System and method for performing high-speed video capture and depth estimation using array cameras

摘要

High-speed video capture and depth estimation using array cameras is disclosed. Actual scenes typically include objects located at different distances from the camera. Therefore, estimating the depth during video capture by the array camera may result in smoother rendering of video from the captured image data of real scenes. One embodiment of the invention includes cameras that capture images from different viewpoints and an image processing pipeline application that obtains images from groups of cameras, wherein each group of cameras includes a different group of cameras And start capturing image data at the start time staggered with respect to the images. The application then determines the scene dependent geometric corrections that shift the pixels captured from the other viewpoints to the reference point by selecting the reference point and performing disparity searches to identify the disparity where the pixels from the different points of view are most similar do. The corrections can then be used to render the frames of the video.