperforms the most advanced methods in terms of quantity and user research.